2008-08-01
[17:09:00.0000] <Hixie>
maybe i should spend more time fixing the spec and less time responding to sam and insulting the tag.

[17:10:00.0000] <hober>
I must have missed the TAG insult email

[17:10:01.0000] <Hixie>
just sent it

[17:17:00.0000] <gsnedders>
/me sighs

[17:17:01.0000] <gsnedders>
Is it silly the sub. docs is longer than xref and TOC/numbering docs put together?

[17:42:00.0000] <Hixie>
http://www.whatwg.org/issues/data.html now shows labels in browsers that support the html5 fillText() api

[17:44:00.0000] <takkaria>
damn, I have to go and download one now...

[18:16:00.0000] <gsnedders>
/me realises he's completely screwed up 1.0b1 spec-gen docs by having no external links

[18:16:01.0000] <gsnedders>
Nevertheless, go get now!

[18:17:00.0000] <gsnedders>
Also, if there's anyone who should be in the ack but isn't, do email me

[18:18:00.0000] <gsnedders>
http://hg.gsnedders.com/hgwebdir.cgi/spec-gen/rev/fab6bfa129aa (see the bzip/zip/gz links to download)

[18:18:01.0000] <gsnedders>
Oh dear.

[18:18:02.0000] <gsnedders>
The docs say 1.0b1-dev to

[18:18:03.0000] <gsnedders>
*too

[18:19:00.0000] <gsnedders>
I really am too tired :P

[18:44:00.0000] <Hixie>
sam is rich

[18:44:01.0000] <Hixie>
"please don't dismiss me" he says, after not replying to almost any of the questions i ask him

[18:45:00.0000] <Hixie>
like, i write an e-mail "here's how you could help us move forward: X. So far you haven't helped us move forward."

[18:45:01.0000] <Hixie>
and he replies "You say I haven't helped you move forward! Whine whine whine."

[18:45:02.0000] <takkaria>
I think that thread is a waste of your time, fwiw, and you should probably stop replying

[18:45:03.0000] <Hixie>
i'm gonna see if he replies to the questions i asked

[18:45:04.0000] <Hixie>
if he does, we could make progress

[18:46:00.0000] <Hixie>
if he doesn't, i'll add him to my filter that labels e-mails as being "AAA IMPORTANT/CRITICAL"

[18:53:00.0000] <takkaria>
Hixie, I assume you saw the posts about getting WF2 integrated?  have you plans to do that soon now that we seem to have some consensus on it?

[18:54:00.0000] <Hixie>
yeah gonna do that after we publish next month

[18:55:00.0000] <takkaria>
awesome. :)

[20:05:00.0000] <kangax>
is it possible to modify image opacity (after it was rendered on canvas)?

[01:35:00.0000] <hsivonen>
apparently, browsers don't treat a bogus internal encoding decl after a BOM as an error in XML: http://upload.wikimedia.org/wikipedia/commons/3/3a/Bahia_Municip_Itapicuru.svg

[01:36:00.0000] <hsivonen>
nzkoz: you were looking for me?

[02:03:00.0000] <hsivonen>
I had dinner with friends who write software.

[02:04:00.0000] <hsivonen>
it seems to me that when people who have had to deal with Namespaces in XML can talk freely, they never have anecdotes about how Namespaces have helped them

[02:04:01.0000] <hsivonen>
instead, they have negative comments

[02:05:00.0000] <hsivonen>
OTOH, devil's advocate scenarios where Namespaces could help come from people who don't have to deal with Namespaces as part of their work

[02:15:00.0000] <othermaciej>
Namespaces are an example of the Fundamental Software Engineering Error

[02:15:01.0000] <othermaciej>
which is that something too terrible to actually use can be fixed by adding a level of indirection

[02:16:00.0000] <othermaciej>
sometimes that is true but software engineers try to do it even when it clearly is not

[02:23:00.0000] <hsivonen>
othermaciej: do you mean that URI-based extensibility is the too terrible thing in this case?

[02:23:01.0000] <othermaciej>
using URIs as a namespace identifier for tags in a markup language

[02:23:02.0000] <othermaciej>
is the terrible thing

[02:24:00.0000] <othermaciej>
if you had to mention the URI on every tag it would be clearly unusable

[02:24:01.0000] <othermaciej>
but since URIs are *obviously* the one true form of unique identifier, you add a level of indirection instead of rethinking why you are using them

[02:25:00.0000] <othermaciej>
or why URIs that are not meant to be dereferenced should start with http: and have a hostname

[02:38:00.0000] <hsivonen>
Hixie: if Google Translate isn't observing <code> now, why would it observe some other "do not translate" marker?

[02:41:00.0000] <hsivonen>
I wonder if Web authors would bother to annotate their stuff for machine translation

[02:42:00.0000] <Philip`>
If sometimes there are <code>s it ought to translate, it could just default to not translating and have some popup UI when you move the mouse over that text to offer to translate it

[02:46:00.0000] <Hixie>
hsivonen: good question

[02:47:00.0000] <Hixie>
hsivonen: though it wouldn't help with the wikipedia example, since that's not marked up right

[02:53:00.0000] <hsivonen>
/me expects a new round of Distributed Extensibility around ITS

[03:15:00.0000] <hsivonen>
In case anyone is wondering about Validator.nu weirdness, the DNS server that Validator.nu use for resolving addresses for outgoing connections is being really slow to respond today

[03:15:01.0000] <Hixie>
is it being attacked?

[03:15:02.0000] <Hixie>
i hear there are attacks going on now

[03:15:03.0000] <hsivonen>
I don't know.

[03:19:00.0000] <virtuelv>
are the mozilla devs still trapped in Whistler?

[03:19:01.0000] <virtuelv>
https://bugzilla.mozilla.org/show_bug.cgi?id=448604

[03:21:00.0000] <MikeSmith>
hsivonen: I just updated my local validator.nu and now getting "Exception in thread "main" java.lang.NoClassDefFoundError: org/mortbay/jetty/Connector" error

[03:22:00.0000] <hsivonen>
MikeSmith: did you run build.py with target 'all' or 'dldeps'?

[03:22:01.0000] <MikeSmith>
with "run"

[03:22:02.0000] <hsivonen>
MikeSmith: try dldeps first and then run again

[03:22:03.0000] <MikeSmith>
k

[03:23:00.0000] <MikeSmith>
OK, I see it's downloading the new dependencies now

[03:27:00.0000] <MikeSmith>
hsivonen: btw, the dldeps can sometimes be a PITA because certain downloads often fail with "Connection reset by peer" messages, and the download doesn't retry, so I have to go back and retry it manually

[03:27:01.0000] <MikeSmith>
happening now with the http://download.icu-project.org/files/icu4j/4.0/icu4j-4_0.jar download

[03:28:00.0000] <hsivonen>
MikeSmith: yeah, I'm experiencing problems with DNS right now. It has worked until now, so there hasn't been a need to make it retry before...

[03:28:01.0000] <MikeSmith>
OK

[03:28:02.0000] <hsivonen>
but yeah, I should probably make it retry

[03:33:00.0000] <Hixie>
MikeSmith: the subversion link is so that people can use a svn client to get the complete log, blame, diffs, etc (it's not supposed to be accessed from a browser)

[03:34:00.0000] <Hixie>
MikeSmith: would it make sense to expose it as a non-hyperlinked url, maybe?

[03:34:01.0000] <Hixie>
MikeSmith: (other changes look fine)

[03:35:00.0000] <MikeSmith>
great

[03:35:01.0000] <MikeSmith>
Hixie: yeah, I think a non-hyperlinked "svn checkout http://svn.whatwg.org/webapps/" would be good 

[03:35:02.0000] <Hixie>
cool, will do that then

[03:35:03.0000] <MikeSmith>
thanks

[03:38:00.0000] <Philip`>
'svn blame' isn't very useful, since it blames Hixie for everything

[03:38:01.0000] <gDashiva>
You mean it isn't all his fault?

[03:38:02.0000] <Hixie>
i use it a lot to track which version number a line was last edited in

[03:39:00.0000] <Philip`>
gDashiva: It is, but we know that already

[03:39:01.0000] <hsivonen>
Philip`: I deployed a new XML serializer. Feel free to try to break it.

[03:40:00.0000] <Philip`>
hsivonen: I might have a look when I have fewer urgent things to work on :-)

[03:41:00.0000] <Hixie>
MikeSmith: "choice of means" kind of sounds kooky to me so i'm changing that paragraph

[03:42:00.0000] <Hixie>
now it just reads:   <p>There are various ways to follow the change history for the specification:</p>

[03:42:01.0000] <MikeSmith>
Hixie: yeah, sounded funny to me too :) I just couldn't think of better wording..

[03:42:02.0000] <MikeSmith>
your revision sounds great to me

[03:45:00.0000] <Hixie>
i also changed your <dd><ul><li> construct to just a list of <dd>s, since <dl>s can have multiple <dd>s per <dt>s

[03:46:00.0000] <MikeSmith>
yeah, that's cleaner

[03:47:00.0000] <Hixie>
hey there's no link to the issues list either

[03:47:01.0000] <Hixie>
should we add taht?

[03:47:02.0000] <Hixie>
i guess i forgot to add it when i added it to the whatwg copy when daniel asked

[03:49:00.0000] <MikeSmith>
Hixie: yeah, seems like that would definitely be good to have too

[03:51:00.0000] <Hixie>
ok here's what i have so far: http://www.whatwg.org/specs/web-apps/current-work/.w3c/Overview.html

[03:52:00.0000] <MikeSmith>
/me looks now

[03:52:01.0000] <Hixie>
/me isn't sure he likes the text of the "HTML 5 bug/issue-tracking service" link but doesn't have a better suggestion

[03:52:02.0000] <Hixie>
too many capitals, numbers, and types of punctuation in short successon

[03:52:03.0000] <Hixie>
succession

[03:54:00.0000] <MikeSmith>
Hixie: yeah, that "HTML 5 bug/issue-tracking service" wording definitely klunky

[03:54:01.0000] <MikeSmith>
anyway, revised SOTD overall looks great 

[03:54:02.0000] <Hixie>
how about just "our public bug tracker"?

[03:54:03.0000] <Hixie>
or database

[03:55:00.0000] <MikeSmith>
"public bug database" sound best, i think

[03:56:00.0000] <MikeSmith>
"submit them to our public bug database"

[03:56:01.0000] <Hixie>
ok: http://www.whatwg.org/specs/web-apps/current-work/.w3c/Overview.html

[03:56:02.0000] <Hixie>
oh you think s/using/to/? i can do that too if you want

[03:56:03.0000] <MikeSmith>
beautiful

[03:57:00.0000] <MikeSmith>
nah, "using" is fine

[03:57:01.0000] <Hixie>
oh i should probably update the link to the bug list to not be the list that i use but hte list that includes all the bugs i am hiding from myself too!

[03:57:02.0000] <Hixie>
e.g. the ones i reassign to you :-)

[03:57:03.0000] <MikeSmith>
heh, yeah

[03:58:00.0000] <hsivonen>
hmm. the resolf.conf on the validator.nu machine is interesting

[04:00:00.0000] <MikeSmith>
Hixie: http://www.w3.org/Bugs/Public/buglist.cgi?quicksearch=ALL+product:HTML+-status:RESOLVED+-status:CLOSED

[04:00:01.0000] <MikeSmith>
I think

[04:00:02.0000] <MikeSmith>
for the "bug database" link

[04:00:03.0000] <MikeSmith>
hmm, though I see that picks up the authoring-guide also

[04:01:00.0000] <MikeSmith>
anyway, some form that quicksearch feature would seem best

[04:03:00.0000] <Hixie>
http://www.w3.org/Bugs/Public/buglist.cgi?component=Spec%20bugs&component=Spec%20proposals&product=HTML%20WG&resolution=NEEDSINFO&resolution=LATER&resolution=REMIND&resolution=---&order=bugs.resolution%2Cbugs.priority%2C%20bugs.bug_severity

[04:04:00.0000] <Hixie>
checked in

[04:39:00.0000] <Lachy>
wow, so Dean Edridge might be becoming an editor? http://lists.w3.org/Archives/Public/www-archive/2008Aug/0002.html

[04:39:01.0000] <Lachy>
it'll be interest to see how well he manages

[04:53:00.0000] <hsivonen>
hmm. Jigsaw is pretty seriously vintage Java...

[04:56:00.0000] <hsivonen>
Namespaces (java packages) don't solve the problem of the contents of the namespace being different in 2000 and 2008

[05:01:00.0000] <MikeSmith>
hsivonen: I have another question about the http://svn.versiondude.net/whattf/syntax/trunk/relaxng HTML5 schema

[05:03:00.0000] <MikeSmith>
does it actually capture the content-model constraints around phrasing prose/phrase/flow content?

[05:03:01.0000] <hsivonen>
MikeSmith: it should

[05:03:02.0000] <hsivonen>
subject to bugs, of course

[05:03:03.0000] <hsivonen>
MikeSmith: however, exclusions are handled in Schematron

[05:04:00.0000] <hsivonen>
and it doesn't capture the new transparent <a> thing yet

[05:04:01.0000] <MikeSmith>
OK. maybe I need to look at the assertions. I haven't much yet.

[05:04:02.0000] <hsivonen>
MikeSmith: do you have a test case that misvalidates?

[05:04:03.0000] <MikeSmith>
for a specific for example, where is the constraint that a <p> can't have a <ul> as a child?

[05:05:00.0000] <hsivonen>
/me looks

[05:05:01.0000] <MikeSmith>
hsivonen: <p><ul><li>foo</li></ul></p> doesn't misvalidate, but validator.nu doesn't actually seem to get to the point of validating it

[05:06:00.0000] <MikeSmith>
because it seems that the parser fixes it before it gets to the validation stage

[05:06:01.0000] <hsivonen>
MikeSmith: in block.rnc, p.inner is defined to be ( common.inner.phrase )

[05:07:00.0000] <MikeSmith>
hsivonen: right, and common.inner.phrase = text & common.elem.phrase*

[05:07:01.0000] <MikeSmith>
and common.elem.phrase = common.elem.embedded

[05:07:02.0000] <hsivonen>
MikeSmith: yeah, in that case, stuff happens according to the parsing algorithm before it reaches the schema layer

[05:07:03.0000] <MikeSmith>
and common.elem.embedded = notAllowed 

[05:07:04.0000] <MikeSmith>
hsivonen: right, that's what I meant for that particular case

[05:07:05.0000] <MikeSmith>
the </p> gets implied

[05:08:00.0000] <MikeSmith>
before the <ul>

[05:08:01.0000] <hsivonen>
that has nothing to do with the schema

[05:08:02.0000] <MikeSmith>
right, I understand that

[05:09:00.0000] <MikeSmith>
it seems like with a conformant HTML5 parser, there are many such cases

[05:09:01.0000] <MikeSmith>
with one consequence being that the error messages aren't going to be very helpful

[05:09:02.0000] <hsivonen>
but for XHTML5, the restriction is that common.elem.prose |= ul.elem does not end up augmenting common.inner.phrase

[05:10:00.0000] <MikeSmith>
OK

[05:10:01.0000] <hsivonen>
MikeSmith: there's a pending feature request to get warnings on implied tags

[05:10:02.0000] <MikeSmith>
ah

[05:10:03.0000] <MikeSmith>
that would be great to have

[05:11:00.0000] <MikeSmith>
ideally I think a user should see a message saying, e.g., "the <p> element cannot contain a <ul> as a child"

[05:11:01.0000] <MikeSmith>
or whatever

[05:11:02.0000] <MikeSmith>
that makes it very explicit

[05:12:00.0000] <hsivonen>
MikeSmith: the thing is, that omitting </p> is a legitimate way to end the <p>

[05:12:01.0000] <MikeSmith>
hmm, yeah, I realize that now

[05:13:00.0000] <MikeSmith>
god, all this stuff must make building a conformance checker a major PITA

[05:13:01.0000] <MikeSmith>
:)

[05:15:00.0000] <hsivonen>
right now, the PITA is that Jigsaw doesn't print informative diagnostics when stuff fails :-)

[05:16:00.0000] <MikeSmith>
I saw you had mentioned Jigsaw but I'm clueless so far about what you need it for

[05:16:01.0000] <MikeSmith>
what problem does it potentially solve for you?

[05:16:02.0000] <hsivonen>
MikeSmith: getting the W3C run an instance of Validator.nu under their preferred container

[05:17:00.0000] <MikeSmith>
ah

[05:17:01.0000] <MikeSmith>
that would definitely be really nice to have

[05:20:00.0000] <MikeSmith>
hsivonen: getting back to the HTML5 schema, am I confused, or is it the case that if you expand the content-model references out, common.inner.phrase just amounts to text & notAllowed

[05:20:01.0000] <hsivonen>
hmm. interesting. when the servlet-relative path is "/", Jigsaw gives it as null

[05:21:00.0000] <hsivonen>
MikeSmith: each phrase-level element definition augments that stub definition

[05:21:01.0000] <MikeSmith>
OK

[05:23:00.0000] <MikeSmith>
hsivonen: I see now... I just need to quit being lazy and to actually read the schema

[05:34:00.0000] <hsivonen>
do all browsers default to submitting the form to base uri if the action attribute on the form is missing?

[05:34:01.0000] <MikeSmith>
hsivonen: do you know of any tools that are able to generate a flattened version of an rng/rnc schema with the combine=choice definitions for a pattern actually combined into a single definition?

[05:35:00.0000] <hsivonen>
MikeSmith: I'm not aware of such a tool, but here's a guess

[05:35:01.0000] <hsivonen>
you might get that result if

[05:35:02.0000] <hsivonen>
you run Trang to convert the schema to RELAX NG XML syntax

[05:36:00.0000] <hsivonen>
and then run Kohsuke Kawaguchi's schema converter to convert the schema from RELAX NG to RELAX NG

[05:37:00.0000] <hsivonen>
but that's just a guess

[05:37:01.0000] <hsivonen>
then you could run Trang againg to compact syntax to make the result human-readable :-)

[05:37:02.0000] <hsivonen>
MikeSmith: Trang preserves the structure of the schema

[05:38:00.0000] <MikeSmith>
yeah, tried trang .. doesn't do it, unfortunately -- or fortunately, depending on how you look at it. trang faithfully preserves the RNC structure in RNG output in such a way that is seems like it's actually round-trippable

[05:38:01.0000] <hsivonen>
MikeSmith:  Kohsuke Kawaguchi's converter builds an abstact model and reserializes it without preserving structure

[05:38:02.0000] <hsivonen>
but IIRC, him tool doesn't read compact syntax

[05:39:00.0000] <MikeSmith>
I tried Dave Tolpin's incelim and it doesn't combine them either

[05:39:01.0000] <hsivonen>
hence, the need to use Trang, too

[05:39:02.0000] <MikeSmith>
hsivonen: OK

[05:39:03.0000] <MikeSmith>
will try Kohsuke's tool

[05:40:00.0000] <MikeSmith>
/me apologizes again for not actually reading carefullywhat hsivonen wrote above

[05:40:01.0000] <MikeSmith>
I'll shut up now :)

[05:40:02.0000] <MikeSmith>
for a while at least

[05:46:00.0000] <hsivonen>
w00t. I got Validator.nu to run inside Jigsaw. (without file upload support, without gzip support and without non-ASCII input support)

[05:49:00.0000] <MikeSmith>
hsivonen: congats

[05:49:01.0000] <hsivonen>
MikeSmith: thanks. now I need to document what I did. :-)

[09:18:00.0000] <MikeSmith>
hsivonen: any clues on getting Kohsuke's rngconv working with the HTML datatype library?

[09:19:00.0000] <MikeSmith>
hsivonen: I'm trying to run a conversion, but I'm getting "http://whattf.org/datatype-draft" is not a recognized data type vocabulary"

[09:29:00.0000] <tantek>
hsivonen, othermaciej just saw your dialog re: namespaces earlier (last night) http://krijnhoetmer.nl/irc-logs/whatwg/20080801#l-154

[09:30:00.0000] <tantek>
feel free to add a new section (or sections), like "implementation experience" and/or "fundamental software engineering error" to http://microformats.org/wiki/namespaces-considered-harmful

[09:34:00.0000] <hsivonen>
MikeSmith: no clue. do you have the library in classpath?

[09:35:00.0000] <hsivonen>
tantek: I recently started a wiki page, too: http://wiki.whatwg.org/wiki/Namespace_confusion

[09:35:01.0000] <hsivonen>
not much there yet

[09:36:00.0000] <tantek>
still, a good collection

[09:36:01.0000] <tantek>
feel free to link to your page also from http://microformats.org/wiki/namespaces-considered-harmful

[09:37:00.0000] <hsivonen>
tantek: ok. I will. (gotta run now, though)

[09:44:00.0000] <MikeSmith>
hsivonen: yeah, I got the dist/html5-datatypes.jar subdir of my http://svn.versiondude.net/whattf/syntax/trunk/relaxng/datatype/java working directory

[09:44:01.0000] <MikeSmith>
it's just that one jar file, right?

[12:08:00.0000] <hober>
/me wonders if getting involved with this thread was a good idea after all

[12:35:00.0000] <takkaria>
it wasn't

[12:36:00.0000] <takkaria>
neither is the way trackback/pingback have been brought up at all

[12:42:00.0000] <hober>
maybe I should set a cron job to email http://xkcd.com/386/ to me every morning...

[12:45:00.0000] <takkaria>
the debate on extensibility is fundamentally a religious one, I don't see how either side will ever buckle

[12:46:00.0000] <takkaria>
I have this suspicion that HTML5 will never become a W3C recommendation as a result of this and other permadiscussions

[14:29:00.0000] <takkaria>
libxml2's APIs suck a little bit

[15:13:00.0000] <Lachy>
"And authors want to add metadata. Instead of forcing it into containers that haven't been designed for it (@title, @data-*), let them do it properly." -- http://lists.w3.org/Archives/Public/public-html/2008Aug/0023.html

[15:13:01.0000] <Lachy>
I don't get what other way would be considered the proper way to embed metadata, beyond the mechanisms designed for adding metadata?!

[15:15:00.0000] <Lachy>
if, as Julian claims, title and data-* weren't designed for adding some type of metadata, then I must be missing something.


2008-08-02
[18:13:00.0000] <Lachy>
Hixie, why does it matter if an image is stretched or not, for the purpose of conformance?

[18:13:01.0000] <Hixie>
why does it matter if a tag is closed or not, for the purpose of conformance?

[18:14:00.0000] <Hixie>
in almost all cases, stretching an image is a mistake. in the remaining cases where it is intentional, it's dubious practice. this argues for the author being notified of the problem.

[18:15:00.0000] <Hixie>
same as with anything else that is a conformance error

[20:39:00.0000] <kangax>
Could anyone point to an SVG to Canvas converter?

[01:48:00.0000] <hsivonen>
MikeSmith: yeah, it's just that one jar file

[01:48:01.0000] <MikeSmith>
hsivonen: yeah, I got it figured out finally

[01:49:00.0000] <MikeSmith>
I was using "jar -r" to run it

[01:49:01.0000] <MikeSmith>
and that caused it to ignore the datatype library despite the fact I had it in my classpath

[01:50:00.0000] <hsivonen>
MikeSmith: the -jar switch sucks

[01:50:01.0000] <MikeSmith>
yeah

[01:50:02.0000] <MikeSmith>
I should have known better than to try it 

[01:52:00.0000] <MikeSmith>
anyway, I did manage to be able to run it, but not without fatal errors

[01:52:01.0000] <MikeSmith>
running against the HTML5 schema, I get java.lang.StackOverflowError

[01:53:00.0000] <MikeSmith>
at com.sun.msv.grammar.ExpressionCloner.onChoice(ExpressionCloner.java:37)

[01:54:00.0000] <MikeSmith>
anyway, I'm giving up on it

[01:55:00.0000] <MikeSmith>
and instead just writing a stylesheet to pre-process the schema and combine all the choice=combine stuff

[02:06:00.0000] <hsivonen>
MikeSmith: msv is no good with the default stack size of hotspot

[02:07:00.0000] <hsivonen>
-XX:ThreadStackSize=2048 

[02:10:00.0000] <MikeSmith>
hsivonen: OK, trying now

[02:12:00.0000] <MikeSmith>
getting same error even with that switch

[02:13:00.0000] <MikeSmith>
but anyway, the idea is kind of a no-go regardless, due to that fact that it's going to restructure the whole schema

[02:14:00.0000] <MikeSmith>
the only thing I really want is to just consolidate the @combine=choice stuff, which I think can manage to do with XSLT

[03:18:00.0000] <Lachy>
oh, I wish what I wrote in IRC didn't get dragged into the thread on the mailing list :-(. I chose not to respond on the list for a reason.

[03:22:00.0000] <Hixie>
i hope mike is happy: http://www.whatwg.org/specs/web-apps/current-work/#fetching :-)

[03:25:00.0000] <MikeSmith>
/me smiles

[03:25:01.0000] <MikeSmith>
Hixie: cool

[03:27:00.0000] <MikeSmith>
"At a time convenient to the user and the user agent" is an interesting phrase

[03:30:00.0000] <Lachy>
heh. It makes it sound like my browser should book an appointment with me to do the download :-)

[05:21:00.0000] <Hixie>
nn

[15:26:00.0000] <Hixie>
i think i might make alt="" required and say that when you don't know what the image is, you have to say what kind of image it is (e.g. "uploaded image", "photo", "thumbnil", or whatever) and put that in braces in the alt="" attribute, as in alt="{photo}"

[15:26:01.0000] <Hixie>
and that that is never allowd inside a link

[15:27:00.0000] <Hixie>
(in a link it should just give the text that is appropriate for the link, e.g. alt="View image" if it's a link to the image)

[15:29:00.0000] <Hixie>
this seems to handle all the cases that allowing alt="" to be omitted does, while verly mildly improving the accessibility of those images, and being slightly more compatible with legacy UAs

[15:29:01.0000] <Hixie>
(it affects some legacy pages, though not many, and certainly no more than the missing-alt-altogether case would)

[15:40:00.0000] <takkaria>
seems like a wise decision

[15:47:00.0000] <Lachy>
Hixie, why use curly braces instead of square brackets?

[15:48:00.0000] <jcranmer>
Lachy: it's more exotic

[15:48:01.0000] <jcranmer>
and probably less likely to conflict

[15:49:00.0000] <Lachy>
ok, makes sense

[16:02:00.0000] <Hixie>
Lachy: [] are used a lot already

[16:02:01.0000] <Hixie>
Lachy: {} not so much

[16:02:02.0000] <Hixie>
Lachy: iirc <> is used even less, but that causes problems in xml or something

[16:02:03.0000] <Hixie>
/me logs on to his vpn to get the numbers

[16:03:00.0000] <jcranmer>
&lt; &gt;, if anyone remembers

[16:05:00.0000] <Hixie>
ok here are the stats as percentages of total pages scanned

[16:05:01.0000] <Hixie>
pages that had an img that wasn't in a link and had a value that followed the pattern [...]: 0.45%

[16:05:02.0000] <Hixie>
pages that had an img that wasn't in a link and had a value that followed the pattern (...): 0.13%

[16:06:00.0000] <Hixie>
pages that had an img that wasn't in a link and had a value that followed the pattern {...}: 0.035%

[16:06:01.0000] <Hixie>
pages that had an img that wasn't in a link and had a value that followed the pattern <...>: 0.033%

[16:07:00.0000] <Hixie>
most common alt={...} value was alt={alpha}, most of which it seems came from pages created by one particular conversion tool

[16:08:00.0000] <jcranmer>
I'm surprised (...) is that low

[16:09:00.0000] <Hixie>
number of pages with <img>: 94%

[16:09:01.0000] <Hixie>
number of pages with <img alt>: 82%

[16:09:02.0000] <Hixie>
number of pages with <img alt> with non-empty alt: 77%

[16:10:00.0000] <Hixie>
number of pages with <img> that had at least one without an alt="": 67%

[16:11:00.0000] <Hixie>
number of pages with <img alt> and that had all their <img> with alt="": 27%

[16:11:01.0000] <Hixie>
number of pages that had at least one <img> element but no <img> elements with alt="": 11%

[16:12:00.0000] <jcranmer>
which comes out to somewhere between 29%-71% of images that don't have alt

[16:12:01.0000] <Lachy>
ok, well, it nicely solves the problem of distinguishing between legitmate and guessed alt text, so it seems like a reasonable solution

[16:13:00.0000] <Hixie>
most of the alt=""s with the pattern <...> seemed to be mistakes, e.g. 0.0015% of pages had alt="span style='background-color: #CCFFFF'>Visa</span>"

[16:13:01.0000] <Lachy>
so what would authoring tools like Dreamweaver be requried to insert by default? Would alt="{image}" be ok?

[16:14:00.0000] <Hixie>
if the author knows what the image is, then alt="{...}" is never "ok"

[16:14:01.0000] <jcranmer>
I think apache's autoindex uses "[ TXT ]" as alt..

[16:14:02.0000] <Hixie>
but yeah, i guess that would be a reasonable default for the case where today they just omit alt="" or have alt="" empty without probable cause

[16:15:00.0000] <Lachy>
yeah, I know that. But by default when a user just drags and drops an image into the WYSIWYG editor, and doesn't enter anything for alt text into the prompt

[16:15:01.0000] <jcranmer>
actually, just "[TXT]"

[16:15:02.0000] <Hixie>
jcranmer: second most common [...] alt value was [DIR] 0.085% (first was [new] 0.095%, third was [NEW] 0.066%)

[16:16:00.0000] <jcranmer>
[DIR] is used in apache's autoindex

[16:16:01.0000] <Hixie>
jcranmer: [TXT] was 0.024%

[16:16:02.0000] <Hixie>
less common than [cpp] and [flash]

[16:16:03.0000] <Hixie>
and [*]

[16:17:00.0000] <Hixie>
most common (...) values were (+), (-) and (?)

[16:17:01.0000] <Lachy>
what would [cpp] be used for?

[16:17:02.0000] <jcranmer>
\[[A-Z]+\] is quite possibly autoindexing

[16:17:03.0000] <jcranmer>
I don't know what IIS uses, if anything

[16:17:04.0000] <Hixie>
(0.038%, 0.037%, and 0.0095% respectively)

[16:18:00.0000] <Hixie>
yeah apache's autoindexing was well represented in these results

[16:18:01.0000] <Hixie>
[spoiler] was common too

[16:18:02.0000] <jcranmer>
what does [] look like if you exclude apache, which is probably a valid use case?

[16:19:00.0000] <jcranmer>
although it would have to be ~30% of all stuff to change the rankings

[16:19:01.0000] <Hixie>
valid how?

[16:19:02.0000] <Hixie>
not sure what you mean

[16:20:00.0000] <webben>
Hixie: Hmm what turned out to be wrong with using an attribute to signal the poverty of alt text, rather than resorting to odd syntax inside the alt attribute?

[16:20:01.0000] <webben>
especially given the use-case is automatic insertion rather than hand-authoring

[16:21:00.0000] <Hixie>
jcranmer: the [...] values were (roughly in order): [new], [DIR], [NEW], [   ], [*], [b], [img], [i], [url], [u], [email], [quote], [flash], [fixed], [spoiler], [cpp], [strike], [TXT], [IMG], [ICO], [M], ...

[16:22:00.0000] <jcranmer>
five of which are definitely apache

[16:22:01.0000] <Hixie>
webben: it seems more likely to be misused

[16:22:02.0000] <Hixie>
webben: e.g. through ignorant copy-paste

[16:22:03.0000] <Hixie>
webben: also, coming up with a name was difficult

[16:22:04.0000] <jcranmer>
another seven of which seem to be BB-code

[16:23:00.0000] <webben>
I'd have thought exactly the opposite was the case.

[16:23:01.0000] <jcranmer>
[img alt="[b]SEE THIS[/b]"] ?

[16:23:02.0000] <webben>
that a weird syntax inside alt is utterly opaque

[16:23:03.0000] <jcranmer>
although the non-existence of [/b] does seem to invalidate that theory...

[16:24:00.0000] <webben>
Hixie: Is there a list of proposed names anywhere?

[16:24:01.0000] <Hixie>
webben: not a convenient list, no

[16:25:00.0000] <Lachy>
webben: noalt, important

[16:25:01.0000] <Lachy>
I can't remember the others

[16:25:02.0000] <webben>
yeah, well, I agree those aren't good names ;)

[16:25:03.0000] <Hixie>
webben: what kind of proposal did you have in mind? <img alt="..." alt-is-not-actually-a-description-but-a-category="true"> ?

[16:25:04.0000] <Hixie>
webben: (with a better name obviously!)

[16:26:00.0000] <webben>
well, you wouldn't need the ="true" (presumably?) but yeah.

[16:27:00.0000] <Hixie>
that was basically my importantimage="" proposal, but nobody could come up with a good attribute name.

[16:28:00.0000] <webben>
missing-text-equivalent

[16:28:01.0000] <Dashiva>
please-sir-can-i-have-some-more-validation

[16:29:00.0000] <Hixie>
webben: <img alt="Photo" missing-text-equivalent> vs <img alt="{Photo}"> seems like a tossup as to which is better

[16:29:01.0000] <webben>
that name's still probably not ideal, but I think the former is easily better.

[16:30:00.0000] <webben>
/me would suggest testing it with some newbie authors and asking them what they think these syntaxes mean

[16:30:01.0000] <webben>
actually that's not a fair test

[16:30:02.0000] <webben>
since that clues them in that {} is a syntax, which is counter-intuitive

[16:31:00.0000] <Hixie>
heh

[16:32:00.0000] <Hixie>
i wish i could remember why i had decided to look at alt={...} rather than importantimage=""

[16:32:01.0000] <Hixie>
there was some more serious problem than the name, iirc, but i don't recall what

[16:33:00.0000] <Dashiva>
(You could use self-documenting decisions, that way you won't have to document anything)

[16:33:01.0000] <Hixie>
not sure what that would mean here :-)

[16:34:00.0000] <Hixie>
the decision wasn't written down anywhere, i just did it

[16:35:00.0000] <Lachy>
if we documented every decision, that would take the fun out of rehashing old arguments!

[16:39:00.0000] <Dashiva>
Hixie: I just felt like taking a jab at self-documenting code.

[16:39:01.0000] <Hixie>
heh

[16:42:00.0000] <Hixie>
webben: missing-text-eqivalent doesn't really convey the right message, i'm not sure it actually helps vs {...}

[16:42:01.0000] <webben>
Hixie: is " alt-is-not-actually-a-description-but-a-category" the right message?

[16:42:02.0000] <Lachy>
Hixie, are you speccing the {...} feature now?

[16:43:00.0000] <Hixie>
the message is alt-is-not-actually-a-description-but-a-category-because-we-do-not-know-what-the-image-actually-is

[16:43:01.0000] <Hixie>
Lachy: i'm looking at it. it's one of the folders with the most messages

[16:44:00.0000] <webben>
Hixie: alt-is-category-only ?

[16:44:01.0000] <Hixie>
webben: i guess the reason i prefer {...} is that if we're going to come up with some mostly opaque syntax, i'd rather pick the most compact

[16:45:00.0000] <Hixie>
webben: that's pretty long, and still doesn't really help, i mean, what's a category? does that mean i can just do this on all my images? etc.

[16:46:00.0000] <webben>
I think the what's a category question is answered by the alt content itself.

[16:47:00.0000] <Lachy>
I like the {} syntax better because it looks ugly enough to discourage authors from using it on all their images, yet simple enough to be used where appropriate

[16:47:01.0000] <Lachy>
would alt="{}" be considered non-conforming?

[16:48:00.0000] <webben>
While {} might discourage authors using {}, it doesn't make it clear the alt is suboptimal. Consequently newbies could easily take away the message that alt="Photo" is a good alt.

[16:49:00.0000] <webben>
missing-text-equivalent at least hints that something's wrong.

[16:49:01.0000] <Lachy>
and which of these regexes best matches the proposed syntax: /^\{.+\}$/ or /^\{[^\}]+\}$/

[16:51:00.0000] <Hixie>
webben: the stats indicate that authors already think omitting alt altogther is fine

[16:51:01.0000] <Hixie>
webben: so i don't think this will make it any worse

[16:51:02.0000] <Hixie>
Lachy: i was just thinking /^\{.*\}$/

[16:51:03.0000] <webben>
Hixie: I don't see the logical connection between those stats and the effect of any given example.

[16:52:00.0000] <Hixie>
webben: i'm saying that nothing could make the current authoring practices worse

[16:52:01.0000] <Lachy>
Hixie, so then alt="{}" would be conforming, even though it's almost completely useless to UAs since it says nothing about what type of image it is?

[16:52:02.0000] <webben>
I don't think not making it any worse is the bar we should be setting. ;)

[16:53:00.0000] <Hixie>
Lachy: alt="{image}" doesn't say anything about what it is either

[16:53:01.0000] <Lachy>
I guess those two could be considered equivalent then

[16:53:02.0000] <webben>
fwiw it could easily be worse.

[16:53:03.0000] <Hixie>
webben: i don't think having a few people stick misteriously named attributes on images is going to improve things either

[16:54:00.0000] <webben>
/me doesn't really see why not.

[16:55:00.0000] <webben>
if the problem is mysteriousness, then go for a big huge long name.

[16:55:01.0000] <Hixie>
then nobody will use it

[16:55:02.0000] <webben>
why would nobody use it?

[16:55:03.0000] <webben>
nobody would use it /accidentally/ ... which is precisely what you're trying to avoid

[16:56:00.0000] <Hixie>
it's a psychology thing -- people just don't seem to use long keywords, they shy away from them

[16:56:01.0000] <Hixie>
i don't know why

[16:56:02.0000] <webben>
We're not talking about hand-authoring here. We're talking about sites like Flickr processing thousands of images; and software like DreamWeaver written by professionals.

[16:56:03.0000] <Hixie>
i guess i'm not sold on the idea that the advantage of an attribute over just a compact syntax outweigh the disadvantages... the pros and cons on both sides seem pretty minimal

[16:57:00.0000] <webben>
that's the target audience of this attribute as I understand it.

[16:57:01.0000] <Hixie>
flickr output is hand-authored templates.

[16:57:02.0000] <Hixie>
it's still hand-authored.

[16:57:03.0000] <webben>
the templates are hand-authored; the output isn't.

[16:58:00.0000] <webben>
likewise someone writes the code for DreamWeaver

[16:58:01.0000] <webben>
s/hand-authoring/small-time authoring/ if you like

[16:58:02.0000] <Hixie>
my point is it's not like we can just ignore authoring because there's a computer involved -- it's still hand written at some point

[16:59:00.0000] <Lachy>
webben, even if the attribute were theoretically a better approach (which I'm not convinced it is), then there's still the big problem of finding an appropriate name that accurately represents its meaning for all the vaious use cases


2008-08-03
[17:00:00.0000] <Hixie>
yeah i haven't yet seen a name that i'd be proud to have in a spec with my name at the top

[17:00:01.0000] <webben>
Lachy: If the meaning is ambiguous, then that's true of {} too. If the meaning can be expressed, then it can be expressed in an attribute name.

[17:00:02.0000] <webben>
if {} is ambiguous, that may hint at a problem with the idea

[17:00:03.0000] <webben>
are there two concepts here that need seperating?

[17:01:00.0000] <Hixie>
{}'s advatnage is great compactness, its disadvantage is it's meaning is not intuitive. for an attribute to outweigh its corresponding verbosity disadvantage, it has to have a name that is intuitive

[17:02:00.0000] <Hixie>
s/it's/its/

[17:02:01.0000] <webben>
I think you're underestimating that {} doesn't even look like code, and therefore is likely to be misused.

[17:02:02.0000] <webben>
one can't really dispute an attribute looks like code, even if it's not obvious what it does.

[17:02:03.0000] <Hixie>
why would it be misused more than now?

[17:03:00.0000] <webben>
it doesn't exist now.

[17:03:01.0000] <Hixie>
alt=[...] is used a lot

[17:03:02.0000] <Hixie>
alt={...} is not

[17:03:03.0000] <webben>
yeah, but not as code.

[17:03:04.0000] <Hixie>
you're saying that alt={...} would become more popular for other uses just because we introduce it as meaning something special for one use?

[17:03:05.0000] <Lachy>
webben, you're not making sense

[17:04:00.0000] <webben>
Hixie: Given folks don't normally see alt, that's not actually inconceivable.

[17:04:01.0000] <Hixie>
re what you asked earlier... the concept here is "i don't know what this image is or will be, so i cannot provide a useful equivalent... here's a hint as to what kind of image it is, at least, so that you know it's not meant to be purely decorative"

[17:05:00.0000] <Lachy>
it's possible that once this syntax starts showing up in poorly written tutorials, authors could misunderstand its purpose and begin using it more commonly, yet wrongly

[17:05:01.0000] <Hixie>
webben: it seems pretty unlikely to me. We know that attributes that people use get copied and pasted around even without reason, too, so i don't see why it would happen any less to an attribute than to a special syntax in an attribute.

[17:06:00.0000] <Hixie>
webben: maybe it's in fact more likely that authors who don't know about this would not know that the {...} syntax means anything, and would thus in fact not use it, as they think it's ugly :-)

[17:06:01.0000] <Hixie>
webben: whereas they would see an attribute and know that it DID mean something, even if they didn't know what, and so WOULD use it (as we have seen happen with other things)

[17:06:02.0000] <Hixie>
e.g. bits of svg appearing in random places in html documents

[17:07:00.0000] <webben>
I think "bits of svg" are probably a lot more opaque than any of the names suggested for this thing.

[17:07:01.0000] <Hixie>
svg was just one example, it happens with everything

[17:08:00.0000] <Hixie>
hm, when we started this discussion i was pretty much neutral on the issue of importantimage="" vs alt={...} but now i'm definitely leaning more towards the latter

[17:08:01.0000] <webben>
Well, yeah, but you need to work out what makes things happen more, not just look at whether they happen at all.

[17:09:00.0000] <Hixie>
i'm gonna go shopping and will think more on this, but feel free to keep discussing it, i'll read any ideas that come up when i get back

[17:09:01.0000] <webben>
k ; have a good shop :)

[17:09:02.0000] <webben>
/me is probably heading to bed very shortly.

[17:10:00.0000] <Lachy>
I think the {} syntax would be accepted by the community better than importantimage="", because there were a lot of suggestions for using a similar approach with square brackets, and very little support for using importantimage (it was mostly ignored when it was suggested, depite repeatedly pointing to it)

[17:11:00.0000] <Lachy>
Hixie, btw, did your Stargate Continuum DVD arrive yet?

[17:11:01.0000] <Lachy>
if so, what did you think of it?

[17:14:00.0000] <webben>
Lachy: I don't think importantimage was a good name; "" implies unimportant image; and it doesn't convey that something is missing.

[17:15:00.0000] <Lachy>
webben, please elaborate?

[17:15:01.0000] <webben>
sorry alt="" => decorative (unimportant) image.

[17:16:00.0000] <webben>
alt="something" => important image

[17:16:01.0000] <webben>
importantimage => no additional information

[17:16:02.0000] <Lachy>
the same is true for alt={}

[17:17:00.0000] <webben>
yes, but I'm suggesting why importantimage wasn't a good name

[17:18:00.0000] <webben>
missing-text-equivalent does at least add some information, though I'm still not precisely happy with it.

[17:18:01.0000] <Lachy>
oh, ok. I misread your message. I thought you said it was a good name

[17:18:02.0000] <webben>
"alt-is-hint-only" perhaps

[17:19:00.0000] <Lachy>
it's too long

[17:19:01.0000] <webben>
/me is thoroughly unconvinced that long is bad. It's actually a plus AFAICT.

[17:19:02.0000] <Lachy>
and there's a possibillity that some authors could inadvertently write alt-is-only-hint instead of alt-is-hint-only

[17:20:00.0000] <webben>
there's also a possibility authors could write {)

[17:20:01.0000] <Lachy>
so using sentences for attribute names isn't really a good idea, especially when it's possible to transpose words without losing its meaning

[17:20:02.0000] <webben>
or " {

[17:21:00.0000] <Lachy>
it's easier to spot a syntax error like that, than it is to spot incorrectly transposed words in an attribute name

[17:21:01.0000] <webben>
the former cannot be caught by the validator; the second can.

[17:22:00.0000] <webben>
sorry {) is not definitely invalid, but alt-is-only-hint definitely is.

[17:22:01.0000] <webben>
so at best a validator could issue a warning about the former.

[17:22:02.0000] <Lachy>
that's what the image analysis tool in the validator is for.

[17:26:00.0000] <webben>
I think having a non-verifiable syntax would not make for a more user-friendly image analysis tool.

[17:26:01.0000] <webben>
since then you're asking people to check syntax as well as equivalents.

[17:27:00.0000] <webben>
whereas you could have them fix the syntax then check equivalents

[17:31:00.0000] <Lachy>
I'm not really convinced that it's likely for authors to mistype {} as {), because the keys are in different positions on the keyboard, and the braces are right next to each other

[17:32:00.0000] <webben>
{]

[17:32:01.0000] <Lachy>
and in this whole discussion, I've not seen anyone accidentally mistype {}

[17:32:02.0000] <webben>
on my keyboard at least } ] are the same key

[17:33:00.0000] <Lachy>
yeah, that's true, but still unlikely

[17:33:01.0000] <webben>
not sure why that would be unlikely

[17:33:02.0000] <Lachy>
have you ever seen it happen anywhere? If so, how freqently?

[17:34:00.0000] <webben>
i don't have much memory of typos and their frequency full stop

[17:34:01.0000] <webben>
let alone mental data on } ] in particular ;)

[17:35:00.0000] <Lachy>
so, in other words, you're just speculating and trying to put that forth as strong evidence anyway?

[17:35:01.0000] <webben>
/me thinks the whole discussion is very speculative.

[17:36:00.0000] <webben>
/me doesn't figure "there's also a possibility authors could write {)" is an especially strong statement either

[17:38:00.0000] <webben>
I do know I don't want to have to get time-poor editorial or QA staff to understand the ins and outs of this syntax when I could get it checked with a validator /before/ giving them more useful work of inspecting text that is supposed to be a text equivalent.

[17:39:00.0000] <webben>
or rely on their eyes when this can be handled by code.

[17:40:00.0000] <Lachy>
so maybe there are ways of at least flagging mismatched braces as a warning in the validator then

[17:40:01.0000] <webben>
that's still a waste of their time.

[17:40:02.0000] <Lachy>
what?

[17:40:03.0000] <webben>
they'd need to manually inspect the braces

[17:41:00.0000] <Lachy>
so? Braces are used very infrequently within alt text anyway, so it's hardly a lot of time wasted

[17:42:00.0000] <webben>
that's worse, then they'd need to try and remember what that pesky developer said about braces

[17:42:01.0000] <webben>
and () is just like {} right?

[17:42:02.0000] <Lachy>
I don't understand your point

[17:42:03.0000] <webben>
well, not worse, but not good either.

[17:43:00.0000] <webben>
Lachy: Basically, it shifts a job that could be done more reliably by machines onto people.

[17:43:01.0000] <webben>
that's impractical.

[17:44:00.0000] <Lachy>
checking the accuracy of alt text isn't a job that can be reliably done by machines anyway, so having authors manually check those using braces along with all the others isn't that big a deal

[17:44:01.0000] <webben>
this isn't about "checking the accuracy of alt text"

[17:44:02.0000] <Lachy>
of course it is

[17:44:03.0000] <Lachy>
what else is it about?

[17:45:00.0000] <webben>
Let's say you have a source of photos coming into a system.

[17:45:01.0000] <webben>
so you have some code which inspects each one to see if it has some text to use as an equivalent

[17:45:02.0000] <webben>
if it does, it inserts the text; if not, it inserts alt="{Photo}"

[17:47:00.0000] <Lachy>
yeah, and...?

[17:47:01.0000] <webben>
why should humans be checking if the system can spell {Photo} correctly, rather than just getting a total of those with {Photo} and proceeding to check the ones that do have alt text?

[17:49:00.0000] <webben>
I guess the image checker could group them

[17:49:01.0000] <webben>
but that would probably mean you'd end up with false positive

[17:49:02.0000] <webben>
*positives

[17:50:00.0000] <Lachy>
you seem to be making assumptions about the UI of the image inspector. Given the {} syntax, why couldn't the image inspector group those with {Photo} together and provide a count, or whatever else?

[17:50:01.0000] <Lachy>
I don't see how you'd end up with any more false positives using alt="{...}" as you would with some attribute

[17:52:00.0000] <webben>
Lachy: Yeah. Maybe. Would get a lot more complicated if alt's were being autogenerated to contain more information

[17:52:01.0000] <webben>
e.g. {Photo tagged with 'cat'}

[17:54:00.0000] <webben>
in fact, the better your auto-generation of alts, the worse it would get.

[17:55:00.0000] <webben>
well, actually, I guess the checker could just group {} and non-{} together for separate checking

[18:01:00.0000] <Lachy>
webben, yeah, exactly how it would group alt="..." separately from some-special-attribute=""

[18:02:00.0000] <Lachy>
and if, in the process of inspecting the alt="..." group, the author sees a { in there, it would look like something that needed fixing

[18:08:00.0000] <webben>
maybe

[18:09:00.0000] <webben>
well, it would if the validator also warned about it /and/ if {} weren't common for that set of alt's

[18:09:01.0000] <webben>
meh, this also means you'd need code to inspect provided equivalents for { } and decide what to do with it

[18:10:00.0000] <webben>
e.g. does this mean the provided equivalent is actually not an equivalent but someone who actually knows the {} syntax and is assuming your just going to dump into an alt attribute.

[18:10:01.0000] <webben>
or is this actually the equivalent

[18:15:00.0000] <webben>
also, it's cutting into the strings people use for interpolation (e.g. YAHOO.lang.substitute uses {} ) http://developer.yahoo.com/yui/docs/YAHOO.lang.html

[18:16:00.0000] <jcranmer>
any good pages with <video> for testing?

[18:17:00.0000] <takkaria>
jcranmer: wikimedia?

[18:18:00.0000] <takkaria>
jcranmer: http://www.double.co.nz/video_test/ also

[18:23:00.0000] <jcranmer>
/me goes through old CSS mailing list messages and laughs

[18:25:00.0000] <jcranmer>
"Let's drop #id as it's superfluous and writing foo\#bar is too hard for most users"

[18:29:00.0000] <jruderman>
yeah, let's make everyone write [id="baz"] so we don't have escape #

[18:30:00.0000] <jcranmer>
some of his suggestions get even crazier

[18:31:00.0000] <takkaria>
whose suggestions are these?

[18:32:00.0000] <jcranmer>
Dmitry Turin

[18:32:01.0000] <Lachy>
jcranmer, http://lachy.id.au/dev/markup/tests/html5/video/

[18:32:02.0000] <jcranmer>
Lachy: thanks

[18:33:00.0000] <jcranmer>
/me notes that wikimedia didn't do anything obvious to get to a <video> tag

[18:34:00.0000] <Lachy>
jcranmer, got a pointer to that suggestion of his?

[18:35:00.0000] <Lachy>
the #id one

[18:36:00.0000] <jcranmer>
Lachy: material around 1/10-ish/2008

[18:37:00.0000] <Lachy>
is that around January 10th?

[18:37:01.0000] <jcranmer>
er, yes

[18:37:02.0000] <Lachy>
/me has difficulty interpreting american date formats

[18:37:03.0000] <jcranmer>
sorry, I forget that not everyone uses American date format

[18:37:04.0000] <jcranmer>
2008-01-10-ish

[18:38:00.0000] <Lachy>
no-one should ever use it. It's horrible, it has the numbers in a weird order

[18:38:01.0000] <takkaria>
Dmitry suggested dropping # *this year*?

[18:39:00.0000] <jcranmer>
Lachy: force of habit

[18:40:00.0000] <Lachy>
LOL! That's related to his SQL5 proposal :-)

[18:40:01.0000] <jcranmer>
for a moment, I thought he was the crackpot who wanted to wave all problems away magically

[18:40:02.0000] <jcranmer>
wrong crackpot, though

[18:40:03.0000] <Lachy>
or is it SQL 50?

[18:41:00.0000] <jcranmer>
5

[18:41:01.0000] <jcranmer>
SQL5, HTML6, Unicode7, Computer2

[18:41:02.0000] <Lachy>
the domain he uses is sql50.euro.ru

[18:41:03.0000] <Lachy>
hmm, the page he linked to is gone

[18:42:00.0000] <Lachy>
http://lists.w3.org/Archives/Public/www-style/2008Jan/0188.html

[18:43:00.0000] <Lachy>
it's unfortunate he chose the number 5 though. We've been using it to represent the spec versions that are actually good. (HTML5, XML5, HTTP5, DOM5, etc.)

[18:43:01.0000] <jcranmer>
/me laughs at his Computer 2 stuff

[18:43:02.0000] <jcranmer>
it seems to me that he understands very little of TCP/IP

[18:43:03.0000] <jcranmer>
and certainly nothing of IPv6

[18:44:00.0000] <takkaria>
I like that Computer 2.0 "It gives large jump in development of economy."

[18:44:01.0000] <jcranmer>
and some of those proposals are incapable of handling the future well

[18:45:00.0000] <jcranmer>
takkaria: I think he's using automatic translation

[18:46:00.0000] <takkaria>
I think he just speaks bad English

[18:46:01.0000] <jcranmer>
ah, here's something interesting

[18:46:02.0000] <Lachy>
I like his 6D mouse :-) http://computer20.euro.ru/site/computer20/en/author/6d_eng.htm

[18:46:03.0000] <jcranmer>
he's proposing to get rid of HTTP, FTP, SMTP, POP, etc.

[18:46:04.0000] <jcranmer>
replaced by

[18:46:05.0000] <jcranmer>
... wait for it ...

[18:46:06.0000] <jcranmer>
SQL5

[18:46:07.0000] <Lachy>
of course. makes perfect sense

[18:47:00.0000] <jcranmer>
apparantely because SQL5 supports "continue downloading of file at breakdown of connection"

[18:47:01.0000] <Lachy>
I don't see why you're laughing at these obviously fantastic proposals! ;-)

[18:48:00.0000] <Lachy>
really? That'd it be great if I could still surf the web without having a working internet connection. Then I wouldn't have to complain to my ISP so frequently

[18:49:00.0000] <takkaria>
he thinks people should communicate over XML, too

[18:49:01.0000] <takkaria>
but not actual XML, weird unquoted XML

[18:50:00.0000] <jcranmer>
you mean SGML without the wacky stuff?

[18:51:00.0000] <takkaria>
I really don't know what he means

[18:51:01.0000] <Lachy>
he wants more wacky stuff, like strange characters in tag names

[18:51:02.0000] <jcranmer>
well, obviously, he wants his stuff

[18:51:03.0000] <takkaria>
he's already started work on HTML6 too

[18:52:00.0000] <Lachy>
"I also propose to allow any spec-symbol &, ^, @, ~, %, $ in name of a tag for future extentions."

[18:52:01.0000] <jcranmer>
that &'ll be a fun one

[18:54:00.0000] <takkaria>
btw, there is almost an html5 browser that you can use

[18:54:01.0000] <takkaria>
well.  s/an html5 browser/a browser using the html5 parsing algorithm/

[18:55:00.0000] <takkaria>
no javascript, but at least it's an implementation

[18:55:01.0000] <takkaria>
it's not quite ready yet since I'm still hunting down bugs. :)

[18:56:00.0000] <jruderman>
wouldn't it be easier to shoehorn your html5 parser into gecko or webkit than make a new browser around it?

[18:57:00.0000] <takkaria>
I'm not writing a new web browser around it, it already existed

[18:57:01.0000] <takkaria>
the browser in question is NetSurf (http://www.netsurf-browser.org/)

[18:57:02.0000] <jruderman>
ahh

[18:58:00.0000] <takkaria>
NS currently uses libxml2, so all I have to do is to bind Hubbub (the C html5 parser I've been working on) to libxml2, and NS should work

[18:59:00.0000] <takkaria>
hooking it into gecko looks like it would be hard work; WebKit looks slightly easier but I would think some fairly chunky rewriting would still have to take place

[19:01:00.0000] <jruderman>
i'm sure mrbkap would be happy to help you, so he doesn't have to do the whole thing himself ;)

[19:03:00.0000] <jcranmer>
IIUC, HTML 5 sandboxing works like this:

[19:03:01.0000] <jcranmer>
<iframe src="yada yada yada" sandbox="" /> ?

[19:04:00.0000] <jcranmer>
and, ideally, no matter how gobbeldy-gook (sp?) the source is, how virulent the JS, it shouldn't be able to mess up anything else in the document?

[19:05:00.0000] <takkaria>
jruderman: hmm, I've been wondering about implementing HTML5 in gecko, I don't know anything of gecko's yet yet though

[19:05:01.0000] <takkaria>
or, er, C++

[19:05:02.0000] <takkaria>
s/gecko's yet/gecko's code/

[19:05:03.0000] <jcranmer>
takkaria: layout and content are >50% of where it's at

[19:06:00.0000] <takkaria>
mm

[19:06:01.0000] <takkaria>
I would be especially interested in writing another HTML parser. :)

[19:06:02.0000] <jcranmer>
there

[19:06:03.0000] <jcranmer>
's also a parser/htmlparser, but I can't profess knowledge about that, including, critically, its use

[19:07:00.0000] <jruderman>
takkaria: join #developers on irc.mozilla.org and look for mrbkap and bz

[19:07:01.0000] <takkaria>
I'll remember that for later

[19:07:02.0000] <jcranmer>
dbaron or roc would probably work well, too

[19:07:03.0000] <takkaria>
it's a few months away before I'll be up for doing anything of that sort

[19:07:04.0000] <jruderman>
takkaria: they can tell you what they know about the existing html parser, how it hooks into the rest of gecko, and how much of it is likely to need rewriting

[19:08:00.0000] <jruderman>
https://bugzilla.mozilla.org/show_bug.cgi?id=373864

[19:09:00.0000] <takkaria>
jruderman: also, thanks for the burning edge, I started reading it about roughly when it started and it's still in my feedreader. :)

[19:10:00.0000] <jruderman>
cool, so i didn't lose all my readers when i went dark for a month? ;)

[19:10:01.0000] <jcranmer>
I'm just the guy who works on mailnews code which rivals gsnedders in age

[19:10:02.0000] <jcranmer>
(some of which, in any case)

[19:11:00.0000] <takkaria>
mm, crufty C++

[19:12:00.0000] <jcranmer>
I *wish* it were C++

[19:12:01.0000] <jcranmer>
it's written in yet-another-person-implementing-C++-in-C-code

[19:13:00.0000] <jcranmer>
and the other part is written in C++ by someone who thought that 15 parameters was too few for a function

[19:14:00.0000] <takkaria>
my other programming project is a roguelike game, written in C, whose code in places still looks like the VMS Pascal it was converted from tenety years ago

[19:14:01.0000] <takkaria>
s/tenety/twenty/

[19:22:00.0000] <Lachy>
http://arstechnica.com/news.ars/post/20080801-newly-found-hybrid-attack-embeds-java-applet-in-gif-file.html

[19:23:00.0000] <Lachy>
if that exploit is merely renaming a java applet with a .gif file extension, and then relying on the fact that browsers ignore content type for <applet>, then I won't be surprised.

[19:24:00.0000] <Lachy>
but if that's all it is, then it could hardly be considered an exploit

[19:25:00.0000] <Lachy>
so the bigger question is, what markup needs to be used on the client side to make it run the file as a java applet, instead of an image, if it isn't <applet> or <object>?

[02:40:00.0000] <jacobolus>
Hixie: re: WebSocket: did you ever read http://tools.ietf.org/html/rfc3093 ??

[02:40:01.0000] <jacobolus>
it strikes me that we are finally making the dream a reality!

[04:06:00.0000] <Lachy>
Wow! http://adweek.blogs.com/adfreak/2008/07/then-well-grab.html

[04:09:00.0000] <webben>
hah :)

[04:20:00.0000] <Hixie>
i posted that image in #whatwg a few weeks back :-0

[04:21:00.0000] <Lachy>
ok

[04:21:01.0000] <Hixie>
:-), even

[04:22:00.0000] <Lachy>
I tried to find the right letters in character palette. If these are the right ones: 䬸厅 then google translated the second symbol translates to office, but it couldn't translate the first.

[04:24:00.0000] <Hixie>
what's the codepoint of the first?

[04:24:01.0000] <Hixie>
look it up in the unihan table

[04:25:00.0000] <Hixie>
there are some translations there

[04:28:00.0000] <Lachy>
I'll have to find it again

[04:28:01.0000] <Lachy>
U+4B38 U+5385

[04:29:00.0000] <Lachy>
where do I find the unihan table?

[04:33:00.0000] <heycam>
the first means "meal"

[04:33:01.0000] <heycam>
(i think)

[04:33:02.0000] <Lachy>
actually, I'd found the wrong letter. 餐厅 translates to restaurant.

[04:34:00.0000] <Lachy>
U+9910 looks very similar to U+4B38

[04:35:00.0000] <Lachy>
individually, the 2 symbols translate to "meal" and "office", respectively

[04:36:00.0000] <Lachy>
I can't find the other 2 symbols in the picture, cause they're obscured by something in front of them

[04:36:01.0000] <heycam>
i'm pretty sure they're the same characters

[04:36:02.0000] <Lachy>
they look slightly different

[04:36:03.0000] <Lachy>
 maybe it's just a different font

[04:36:04.0000] <heycam>
different font

[04:36:05.0000] <heycam>
yeah

[04:37:00.0000] <Lachy>
see, I never know with this weird language they all look alike to me anyway

[04:37:01.0000] <heycam>
:)

[04:40:00.0000] <Philip`>
That's why translation should be performed by competent people, not by non-competent people or non-people :-)

[04:51:00.0000] <Philip`>
(Someone says "The Chinese text on the banner (can1 ting1) is simply a generic term for "dining hall" or "cafeteria"")

[06:40:00.0000] <hsivonen>
Re: HTML5 parsing in C++: the core of the Validator.nu parser is intentionally written in a subset of Java that can be mapped to C++ mechanically.

[07:27:00.0000] <jgraham>
/me wonders what the advantage of alt={Photo} is over title=Photo if the idea is to provide a description

[07:30:00.0000] <Dashiva>
The idea is to satisfy the alt-must-be-mandatory crowd

[07:30:01.0000] <Philip`>
s/crowd/arguments/

[07:32:00.0000] <jgraham>
that can be solved by the sentence "An <img> element must either have an alt attribute providing a textual equivalent for the image, or a title attribute providing a description of the image"

[07:32:01.0000] <jgraham>
(or both)

[07:32:02.0000] <Dashiva>
The arguments can be satisfied without mandatory alt, the crowd cannot

[07:37:00.0000] <Philip`>
jgraham: I wouldn't want Flickr to say <img title=Photo> because it'd result in tooltips saying "Photo" and would make me think it was ugly and stupid

[07:37:01.0000] <jgraham>
Philip`: It could say Photo:My cat playing with string

[07:37:02.0000] <Philip`>
whereas <img alt={Photo}> only looks ugly and stupid to users without images and users with IE <= 7 (I think), and I'm not one of those users so I wouldn't care

[07:41:00.0000] <jgraham>
(I'm also wondering if promoting URIs in classnames and/or attribute names is a good idea. It seems to me that URIs are too unweieldy to use as prefixes and URLs in particular are bad because they rely on the DNS system for registration, even though prefixes are permanent whereas domain registrations only last for a finite period of time)

[07:42:00.0000] <jgraham>
(this complete the set of 3 things that I have wondered about recently)

[07:44:00.0000] <Philip`>
/me wonders whether w3.org will continue being reregistered forever, until DNS eventually dies out

[09:53:00.0000] <takkaria>
voila, I have a build of an HTML5-parsing-algorithm-supporting browser

[09:57:00.0000] <jgraham>
takkaria: Next steps: ship it to a few hundred thousand users with a wide range of interests and get them to file some bugs :)

[09:58:00.0000] <takkaria>
I guess the step after that is "profit!!!"

[09:59:00.0000] <takkaria>
hmm, JFS isn't the filesystem you want if you want a directory containing 36k other directories

[10:58:00.0000] <jgraham>
takkaria: So presumably you finished the libxml2 interface? How hard would it be to make a libxml2 build that used hubbub rather than the built-in thing to do the HTML parsing?

[11:17:00.0000] <takkaria>
jgraham: I can't imagine that hard at all

[11:17:01.0000] <takkaria>
the libxml2 interface is finished, yes, but it's not really that efficient

[11:18:00.0000] <takkaria>
one problem is that hubbub's emitted strings are (pointer,length) pairs, and libxml's are NUL-terminated strings

[11:18:01.0000] <takkaria>
so in order to set properties, some kind of strndup() has to be performed

[11:19:00.0000] <takkaria>
which is rather unfortunate

[11:20:00.0000] <takkaria>
this is what the timings look like at the moment: http://pastebin.com/m35aeca22

[11:42:00.0000] <jgraham>
So hubbub is a factor of ~3 slower than libxml2? Room for improvement but not bad for new code

[12:20:00.0000] <takkaria>
jgraham: by adding a really basic speedup to the test treebuilder I'm using, hubbub is 1.8s on html5, and I suspect there will be other simple optimisations that make it faster

[12:28:00.0000] <Lachy>
if I'm reading the URLs section of the spec right, then given href="http://example.com/?foo=&#x4B38;" in a document encoded as ISO-8859-1, then that resolves to "http://example.com/?foo=?" because &#x4B38; can't be encoded in that encoding. Is that correct?

[12:30:00.0000] <Lachy>
but in a UTF-8 document, it would be resolved to "http://example.com/?foo=%E9%A4%90"

[14:38:00.0000] <Hixie>
jgraham_: the advantage is that the UA can provide a clear indication that there is an image

[14:38:01.0000] <Hixie>
jgraham_: which it doesn't have to do if there is alt text normally

[14:40:00.0000] <Philip`>
Hixie: The UA could do the same if the spec defined the appropriate behaviour for <img title=...> instead of <img alt={...}>

[14:42:00.0000] <hsivonen>
jgraham_: thanks for pointing out the existing translation hints that translation engines could be using (but don't)

[14:44:00.0000] <hsivonen>
perhaps the tendency of people to look for layout features, accessibility features and translation control features by caegory is evidence that the semantic markup thing isn't going all that well

[14:44:01.0000] <Hixie>
Philip`: The title="" gives the caption/credit/etc of the image, not the kind of image

[14:44:02.0000] <hsivonen>
s/caegory/category/

[14:54:00.0000] <webben>
hsivonen: I think it's probably more a sign of HTML being used as a UI language as well as a content language.

[14:54:01.0000] <webben>
at least for accessibility features

[15:31:00.0000] <hsivonen>
http://nothing-more.blogspot.com/2004/10/loving-and-hating-xml-namespaces_21.html is particularly interesting, because the author has been "a Lead Developer, lording over MSXML and System.Xml development"

[15:57:00.0000] <Lachy>
still reading that, but it doesn't appear to be a major problem editing namespaces in places where prefixes really are effectively meaningless, such as within the DOM. But the big problems occur when you want to serialise it, where the prefixes suddenly become meaningful and there are conflicts

[16:00:00.0000] <Lachy>
some of the problems could possibly have been mitigated if prefixes had to be declared document-wide, and had to be unique

[16:05:00.0000] <Dashiva>
Wouldn't that ruin the whole "put some xml inside some other xml" thing?

[16:09:00.0000] <Lachy>
how so?

[16:10:00.0000] <Lachy>
you could still declare multiple namespaces, but it would prevent silly things like this <x:foo xmlns:x="a"><x:bar xmlns:x="b"/></x:foo>

[16:10:01.0000] <Dashiva>
Like if you're doing a feed, you'd have to extract all the namespaces used, and solve potential conflicts

[16:10:02.0000] <Lachy>
yes, you would have to ensure there are no conflicts

[16:11:00.0000] <Lachy>
but you get other problems, as outlined in that blog entry, when conflicts arise

[16:11:01.0000] <jgraham>
Hixie: Why would you want to put the caption / credit /etc. in @title rather than in a user-visible form using @figure

[16:11:02.0000] <Dashiva>
And while you could say it's not a supported feature, automatic renaming for conflicts would break things relying on the prefix

[16:12:00.0000] <jgraham>
In practice, i think that Philip`'s objection about tooltips is stronger, although given the IE behaviour with alt, not so much stronger

[16:12:01.0000] <Lachy>
even with namespaces as they are today, nothing should really ever rely on the prefix

[16:12:02.0000] <Dashiva>
Yeah, but that's very violated should

[16:13:00.0000] <Philip`>
jgraham: Because having a user-visibly-by-default caption/credit/etc is really ugly in cases like http://diveintomark.org/archives/2007/03/05/cc-dogs

[16:13:01.0000] <Philip`>
s/visibly/visible/

[16:14:00.0000] <jgraham>
Hixie: I think the fact that there is an <img> tag allows the UA to indicate that there is an image. Once you know that there is an image and there isn't any textual alternative the title seems like it will provide the mst information about what you are missing out on

[16:14:01.0000] <Dashiva>
Lachy: Certain feed readers only working with no prefix or atom: to mention one ;)

[16:16:00.0000] <jgraham>
Philip`: The caption/credit on that page /is/ user visible. Even if you wanted to hide it @title wouldn't help, something like <details> would. 

[16:17:00.0000] <jgraham>
(the fact that @title is used to indicate what happens when you click the image rather than to give a title is a bit sad but I'm pretty convinced that a screenreader user could work out what was going on on that page without any alt text)

[16:18:00.0000] <Philip`>
jgraham: It is visible, and hence it is really ugly, and so it'd be nicer if it wasn't visible

[16:19:00.0000] <Philip`>
and @title would help if it was split into lots of <img>s, or used an image map or whatever

[16:19:01.0000] <jgraham>
Philip`: I think it's a really poor example of where title might help :)


2008-08-04
[18:51:00.0000] <Philip`>
takkaria: Why shouldn't <h a='&COPY'> get turned into U+00A9?

[18:53:00.0000] <takkaria>
Y is U+0059

[18:53:01.0000] <takkaria>
"If the character reference is being consumed as part of an attribute, and the last character matched is not a U+003B SEMICOLON (;), and the next character is in the range ... U+0041 LATIN CAPITAL LETTER A to U+005A LATIN CAPITAL LETTER Z, .... then, for historical reasons, all the characters that were matched after the U+0026 AMPERSAND (&) must be unconsumed, and nothing is returned."

[18:54:00.0000] <Philip`>
It matches &COPY and then the last character is Y which is not ;, and the next character is ' which is not in those ranges so it doesn't get unconsumed

[18:55:00.0000] <takkaria>
ah, I've gone and misread the spec, haven't I?

[18:55:01.0000] <takkaria>
there's a good reason people go to bed before 2am, I see

[18:55:02.0000] <Philip`>
It's still before 2am now :-)

[18:56:00.0000] <Philip`>
There are good reasons to not go to bed before 2am too, such as having to write several hundred words before a deadline tomorrow

[19:00:00.0000] <takkaria>
yeah, agreed, but I think my habit of coding well into the night is not helpful in terms of the number of mistakes I make. :)

[19:03:00.0000] <Philip`>
You'll have to compute the tradeoff between increased productivity due to spend more time working on it, and decreased productivity due to having more bugs

[19:12:00.0000] <takkaria>
I find I get more done in the night but it's of lower quality

[19:12:01.0000] <takkaria>
during the day I take more care over details

[19:14:00.0000] <Philip`>
If you only worry about details then you'll never get any of the big stuff done at all

[19:14:01.0000] <takkaria>
yeah, and for this reason I tend to prefer coding at night

[19:14:02.0000] <jcranmer>
quality, quantity, and time of day have no correlation for me

[19:15:00.0000] <Philip`>
The internet didn't become popular by being very carefully coded and engineered in every detail

[19:15:01.0000] <jcranmer>
the key correlation appears to be between quality and quantity as dependent variables on the independent variable of "time since I started work on this"

[19:15:02.0000] <Philip`>
so clearly you should just hack everything together as quickly as possible :-)

[19:16:00.0000] <jcranmer>
where quantity and quality start high at t = 0, quality starts to dip, followed by quantity some time later, followed by both going to 0

[19:16:01.0000] <jcranmer>
followed by a meteoric rise in quality and then a similar rise in quantity, with both forming a not-quite-in-phase cyclical graph

[19:17:00.0000] <jcranmer>
Quantity = max(A * cos (B * x) + C, 0)

[19:17:01.0000] <jcranmer>
Quality = max(A * cox (B * x + D) + C, 0)

[19:18:00.0000] <jcranmer>
s/(?<\w)x/t/

[19:18:01.0000] <takkaria>
interesting

[19:19:00.0000] <jcranmer>
s/A/A(t)/, where A(t) is a function which smooths out the peaks somewhat

[19:19:01.0000] <jcranmer>
and B(t) is a function which induces non-constant recurrences between peaks and troughs

[19:19:02.0000] <jcranmer>
and C = 0

[19:20:00.0000] <takkaria>
:)

[19:20:01.0000] <Philip`>
So you're basically saying that quantity and quality are defined by the undefined arbitrary functions A and B? :-)

[19:20:02.0000] <jcranmer>
A and B as constants are rough approximations

[19:22:00.0000] <jcranmer>
just like elementary physics problems use the approximation sin(x) = x for small values of x to simplify a differential equation to a linear one

[19:24:00.0000] <Philip`>
What frame of reference are you measuring t in? If you travelled in a relativistic spaceship, would you be any more productive?

[19:25:00.0000] <jcranmer>
Philip`: my frame of reference

[19:25:01.0000] <jcranmer>
or, should I say, the frame of my internal clock

[19:26:00.0000] <Philip`>
What would happen if you surgically extracted your internal clock and put it on the spaceship?

[19:26:01.0000] <jcranmer>
which experiences relativistic effects just from flying in an airplane :-)

[19:26:02.0000] <jcranmer>
(although neither General nor Special Relativity, but Joshua Relativity)

[19:29:00.0000] <Hixie>
jgraham: there are many cases where you want the image to just not be mentioned

[19:29:01.0000] <Hixie>
jgraham: and just replaced with its alternative text

[22:56:00.0000] <Hixie>
reading through alt="" e-mail, it seems my life would be a lot easier if i was an accessibility expert

[22:56:01.0000] <Hixie>
i could just say things without having to defend them with data

[22:57:00.0000] <Hixie>
i wonder how i become an accessibility expert

[22:57:01.0000] <roc>
gouge your own eyes out

[22:58:00.0000] <Hixie>
the first hit for "how to become an accessibility expert" on google is a powerpoint presentation "Become an Accessibility Expert in 50 min"

[22:58:01.0000] <Hixie>
no, many accessibility experts are not blind

[22:58:02.0000] <roc>
I didn't say it was the only way

[22:59:00.0000] <roc>
I bet you can do it in less than 50 minutes

[02:11:00.0000] <Dashiva>
Darths & Droids comes out in favor of text equivalents not being equivalent enough. I wonder if anyone will cite that :)

[02:56:00.0000] <jgraham>
Hixie: If you  just want the image replaced by its alternate text i follows you know what the alternate text is so you put it in the alt attribute and there's no problem

[02:57:00.0000] <Hixie>
jgraham: my point is that you wouldn't necessarily even want the UA to (by default) inform the user that there is an image in that case

[03:05:00.0000] <jgraham>
Hixie: I don't disagree with that. It's the case where there is no alt attribute where a UA could announce there is an image and read the title attribute

[03:06:00.0000] <Hixie>
jgraham: it could always read the title attribute, that seems orthogonal

[03:08:00.0000] <jgraham>
Not really. If I have <img alt="XYZ Systems" title="XYZ systems logo"> I don't expect the UA to read both @alt and @title

[03:10:00.0000] <jgraham>
Whereas for <img title="XYZ Systems Logo"> reading Image, XYZ Systems Logo seems quite helpful

[03:10:01.0000] <Hixie>
<img title="XYZ Systems Logo"> is non-conforming

[03:10:02.0000] <Hixie>
and in both cases, the title="" attribute's value sucks

[03:11:00.0000] <jgraham>
We get o choose what is non-conforming :) 

[03:11:01.0000] <jgraham>
Also, why does it suck?

[03:11:02.0000] <Hixie>
i chose it to be non-conforming :-) it has no reason to be conforming

[03:11:03.0000] <Hixie>
the title attribute doesn't help anything

[03:12:00.0000] <Hixie>
in that case

[03:12:01.0000] <Hixie>
i mean, it's not _wrong_

[03:12:02.0000] <jgraham>
(it should be non conforming in this case because there is obvious alt text. But thre are cases like flickr where the photo is likely to have a title but no alternate text avaliable)

[03:13:00.0000] <Hixie>
yes, in those cases it would be fine for the UA to mention that there is potentially useful title="" text

[03:13:01.0000] <Hixie>
that's just part of whatever helpful stuff it wants to give

[03:14:00.0000] <jgraham>
What other helpful stuff does it want to give?

[03:15:00.0000] <jgraham>
The only other information likely to be avaliable is a category of image like Photo or Logo or whatever

[03:16:00.0000] <jgraham>
Is that really useful enugh to invent a whole microsyntex for alt that causes it to duplicate much of the functionality of @title?

[03:16:01.0000] <Hixie>
it might know what the image is through image analysis (e.g. it might notice the image is fully transparent or 1x1) or it might notice that the image is one that it has seen before on another page where it had alt="" text, or it might be one that it recognises has a face in it, or any number of other things

[03:16:02.0000] <Hixie>
i don't think alt="{photo}" is information that would ever be appropriate in title=""

[03:22:00.0000] <jgraham>
Hixie: if the UA notices something through image analysis it seems fine to present that alongside / instead of @title (assumng no @alt). I'm not sure that changes the fact that <img alt={photo}> isn't providing much useful informaion and, in particular is probably less useful than an image with a good title (which is often easier to enter in a CMS than a text alternative)

[03:22:01.0000] <Hixie>
i agree that the title, if available, would be fine information to make available

[03:22:02.0000] <Hixie>
in most cases (e.g. flickr) i imagine all such information would be on the page and not in the title=""

[03:23:00.0000] <Hixie>
title="" makes more sense for images that aren't the main topic of the page, in which case the alt="" text is usually available anyway

[03:24:00.0000] <Lachy>
Hixie, could you answer this question for me? http://krijnhoetmer.nl/irc-logs/whatwg/20080803#l-507

[03:24:01.0000] <jgraham>
I guess if there really is no good alt text or any good title, having some way to indicate the type of the image might be useful though I would expect it to be wrong much of the time (consider sites that allow users to upload photos or 3D renerings)

[03:25:00.0000] <jgraham>
Hixie: flickr puts the title of the photo in @title

[03:25:01.0000] <Hixie>
Lachy: "yes"

[03:25:02.0000] <Hixie>
Lachy: (is the answer to the question. unknown characters become "?".)

[03:26:00.0000] <Hixie>
jgraham: yeah, it also puts it in alt="", and in the header, and all over the place

[03:26:01.0000] <Lachy>
Hixie, ok. Cause that didn't match the behaviour of Firefox or Safari. Was that intentional? (I didn't test IE yet)

[03:26:02.0000] <Hixie>
jgraham: listening to that site in a screen reader is like being in an echo chamber

[03:26:03.0000] <Hixie>
Lachy: they convert unknown chars to &12345; right?

[03:26:04.0000] <Hixie>
Lachy: oh wait

[03:26:05.0000] <Hixie>
Lachy: firefox switches to UTF-8

[03:26:06.0000] <Philip`>
Lachy: Firefox seems crazy and switches between document-encoding and UTF-8 depending on what characters there are

[03:26:07.0000] <Hixie>
right?

[03:27:00.0000] <Hixie>
and the others se &12345; or something

[03:27:01.0000] <Hixie>
or is that just for form submission

[03:27:02.0000] <Philip`>
Safari changed recently, so nightly versions of WebKit probably do something different to 3.1

[03:27:03.0000] <Lachy>
Hixie, demo: http://tinyurl.com/6na2xh

[03:27:04.0000] <Hixie>
anyway either way iirc i looked at what they did, decided they were all on crack, and made the spec say "?" because that's the only thing that makes sense

[03:27:05.0000] <Hixie>
yeah safari does the &12345; thing, firefox does utf-8

[03:27:06.0000] <Hixie>
both are insane

[03:28:00.0000] <Lachy>
ok

[03:28:01.0000] <Philip`>
http://lists.w3.org/Archives/Public/public-html/2008Jun/0426.html

[03:28:02.0000] <Hixie>
(we might switch the spec to &12345; in cr, though i will fight it all the way :-) )

[03:29:00.0000] <Lachy>
ok. I thought Firefox converting to UTF-8 seemed reasonable

[03:29:01.0000] <Philip`>
What about character sets that don't contain the "?" character?

[03:29:02.0000] <Hixie>
Lachy: on the server side you have no idea what the encoding is if it changed, so how are you supposed to tell?

[03:29:03.0000] <Hixie>
what's a valid use case for splitting an image into two or more pieces?

[03:29:04.0000] <Hixie>
Philip`: like what?

[03:30:00.0000] <Philip`>
Hixie: Like a hypothetical one I could make up to be awkward

[03:30:01.0000] <Hixie>
/me goes to add that encoding to the spec in the "must not support" column

[03:30:02.0000] <Hixie>
(hypothetically)

[03:31:00.0000] <Lachy>
Hixie, I just thought if it were defined to be UTF-8 for a specified set of document encodings, or else the document encoding for anything else then the server would know by looking at what the spec told it to expect

[03:32:00.0000] <Lachy>
but I don't mind it converting to ?, if that's compatible with the web

[03:32:01.0000] <Philip`>
Hixie: <p>We have had <img src=counter/1.gif><img src=counter/4.gif><img src=counter/3.gif> visitors since 1998</p>, since it's often unacceptable to require server-side image generation (because that's complex and inefficient), where the appropriate alt text is a single string "143" and is not three distinct digit strings

[03:33:00.0000] <Hixie>
lachy: what firefox does is look at the characters in the url and decide on the encoding based on whether it can convert them or not

[03:33:01.0000] <Hixie>
Lachy: it's not based on the document encoding per se

[03:33:02.0000] <Lachy>
oh, ok. That's definitely insane

[03:33:03.0000] <Hixie>
Philip`: any other cases?

[03:34:00.0000] <Philip`>
Hixie: Do you count slicing images to create image maps (without using <map>, because e.g. you want one slice to be replaced dynamically when you move the mouse over it) to be a valid use case?

[03:35:00.0000] <Hixie>
no, because in those cases you'd want the alternative text in each image, not all in one

[03:36:00.0000] <Philip`>
Repeat the whole text in each <img>? That sounds like it would be annoyingly repetitive to anyone who is reading the page without images

[03:37:00.0000] <Hixie>
no, you'd want each link to have its appropriate text

[03:39:00.0000] <Philip`>
What if the image is a smiley face and the nose changes colour when you move the mouse over it, and so the image is sliced up so it only has to download the small nose section when it updates? The appropriate alt text for the whole thing would be something like "Smiley face" or whatever, and there wouldn't be appropriate text for each individual <img>

[03:42:00.0000] <Lachy>
woah, Firefox will encode all characters as UTF-8 if there are any non-ISO-8859-1 characters elsewhere in the query string. e.g. ?&copy; vs. ?&#x4B38;&copy;

[03:42:01.0000] <Hixie>
Lachy: correct

[03:42:02.0000] <Hixie>
Philip`: seems contrived, but ok :-) thanks

[03:43:00.0000] <Philip`>
There are probably better examples but I can't think of them right now :-p

[04:44:00.0000] <nzkoz>
hsivonen: Are you around?

[04:45:00.0000] <hsivonen>
nzkoz: yes

[04:46:00.0000] <nzkoz>
hey, if you have a moment I have a few questions about the validator.nu service, I've noticed some changes in the responses lately (since its downtime) 

[04:47:00.0000] <nzkoz>
Things like http://xrl.us/omobx which now only give a few errors (one of which is fatal( where previously it enumerated quite a few more?

[04:48:00.0000] <nzkoz>
was there a major change that happened that I should do some reseearch on?

[04:49:00.0000] <hsivonen>
nzkoz: I fixed paragraph end inference inside list items, at least

[04:50:00.0000] <hsivonen>
nzkoz: do you remember what the old errors were about?

[04:51:00.0000] <hsivonen>
nzkoz: I haven't changed the XHTML 1.0 schemas lately except to tweak xml:space issues.

[04:51:01.0000] <nzkoz>
let me dig up an old response

[04:54:00.0000] <Hixie>
why aren't my e-mails making the list

[04:54:01.0000] <Hixie>
lists

[04:54:02.0000] <Hixie>
why aren't my e-mails going anywhere

[04:54:03.0000] <Hixie>
wtf

[04:54:04.0000] <nzkoz>
hsivonen: previously we got back: Attribute “xml:lang” not allowed on element “html” at this point.

[04:54:05.0000] <nzkoz>
then additional errors

[04:54:06.0000] <nzkoz>
now that seems to be sufficient to stop parsing?

[05:00:00.0000] <nzkoz>
hsivonen: yeah, at least in the 4 old responses I've dug up, the difference appears to be it stops reporting errors after it hits an xml:lang, previously it continued

[05:01:00.0000] <hsivonen>
nzkoz: oh. good point. I seem to have a bad config for XML violations.

[05:01:01.0000] <hsivonen>
nzkoz: thanks. I'll fix.

[05:03:00.0000] <nzkoz>
hsivonen: cheers, glad I could help.  I'm scripting the json api and it's really useful.  We're using gzip compression and setting a user agent, I assume you'd notify us if it were causing trouble?

[05:04:00.0000] <hsivonen>
nzkoz: yeah, that's OK.

[05:11:00.0000] <Hixie>
1893 pending messages!

[05:11:01.0000] <tusho>
    <Hixie> i think i might make alt=”" required and say that when you don’t know what the image is, you have to say what kind of image it is (e.g. “uploaded image”, “photo”, “thumbnil”, or whatever) and put that in braces in the alt=”" attribute, as in alt=”{photo}”

[05:11:02.0000] <tusho>
seems reasonable

[05:11:03.0000] <tusho>
how does it interact with screen readers

[05:12:00.0000] <Hixie>
they say "image photo" or some such probably

[05:12:01.0000] <hsivonen>
does anyone happen to remember which element don't have class/id allowed in XHTML 1.0 and HTML 4.01?

[05:13:00.0000] <tusho>
Hixie: they pronounce punctuation

[05:13:01.0000] <tusho>
I believe

[05:13:02.0000] <tusho>
image opening curly brace photo closing curly brace

[05:13:03.0000] <Hixie>
hsivonen: i sent you an e-mail with an archive link which is not accurate -- when you get the mail, then look around the link i gave for the actual link

[05:14:00.0000] <Hixie>
tusho: well by the time people are doing this on any kind of regular basis, hopefully they'll have caught up with the spec

[05:14:01.0000] <Philip`>
tusho: In the default configuration?

[05:14:02.0000] <tusho>
Hixie: now THAT'S wishful thinking :)

[05:14:03.0000] <tusho>
Philip`: i -think- so

[05:15:00.0000] <Hixie>
this is all for the case where the accessibility story is a lost cause anyway -- the whole point is we have no useful alternative text

[05:15:01.0000] <tusho>
Hixie: decoration

[05:15:02.0000] <hsivonen>
Hixie: ok

[05:15:03.0000] <tusho>
often basically useless to the content images are used to spruce up an article

[05:16:00.0000] <tusho>
for a screen reader, they should just be ignored

[05:16:01.0000] <tusho>
thus alt=""

[05:16:02.0000] <Hixie>
i don't understand why my scripts can send mail but i can't

[05:16:03.0000] <Hixie>
wtf

[05:16:04.0000] <Hixie>
tusho: spec says they should al be alt="" (empty alt), unaffected by this {photo} thing

[05:17:00.0000] <tusho>
Hixie: but you said you might make alt= required

[05:17:01.0000] <Hixie>
for decorative images, it is required, and is required to be empty

[05:17:02.0000] <tusho>
oh, I see

[05:17:03.0000] <tusho>
I misread the quote

[05:17:04.0000] <tusho>
I thought you meant making the alt attribute being mandatory and non-empty

[05:17:05.0000] <Hixie>
no

[05:18:00.0000] <tusho>
ok :)

[05:18:01.0000] <Philip`>
Calling it 'the alt="" attribute' is confusing, since that looks the same as saying decorative images must have alt=""

[05:19:00.0000] <hsivonen>
nzkoz: fixed. (the errors about the id attribute not being allowed in the XHTML 1.0 schema may be bogus. I'll investigate.)

[05:21:00.0000] <nzkoz>
hsivonen: awesome, thanks a bunch.  For future reference is this the best way to raise anything else I notice? or should I use the bugzilla  or...?

[05:22:00.0000] <hsivonen>
nzkoz: bugzilla.validator.nu is the preferred way, since there's a risk of IRC remarks falling through the cracks

[05:24:00.0000] <nzkoz>
k, will do.  Thanks again

[05:31:00.0000] <Philip`>
Hixie: "... one of the images must its alt attribute set ..." is missing a word

[05:32:00.0000] <Hixie>
fixed

[05:32:01.0000] <Hixie>
thanks

[05:42:00.0000] <Hixie>
wish i knew why my e-mail wasn't going anywhere

[05:42:01.0000] <Hixie>
oh well

[05:42:02.0000] <Hixie>
bed time

[05:42:03.0000] <Hixie>
hopefully e-mail will make its way out while i'm sleeping

[05:46:00.0000] <Philip`>
I got commit notices for r1971, 1973, 1983 and 1985, and none of the ones in between

[05:51:00.0000] <mcarter>
hello

[05:53:00.0000] <mcarter>
I don't suppose anyone has any idea what is supposed to happen (if there is a standard) when an HTTP/1.1 XHR response is half over when there is a page navigation. Should it a) shut the connection immediately, or b) wait for the rest of the response (though now irrelevant), and then re-use the connection

[09:37:00.0000] <gDashiva>
I went in the month-change archive trap -again-

[09:37:01.0000] <gDashiva>
s/went/fell

[09:38:00.0000] <Philip`>
It's an effective way to get an occasional respite from the list discussions

[11:19:00.0000] <zcorpan>
aha.. can we resolve the extensibility problem by making document.write work in xml?

[11:25:00.0000] <hsivonen>
/me notes that the XML community already has its own document.write(). It's called disabling output escaping in XSLT.

[11:25:01.0000] <gDashiva>
I thought the XML community didn't want document.write

[11:26:00.0000] <tusho>
document.write = no thanks

[11:26:01.0000] <Philip`>
You could implement by having the server only transmit bytes up to the end of </script>, then having document.write() send an XMLHttpRequest to the server to tell it what bytes it should subsequently transmit to continue the XML document

[11:26:02.0000] <tusho>
a nicer layer over the DOM = yez plz

[11:26:03.0000] <tusho>
mochikit style that is

[11:26:04.0000] <Philip`>
s// it/

[11:26:05.0000] <tusho>
p(strong("hello"), " world")

[11:27:00.0000] <tusho>
maybe: with (DOM) { ... }

[11:27:01.0000] <Philip`>
tusho: Why not use thing.innerHTML = '<p><strong>hello</strong> world</p>' for that?

[11:27:02.0000] <tusho>
Philip`: The DOM is less prone to errors and programmatically cleaner.

[11:27:03.0000] <tusho>
Plus you can't construct an invalid tree with it.

[11:31:00.0000] <zcorpan>
tusho: can't construct an invalid tree with innerHTML either

[11:31:01.0000] <tusho>
ok, wrong words

[11:31:02.0000] <tusho>
still

[11:31:03.0000] <tusho>
my other points stand

[11:34:00.0000] <zcorpan>
i'm not convinced that p(strong("hello"), " world") or  (DOM) { ... } is less prone to error than innerHTML, and programmatically cleaner is just an opinion :)

[11:35:00.0000] <Philip`>
zcorpan: Try setting attributes in innerHTML and having to escape and concatenate everything properly, and then it should be clear which way is cleaner and less error-prone :-)

[11:37:00.0000] <Lachy>
I'm opposed to introducing document.write() into XML. It's unfortunate that it even exists for HTML

[11:37:01.0000] <Philip`>
But it's really useful

[11:37:02.0000] <hsivonen>
Lachy: with a well defined XML5 parsing algorithm including script execution steps, we could define document.write() for XML

[11:37:03.0000] <Lachy>
it encourages poorly structured programming practices

[11:38:00.0000] <Philip`>
It encourages people to write HTML code that works and does what they want and makes them happy, even if someone might say it's poorly structured :-)

[11:39:00.0000] <Lachy>
and makes people work with the DOM by manipuating strings, instead of actually working with the DOM properly, and the fact that it injects characters back into the input stream is bad design

[11:40:00.0000] <Philip`>
Working with the DOM "properly" is really ugly and painful, so it's not surprising that everyone tries to avoid that

[11:40:01.0000] <Lachy>
innerHTML is reasonable, because at least it doesn't mess with the input stream

[11:41:00.0000] <Lachy>
and keeps markup errors in the string isolated to just the element it's appeneded to

[11:41:01.0000] <Philip`>
But then you need to create a container element to set innerHTML, and give it an ID and document.getElementById('whatever') it and think of a new ID for every time you do this, and try not to get confused by all the indirection

[11:42:00.0000] <Lachy>
to e.g. document.write() with an unclosed inline element will affect the rest of the document, unlike how it would work in innerHTML

[11:42:01.0000] <Lachy>
if you create a container element, then you don't need an ID, you already have a reference to it.

[11:42:02.0000] <gDashiva>
The DOM kinda shot itself in the foot by making long, unwieldy identifiers a feature

[11:43:00.0000] <Lachy>
if there's one already in the DOM, then sure, gEBId (or other selector) is needed

[11:43:01.0000] <Lachy>
also, document.write() violates the separation of script and markup principles

[11:43:02.0000] <hsivonen>
gDashiva: the DOM shot itself in many places in many ways

[11:44:00.0000] <Lachy>
gDashiva, agreed

[11:44:01.0000] <Philip`>
Those separation principles rely on indirection, and hence are confusing and annoying

[11:44:02.0000] <Lachy>
your arguments are confusing and annoying

[11:45:00.0000] <Philip`>
That's quite possible

[11:45:01.0000] <Lachy>
there's nothing wrong with a little bit of indirection. It makes code more modular and allows for better reuse

[11:45:02.0000] <gDashiva>
/me laments the lost opportunity for "So's your face"

[11:46:00.0000] <Lachy>
:-D

[11:46:01.0000] <gDashiva>
So instead of document.write, what about a document.emitNode() or something, would that make people happy?

[11:47:00.0000] <Lachy>
how would it work?

[11:47:01.0000] <Philip`>
document.writeWellFormedFragment()?

[11:47:02.0000] <Lachy>
would it append the given node to the script's parent element?

[11:47:03.0000] <gDashiva>
Lachy: Yeah

[11:48:00.0000] <Lachy>
what if the <script> is the root element?

[11:48:01.0000] <gDashiva>
Then you deserve whatever is coming 

[11:48:02.0000] <gDashiva>
Or you could specify it to replace the script element

[11:48:03.0000] <zcorpan>
Lachy: you do the same as document.appendChild() would do

[11:49:00.0000] <Lachy>
what if it's a DocumentFragment, with multiple child nodes? then it still would solve the root element problem/

[11:49:01.0000] <Lachy>
/me goes to test how document.appendChild works in browsers

[11:49:02.0000] <zcorpan>
again, do the same as fragment.appendChild() would do :)

[11:50:00.0000] <Lachy>
it throws an error

[11:50:01.0000] <gDashiva>
The thing I'm trying to clarify is if the problem is emitting markup vs emitting DOM, or if it's emitting itself that's the problem 

[11:50:02.0000] <Philip`>
What if the script removes its own <script> element from the document, and then calls emitNode()?

[11:51:00.0000] <Lachy>
zcorpan, no, I meant if you did document.emitNode(fragment); where fragment is a DocumentFragment

[11:51:01.0000] <zcorpan>
Philip`: then you do the same as document.appendChild() would do ;)

[11:51:02.0000] <zcorpan>
Lachy: ah

[11:51:03.0000] <hsivonen>
if you don't like doc.write but want to do it, you can compile a parser into JS using GWT and parse strings without source text insertion support in the browser

[11:51:04.0000] <zcorpan>
Lachy: well then you act as if you did document.appendChild() multiple times

[11:51:05.0000] <Lachy>
yeah, that would throw an exception

[11:52:00.0000] <zcorpan>
depends on if there's a root element or not

[11:52:01.0000] <Lachy>
what is GWT?

[11:52:02.0000] <hsivonen>
Lachy: Google Web Toolkit

[11:52:03.0000] <zcorpan>
i guess it would insert the first element and then throw, but perhaps it could be specced so that if it throws, nothing happens

[11:53:00.0000] <Philip`>
hsivonen: But only crazy people would do that

[11:53:01.0000] <Philip`>
hsivonen: and only crazy people who don't care about performance

[11:54:00.0000] <hsivonen>
Philip`: actually, if I manage to ship a JS library for retrofitting HTML5 parsing into Firefox 3, Safari 3.1 and Opera 9.5, I'd expect people to use it as a fallback for the past generation when the then-current generation support HTML5 parsing natively

[11:54:01.0000] <Philip`>
hsivonen: How many tens/hundreds of kilobytes of JS code is it?

[11:54:02.0000] <hsivonen>
if using it is as simple as <body><script src=html5.js></script>

[11:55:00.0000] <hsivonen>
Philip`: 128 KB without gzip

[11:55:01.0000] <Lachy>
heh, Hixie's mail from 10:16 this morning finally arrived about 10 hours late

[11:56:00.0000] <Lachy>
the one about the alt attribute stuff

[11:58:00.0000] <hsivonen>
Philip`: 36 KB with gzip

[11:59:00.0000] <hsivonen>
Philip`: my footprint vs. execution speed choices are meant for Java and C++, so it's not optimal for JS footprint

[12:02:00.0000] <hsivonen>
http://derstandard.at/text/?id=1216918402134 

[12:03:00.0000] <hsivonen>
Miguel seems to prefer Microsoft-controlled environments over JS and SVG

[12:08:00.0000] <Philip`>
That does not seem unexpected

[12:13:00.0000] <zcorpan>
Hixie: what does <img src=x alt={> represent when the ua is configured to not show images?

[12:15:00.0000] <Lachy>
zcorpan, that's covered by the 2nd condition in the spec "If the src  attribute is set and the alt attribute is set to a string with at least one character whose first character is not a U+007B LEFT CURLY BRACKET character ({) or whose last character is not a U+007D RIGHT CURLY BRACKET character (})"

[12:15:01.0000] <Lachy>
that's the confusing one I complained about, cause I had the same question initially, till I figured it out before sending my mail

[12:15:02.0000] <zcorpan>
Lachy: how am i reading it wrong?

[12:16:00.0000] <zcorpan>
the first character IS { so it's not covered by that case

[12:16:01.0000] <Lachy>
because the last character is not '}'

[12:16:02.0000] <Philip`>
"{" has a first character that is not "{" or it has a last character that is not "}", so it is covered by that case

[12:16:03.0000] <zcorpan>
Lachy: it says or, not and

[12:17:00.0000] <Lachy>
no, think of it like this:...

[12:17:01.0000] <zcorpan>
hmm

[12:18:00.0000] <Lachy>
if ( isset(src) && isset(alt) && (alt[first-letter] != '{' || alt[last-letter] != '}'))

[12:18:01.0000] <Lachy>
(that's just some sort of pseudo-code, don't try to run it)

[12:19:00.0000] <Philip`>
/me doesn't know of any languages in which "first-letter" is a valid identifier rather than a subtraction expression :-p

[12:19:01.0000] <Lachy>
Philip`, in CSS it is

[12:19:02.0000] <Lachy>
:-)

[12:19:03.0000] <Philip`>
But CSS doesn't have subtraction expressions, so that doesn't count :-)

[12:20:00.0000] <zcorpan>
doesn't it have calc() ?

[12:20:01.0000] <Lachy>
not yet

[12:20:02.0000] <Lachy>
unless someone specced a proposal for it that I missed.

[12:21:00.0000] <zcorpan>
http://www.w3.org/TR/css3-values/#calc ?

[12:21:01.0000] <Lachy>
oh

[12:21:02.0000] <Philip`>
When I say "CSS" I mean "real CSS like what you can actually use in real life and it'll work in sensible web browsers"

[12:25:00.0000] <Lachy>
Philip`, in that case, you can use IE's expression()

[12:26:00.0000] <hsivonen>
is the execution time of expression well defined?

[12:26:01.0000] <hsivonen>
what if an expression has side effects?

[12:26:02.0000] <Philip`>
Lachy: "sensible web browsers"

[12:26:03.0000] <Lachy>
hsivonen, exression isn't defined at all. It's only poorly documented in MSDN, like all of Microsoft's extensions

[12:26:04.0000] <Philip`>
hsivonen: It isn't

[12:27:00.0000] <Philip`>
You can have side effects, and it'll just do whatever IE's layout engine wants to do

[12:27:01.0000] <Lachy>
hsivonen, I expect that IE would probably crash. I know that when i was forced to use it once, any time I did something mildly complicated, IE crashed

[12:28:00.0000] <hsivonen>
great value for users then

[12:28:01.0000] <Lachy>
it's designed for simple expressions, and for that it works ok. but it's not really well designed or implemented at all

[12:30:00.0000] <Lachy>
plus, it creates an attack vector, if a site allows style attributes to be added by users in comments or whatever, without filtering out expressions

[12:33:00.0000] <Lachy>
hah! I thought the Knights Templar only existed in Indian Jones and the Last Crusade. http://www.theregister.co.uk/2008/08/04/knights_templar_pope/

[12:35:00.0000] <tusho>
Philip`: {doesn't know of any languages in which "first-letter" is a valid identifier rather than a subtraction expression :-p}

[12:35:01.0000] <tusho>
lisp

[12:35:02.0000] <tusho>
dyland

[12:35:03.0000] <tusho>
*dylan

[12:35:04.0000] <Philip`>
tusho: Ah, so it's only crazy languages ;-)

[12:40:00.0000] <Lachy>
http://xkcd.com/297/

[12:40:01.0000] <Lachy>
my pseudo-code above did use a lot of parenthesis too. :-)

[12:44:00.0000] <zcorpan>
+   users or very low-bandwidth connections or who pay by the byte, or

[12:44:01.0000] <zcorpan>
i guess people will laugh at that in 10 years

[12:47:00.0000] <Lachy>
zcorpan, what? no-one has such poor connections these days. Even dial up is relatively cheap

[12:47:01.0000] <Philip`>
zcorpan: Ten years ago when people started using things like ADSL, I imagine they would have laughed at the idea of still having very low-bandwidth connections and paying by the byte in the present day

[12:49:00.0000] <Philip`>
but people are still willing to pay for really rubbish internet connections, mostly since there's no competition and they have no other choice

[12:49:01.0000] <zcorpan>
Philip`: yeah, i'm not sure why we still have very low-bandwidth and pay by the byte :)

[12:49:02.0000] <Lachy>
oh, I didn't realise you were quoting hte spec

[12:49:03.0000] <Lachy>
Hixie probably put it in as a bit of a joke

[12:50:00.0000] <jgraham>
Lots of mobile users have low bandwidth, poor connections and pay by the byte

[12:50:01.0000] <zcorpan>
what jgraham said

[12:50:02.0000] <Lachy>
well, by the kilobyte, maybe

[12:51:00.0000] <Philip`>
People are still willing to pay however much they pay for SMS messages - the price hasn't been driven to zero by the improvements in technology, because what's important is how much the service is worth to people, and that's fairly constant over time

[12:51:01.0000] <Lachy>
I suppose they still charge a lot for mobile and wireless connections these days

[12:52:00.0000] <jgraham>
Philip`: there are lots of plans with hundreds of free texts for only £extortionate monthly rates

[12:52:01.0000] <Lachy>
Philip`, people have been complaining about the price of SMS for a while

[12:52:02.0000] <Lachy>
it's the telcos that just won't listen

[12:52:03.0000] <jgraham>
Lachy: I don't recall that happening in the UK

[12:52:04.0000] <jgraham>
where SMS is very very popular

[12:52:05.0000] <Lachy>
I didn't say it wasn't popular. Just expensive

[12:53:00.0000] <Philip`>
Lachy: Why should they listen?

[12:53:01.0000] <Lachy>
Depending on your plan, it can still cost $0.20 per message

[12:53:02.0000] <Lachy>
...in Australia

[12:53:03.0000] <Lachy>
though, at least some offer either unlimited or hundreds of free SMSs

[12:58:00.0000] <Philip`>
I guess the mobile operators looked at the internet running over standard telephone lines and decided that it was clearly a miserable failure since the telephone companies weren't being paid any more than for standard voice calls, and so they're still trying hard to avoid that mistake

[13:01:00.0000] <jgraham>
Philip`: I guess that's not true in the UK because there does seem to be actual competition between operators

[13:02:00.0000] <jgraham>
(well I mean it might be true that they prefer that scenario, but it's hard to imagine how they will be able to maintain it)

[13:02:01.0000] <jgraham>
s/that scenario/the current scenario/

[13:03:00.0000] <Philip`>
/me knows almost nothing about mobile phones, since he's only ever had one (which is probably four or five years old now) and almost never uses it

[13:04:00.0000] <jgraham>
/me keeps losing his phone

[13:04:01.0000] <jgraham>
(well that is to say I did it and then didn't get another one)

[13:05:00.0000] <jgraham>
(because I hardly ever used it anyway and cool stuff like internet access is only sensibly priced if you also buy an expensive contract)

[13:18:00.0000] <zcorpan>
Hixie: the captca example is missing an end tag

[13:23:00.0000] <zcorpan>
given 4.7.2.1.10., alt is still not always required

[13:49:00.0000] <Hixie>
zcorpan: i agree that the patch isn't one we want, but thanks for the headsp

[13:49:01.0000] <Hixie>
zcorpan: (i trust your judgement in these issues)

[13:55:00.0000] <zcorpan>
Hixie: having many mods seems to be effective against spam

[13:56:00.0000] <zcorpan>
Hixie: we've only had spam once and two mods were on it pretty quickly :)

[13:57:00.0000] <Hixie>
nice

[14:00:00.0000] <zcorpan>
http://www.paciellogroup.com/blog/?p=84

[14:07:00.0000] <Lachy>
"no co-operative exchange occured with WAI prior to the changes..." - Is he choosing to ignore excessive amount of discussion about the whole issue previously on public-html and the wai lists?

[14:09:00.0000] <Lachy>
maybe because this specific proposal wasn't discussed on the list recently, before it was added to the spec. But so what? It's a proposal. At least this gives us something to discuss

[14:09:01.0000] <Hixie>
well i'm glad that we're worried about process rather than what the spec says now

[14:09:02.0000] <Hixie>
that's at least a step in the right direction

[14:11:00.0000] <Lachy>
well, he said he hasn't had time to review it all yet. But I find it annoying having so many bureaucracy trolls

[14:16:00.0000] <Lachy>
I'm pretty sure this proposal will be much more well received than the last, since it actually ends up giving a lot more useful information about the image than something like noalt="", or some other reserved keyword like alt="_unknown"

[14:18:00.0000] <Philip`>
And because people who might have minor objections will keep quiet since they don't want to risk reanimating the issue

[14:19:00.0000] <Hixie>
hah

[14:19:01.0000] <Hixie>
didn't stop lachy! :-P

[14:20:00.0000] <Lachy>
Hixie, I think you should rephrase the "In a rare subset of these cases..." and "...if the site has received 8000 photos...". Those particular phrases caused quite a lot of controversy and don't seem to convey their intended meaning well

[14:20:01.0000] <Hixie>
yeah

[14:20:02.0000] <Hixie>
god point

[14:20:03.0000] <Hixie>
good even

[14:22:00.0000] <Lachy>
I think it should just focus on the specific conditions of when using {...} is acceptable. i.e. when the image is obtained without associated alt text, rather than trying to explain it through examples

[14:24:00.0000] <Hixie>
it also applies to inkblot tests and blind people's photos

[14:26:00.0000] <Lachy>
probably also applies to fixed weather cams or traffic cams, which just take and publish intermittent photographs without user intervention

[14:27:00.0000] <Philip`>
And coffee pot cams

[14:27:01.0000] <Lachy>
haven't seen one of those

[14:28:00.0000] <Philip`>
http://en.wikipedia.org/wiki/Trojan_room_coffee_pot

[14:28:01.0000] <Dashiva>
They're common in highly caffeinated nerd clubs

[14:33:00.0000] <Hixie>
ok Lachy, checked in your suggestions 

[14:35:00.0000] <Hixie>
didn't change the e-mail thing, i don't see any reason why it would matter in that case, and it doesn't affect web page conformance anyway

[14:38:00.0000] <Lachy>
Hixie, missing ')' from the end of: If the src  attribute is set and the alt attribute is set to a value that isn't matched by the previous two entries (not empty, not "{...}"

[14:38:01.0000] <Hixie>
oops

[14:39:00.0000] <Lachy>
and you didn't change it for the other one describing the case for when the src attribute is not set

[14:40:00.0000] <Hixie>
right, see e-mail

[14:41:00.0000] <Hixie>
i have GOT to start taking out the "***DHSPAN***" bits of subject lines i send!

[14:42:00.0000] <Dashiva>
What is that anyway?

[14:42:01.0000] <Lachy>
s/DHSPAN/DHSPAM/ == DreamHost SPAM, I figure

[14:42:02.0000] <Hixie>
yeah

[14:43:00.0000] <Hixie>
people who top-post tend to trigger my last spam line of defence

[14:43:01.0000] <Hixie>
which is spam assassin's "i don't think it's spam but it's suspicious so i'll label it as such anyway"

[14:43:02.0000] <Lachy>
Hixie, the email just says you changed the order of them, but you only did it for one. Since they conditions are the basically the same, except one is for when src is present and the other for when it's not, they should be written in almost the same way.

[14:44:00.0000] <Lachy>
so it should be changed to "If the src attribute is *not* set and the alt attribute is set to a value that isn't matched by the previous two entries (not empty, not "{...}" )"

[14:45:00.0000] <Hixie>
oh!

[14:45:01.0000] <Hixie>
i see what you're saying

[14:45:02.0000] <Hixie>
my apologies

[14:45:03.0000] <Lachy>
what did you think I was saying?

[14:47:00.0000] <Lachy>
"...display some sort of indicator that the image is not being rendered, if possible providing..." needs to be fixed grammatically.  It should be "... rendered and, if possible, providing ..."

[14:49:00.0000] <Hixie>
ok all those are fixed now

[14:56:00.0000] <Lachy>
ok, that works even better the way you did it with the "otherwise" condition

[15:05:00.0000] <Lachy>
hmm, something is wrong with my IRC client. It just crashed several times in a row :-(

[15:06:00.0000] <Lachy>
... and it did it again. Whenever I right click on it :-(

[15:06:01.0000] <Hixie>
d'oh

[15:06:02.0000] <Philip`>
Maybe it's a Mac application and they never imagined someone might right-click

[15:07:00.0000] <Lachy>
maybe there's something wrong with the profile or a plist or something.

[15:07:01.0000] <Lachy>
LOL

[15:07:02.0000] <Lachy>
I'm supposed to be able to right click on someones name and select to open a private chat with them

[15:09:00.0000] <Hixie>
hsivonen's data shows that even in the svg files he downloaded, there are html <font> elements :-)

[15:10:00.0000] <Hixie>
though it looks like that can likely be accounted for by the cases where <font> wasn't in the svg namespace? unclear

[15:10:01.0000] <hsivonen>
Hixie: I looked for the local name

[15:11:00.0000] <hsivonen>
I should rerun the analyzer excluding foreignContent stuff

[15:11:01.0000] <hsivonen>
foreignObject that is

[15:12:00.0000] <Philip`>
hsivonen: Do you record a list of which files match each of the features you're searching for?

[15:12:01.0000] <Lachy>
I never thought the HTML5 spec would ever teach me a new Norwegian word :-)

[15:12:02.0000] <hsivonen>
Philip`: i don't

[15:12:03.0000] <Hixie>
i'm hoping the norwegian word is correct

[15:12:04.0000] <Lachy>
according to Google Translate it is

[15:12:05.0000] <Hixie>
my norwegian friends have not yet responded!

[15:13:00.0000] <Lachy>
I'll look it up in my Norwegian dictionary for you

[15:13:01.0000] <Hixie>
yeah, but i used google translate to determine it, so...

[15:13:02.0000] <Lachy>
Hixie, in all your time living in Norway, did you learn much Norwegian?

[15:13:03.0000] <Hixie>
Lachy: not really

[15:13:04.0000] <Hixie>
Lachy: but i didn't try very hard

[15:14:00.0000] <Lachy>
ok

[15:14:01.0000] <Hixie>
hsivonen: i really wasn't expecting the conclusions you drew from that data. that's really interesting.

[15:14:02.0000] <Hixie>
hsivonen: i think i'll make <font> be triggered from attributes

[15:14:03.0000] <Lachy>
nor have I. I know enough to get me through shopping

[15:14:04.0000] <Lachy>
and that's about it

[15:14:05.0000] <hsivonen>
Hixie: what conclusions don't flow from the data?

[15:15:00.0000] <Hixie>
hsivonen: all the conclusions flow from the data

[15:15:01.0000] <Hixie>
hsivonen: i jsut wasn't expecting that data

[15:15:02.0000] <hsivonen>
ah. I thought you weren't expecting the conclusions given the data

[15:15:03.0000] <Hixie>
Lachy: i never found lack of norwegian knowledge to be a practical barrier to living in norway.

[15:16:00.0000] <hsivonen>
Hixie: anything particularly unexpected?

[15:16:01.0000] <Hixie>
hsivonen: that the svgwg proposal would actually break on more pages than the commented out proposal

[15:16:02.0000] <Lachy>
From my dictionay: *bilde* /subst/. n 1. picture

[15:16:03.0000] <Hixie>
also, the sheer number of metadata elements

[15:17:00.0000] <Hixie>
Lachy: not photograph? bummer.

[15:17:01.0000] <Lachy>
the example is "I took a picture of my daughter"

[15:17:02.0000] <Lachy>
I'll look up photo

[15:17:03.0000] <hasather_>
Hixie: that would be "fotografi"

[15:18:00.0000] <Hixie>
isn't that photography, as opposed to photograph?

[15:18:01.0000] <Lachy>
Hixie: *Photo* is either: fotografi, foto, bilde

[15:19:00.0000] <hasather_>
Hixie: oh, sorry, read that wrong, then it's "foto"

[15:19:01.0000] <Hixie>
bummer, i was hoping for something that looked nothing like "photo" :-)

[15:20:00.0000] <Hixie>
i'd have used french but it uses photo too

[15:20:01.0000] <Lachy>
Hixie, fotograph is the same as photograph

[15:20:02.0000] <hsivonen>
Hixie: you could use 'valokuva' for lang=fi

[15:20:03.0000] <Hixie>
aha!

[15:20:04.0000] <Hixie>
henri to the rescue

[15:20:05.0000] <Lachy>
I mean fotografi

[15:21:00.0000] <Lachy>
silly norwegians using an "f" instead of "ph" for an "f" sound is confusing ;-)

[15:21:01.0000] <Hixie>
hsivonen: lang=fi is "Finnish" right? (checking spelling)

[15:21:02.0000] <hsivonen>
Hixie: yes

[15:22:00.0000] <Hixie>
committed

[15:22:01.0000] <Hixie>
thanks

[15:25:00.0000] <Lachy>
I wish there was some way for the spec to dynamically update itself whenver Hixie makes a commit, so I don't need to keep reloading the whole thing

[15:25:01.0000] <Hixie>
hmmm...

[15:26:00.0000] <Hixie>
that could be arranged

[15:26:01.0000] <Lachy>
cool

[15:29:00.0000] <Hixie>
Lachy: when i make a commit, or when i make an edit even before committing it?

[15:31:00.0000] <Lachy>
commits would be sufficient, and would probably work better since the revision number would provide a way for it to ask the server for all updates since the last revision it received.

[15:31:01.0000] <Lachy>
so if I go offline for a little while, then I still get all the updates during that time when I return

[15:31:02.0000] <roc>
sigh

[15:31:03.0000] <Hixie>
my plan is just to reload the page

[15:31:04.0000] <Hixie>
not do DOM-level diffs

[15:31:05.0000] <Lachy>
oh

[15:31:06.0000] <Hixie>
which would be insane amounts of work :-)

[15:31:07.0000] <Hixie>
roc: sigh?

[15:31:08.0000] <Philip`>
Rather than automatically reloading and making readers very confused if the text they just read has vanished for no apparent reason, could it just pop up a notification box with a button to reload?

[15:32:00.0000] <Hixie>
(patches welcome though if you want to do DOM-level diffs)

[15:32:01.0000] <Lachy>
well, I wouldn't want an auto refresh for that.

[15:32:02.0000] <Lachy>
since if I'm reading a section, I don't want it to suddently disappear

[15:33:00.0000] <Hixie>
a notification would be fine too i guess

[15:33:01.0000] <Hixie>
oh hey i know how to do diffs

[15:33:02.0000] <Philip`>
Subscribe to commit-watchers with a mail client that pops up a notification box when you get new mail

[15:33:03.0000] <Lachy>
I wonder how DOM level diffs could be implemented. Might have to parse the old revision and the latest revision with html5lib, then diff the DOMs and send back the changes somehow

[15:33:04.0000] <roc>
http://derstandard.at/?url=/?id=1216918402134 ... it's Miguel de Icaza

[15:34:00.0000] <Hixie>
i could also do it client-side by parsing the new doc into a DOM

[15:34:01.0000] <Hixie>
and doing something clever with the two DOMs

[15:34:02.0000] <Philip`>
By "clever" do you mean "fragile"?

[15:34:03.0000] <Lachy>
Philip`, I'm on commit watchers, but I don't like reading it cause that whole syntax is not really readable

[15:35:00.0000] <Hixie>
roc: so sad to see someone so intelligent fall onto a treadmill like that

[15:35:01.0000] <Philip`>
Lachy: The presence of the email still serves as a notification that you should reload the spec if you're currently reading it

[15:36:00.0000] <Lachy>
if it's done server side, then that would reduce your bandwidth load, but increase processor load. But if it's done client side, then it increases everyone's bandwidth load and client side processing

[15:38:00.0000] <Lachy>
but I guess, walking the DOMs, and comparing the innerHTML of the block level elements, and then updating and flagging the ones that change would work.

[15:39:00.0000] <Philip`>
If one paragraph is added, how would you tell that all the subsequent paragraphs are unchanged (just offset) from the previous version?

[15:40:00.0000] <Lachy>
xmldiff might work http://www.logilab.org/859

[15:40:01.0000] <Lachy>
it says it can work with DOMs

[15:51:00.0000] <Lachy>
xmldiff won't even install

[15:51:01.0000] <Lachy>
$ python setup.py install

[15:51:02.0000] <Lachy>
  File "setup.py", line 23

[15:51:03.0000] <Lachy>
    from __future__ import nested_scopes

[15:51:04.0000] <Lachy>
SyntaxError: from __future__ imports must occur at the beginning of the file

[15:58:00.0000] <Hixie>
woah, did sam just say he was ok with the spec?

[16:03:00.0000] <Lachy>
nice. I like it when people can become happy with the spec, without making any specific changes for them.

[16:06:00.0000] <Philip`>
He only seems to be okay if you ignore the line where he said "Of course, the usuability of such an approach sucks"

[16:11:00.0000] <Hixie>
well yes

[16:11:01.0000] <Hixie>
but as he says

[16:11:02.0000] <Hixie>
that's the point

[16:12:00.0000] <Hixie>
though to be honest i don't see why it sucks so much, it's basically syntactically equivalent and actually has more features than raw element names (you get to pick your own disambiguation scheme, you can pick multiple names, there is better fallback, etc)

[16:14:00.0000] <Lachy>
Hixie, for the marked issues in the spec, we used to have a JavaScript that inserted "Big Issue: " at the beginning of each one. You could just enable that again

[16:15:00.0000] <Lachy>
or maybe change it to insert a ** or some other syntax

[16:18:00.0000] <Lachy>
Hixie, http://status.whatwg.org/mark-issues.js

[16:18:01.0000] <Lachy>
it was called as part of this script http://status.whatwg.org/annotate-web-apps.js

[16:18:02.0000] <Hixie>
ah

[16:18:03.0000] <Philip`>
Isn't it easier to write "Big Issue" in the source, rather than having JS that slowly traverses and dynamically updates the whole document at runtime on each client?

[16:19:00.0000] <Lachy>
Philip`, it could probably be done by the spec gen or something

[16:19:01.0000] <Hixie>
i don't really want the issues to say "big issue" though

[16:20:00.0000] <Hixie>
at least not to sighted users

[16:23:00.0000] <Lachy>
is there some almost-invisible character that would be noticed by screen readers that could convey the intended meaning?

[16:23:01.0000] <Philip`>
<font color=white>Big Issue:</font>

[16:24:00.0000] <Lachy>
Philip`, stylesheets could hide it

[16:24:01.0000] <Hixie>
wouldn't they hide it from screen readers too?

[16:25:00.0000] <Philip`>
(and then Google will ban you for making content visible to search engines and not to normal users)

[16:25:01.0000] <Lachy>
Hixie, not if you use position: absolute; left: -2000px;

[16:25:02.0000] <Hixie>
that's such a hack

[16:25:03.0000] <Hixie>
what's the officially right way to do this?

[16:25:04.0000] <Lachy>
for these in the stylesheet: p.note:before { content: 'Note: '; } p.warning:before { content: '\26A0 Warning! '; }, that probably could be inserted directly into the markup instead of using stylesheets too

[16:25:05.0000] <Hixie>
surely the accessibility experts have come up with a solution

[16:26:00.0000] <Lachy>
Hixie, unfortunately, the correct way to do it would have been display: none;, but screen readers decided to ignore such content

[16:26:01.0000] <Hixie>
i'll wait for gsnedders to come back and have him add stuff for those classes

[16:26:02.0000] <Hixie>
Lachy: yeah, i know, i was being sarcastic. :-)

[16:26:03.0000] <Hixie>
Lachy: the right solution is for screen readers to be replaced by actual speech browsers

[16:27:00.0000] <Hixie>
i should go eat lunch

[16:27:01.0000] <Hixie>
bbiab

[16:27:02.0000] <Lachy>
Hixie, I agree. But others who work in the accessibility field disagree, cause apparently users use the same screen reader for all their applications, not just web browsing

[16:28:00.0000] <Philip`>
It would be great if the market was large enough to support development of a real browser comparable to IE or Firefox but designed for a speech interface, but it doesn't seem to be

[16:28:01.0000] <Philip`>
Oh, actually that wouldn't be great - it would be best if the market disappeared entirely

[16:29:00.0000] <Lachy>
if there was such a browser that ordinary screen readers could hook into, that might work too, instead of trying to hack an accessibility API onto a browser predominately designed for visual users

[16:29:01.0000] <roc>
since 90% of the code is the same for a visual browser and a speech browser, it would make sense to share the code

[16:29:02.0000] <Lachy>
you mean if we exterminated all potential users of screen readers? Sure, that would work. It's a bit cruel though!

[16:29:03.0000] <jcranmer>
if there was a browser that wacked web developers senseless until they actually designed for non-visual users, that would help

[16:30:00.0000] <roc>
and then you have to ask yourself if it makes sense to have one build for visual and one for speech

[16:30:01.0000] <jcranmer>
/me would have to ask MarcoZ what the internet is like with a screen reader

[16:30:02.0000] <Philip`>
Lachy: I was thinking of a more ideal world where people don't have such disabilities, not one where we kill everyone who doesn't conform

[16:31:00.0000] <Philip`>
s/don't have/never acquire/

[16:32:00.0000] <Philip`>
The problem is when there a lot of people who are affected by poor tools, but not enough to justify the development of good tools

[16:32:01.0000] <Philip`>
*when there are ...

[16:32:02.0000] <Lachy>
roc, separate browsers could share the same underlying code, but with UI built from the groud up to support different methods of usage

[16:33:00.0000] <roc>
yeah, ok

[16:33:01.0000] <jcranmer>
the Gecko screen reader browser!

[16:33:02.0000] <Lachy>
in my limited use of screen readers, I found that one of my biggest hassels was that the screen reader overrides many of the normal application keyboard shortcuts

[16:34:00.0000] <roc>
given that things like http://vimperator.mozdev.org/ exist, you probably build what's needed as a Firefox extension

[16:34:01.0000] <Lachy>
if the application itself was designed and built with a screen reader UI in mind, then such conflicts could be avoided

[16:34:02.0000] <jcranmer>
an XULRunner application might be better?


2008-08-05
[18:32:00.0000] <aboodman>
question on MessagePorts: why do they have an ownerWindow?

[18:32:01.0000] <aboodman>
it seems weird that you can reach ownerWindow.openDatabase, for example

[18:37:00.0000] <Hixie>
aboodman: ownerWindow is always the Window in which they are found, so you can always get to the Window anyway

[18:38:00.0000] <Hixie>
i suppose it doesn't have to be made visible to the interface

[18:38:01.0000] <aboodman>
but if I pass a MessagePort through to some other context, what will that context see for port.ownerWindow ?

[18:38:02.0000] <aboodman>
what if it is cross-origin?

[18:38:03.0000] <Hixie>
the ownerWindow will change to the new context

[18:39:00.0000] <aboodman>
oh, i didn't get that part.

[18:39:01.0000] <Hixie>
it's the owner of the port at that time, not the original owner or the creator

[18:39:02.0000] <Hixie>
yeah

[18:39:03.0000] <aboodman>
what is the point of this property?

[18:39:04.0000] <Hixie>
i need to write an intro section

[18:39:05.0000] <Hixie>
i'm not sure why i made it a property

[18:39:06.0000] <Hixie>
the concept is needed for various aspects of how the ports work

[18:39:07.0000] <Hixie>
i don't think we need to actually expose it

[18:39:08.0000] <aboodman>
right

[18:39:09.0000] <aboodman>
ok, thanks

[18:40:00.0000] <Hixie>
if you want it removed, send mail and i'll look at why i put it there in the first place

[18:40:01.0000] <aboodman>
ok

[18:40:02.0000] <Hixie>
right now i'm swamped with dealing with all kinds of whining people

[18:40:03.0000] <aboodman>
ok, can you talk about another aspect of this spec now?

[18:40:04.0000] <Hixie>
sure

[18:40:05.0000] <Hixie>
i can talk abotu anything :-)

[18:40:06.0000] <aboodman>
(i will send mail about it too, but i just love this low latency communication :))

[18:41:00.0000] <aboodman>
ok, web workers, processing model, step 10

[18:41:01.0000] <aboodman>
it says that scripts running in the worker will get suspended when the worker is no longer an active needed worker

[18:42:00.0000] <Hixie>
right

[18:43:00.0000] <aboodman>
so this means that if you say: createWorker("foo.js"), but don't hold a reference to the returned port, foo.js may get paused at any random point

[18:43:01.0000] <aboodman>
(whenever gc happens)

[18:44:00.0000] <aboodman>
let me back up... what is that step for?

[18:44:01.0000] <Hixie>
if you don't hold on to the reference, then step 9 will soon set closing to true

[18:45:00.0000] <Hixie>
and step 10 won't trigger, since it only triggers when closing = false

[18:45:01.0000] <Hixie>
and so the worker will gracefully shutdown

[18:45:02.0000] <Hixie>
the step is intened to pause workers when someone hits the back button and the browser doesn't throw the document away

[18:45:03.0000] <Hixie>
e.g. firefox's bfcache

[18:46:00.0000] <aboodman>
I see.

[18:46:01.0000] <aboodman>
the ports can become inactive separately from being gc'd, i forgota bout that

[18:46:02.0000] <Hixie>
yeah

[18:47:00.0000] <Hixie>
if they try to gc, step 10 doesn't kick in, because step 9 does and step 10 only kicks in when 9 doesn't

[19:07:00.0000] <jcranmer>
Hixie: you should update the implementation status for FF and audio

[19:08:00.0000] <Hixie>
you can do so too :-)

[19:08:01.0000] <Hixie>
what should it say?

[19:08:02.0000] <jcranmer>
experimental support

[19:08:03.0000] <Hixie>
(alt-double-click on a section to edit the status)

[19:09:00.0000] <Hixie>
ok, will change

[19:09:01.0000] <jcranmer>
alt-double click where?

[19:09:02.0000] <Hixie>
anywhere in a section

[19:10:00.0000] <jcranmer>
oh, I have to login first, I presume?

[19:11:00.0000] <Hixie>
yeah

[19:11:01.0000] <Hixie>
if you've ever sent feedback, you'll have a login name and password

[19:11:02.0000] <Hixie>
login name is the e-mail you used to send feedback

[19:11:03.0000] <Hixie>
password will be sent to that address

[19:11:04.0000] <jcranmer>
and it would help if my WM didn't use alt to handle window stuff...

[19:12:00.0000] <Hixie>
heh

[19:13:00.0000] <jcranmer>
/me is surprised he beat roc to an example of <audio> in a blog

[19:16:00.0000] <Hixie>
feel free to add urls to examples and demos in the spec too

[19:20:00.0000] <takkaria>
I was going to go and update bits of the annotations but I had the same alt- being used for my window manager -problem

[19:23:00.0000] <Hixie>
heh

[19:24:00.0000] <jcranmer>
/me gets lazy and solves the problem the easy way

[19:24:01.0000] <Hixie>
you can press ctrl+alt at the same time i think

[19:24:02.0000] <Hixie>
if that helps

[19:25:00.0000] <jcranmer>
ah, that works

[19:25:01.0000] <jcranmer>
no need to do my workaround: edit host files to point to localhost and then do apache rewriting to change the JS file to use something else

[19:29:00.0000] <Hixie>
heh

[19:30:00.0000] <Hixie>
that was your easy way? :-)

[19:30:01.0000] <jcranmer>
easiest way I could think of to edit the source of another page

[19:30:02.0000] <Hixie>
heh

[19:31:00.0000] <Hixie>
i wonder how a script on whatwg.org can work out what revision it is

[19:31:01.0000] <jcranmer>
I have a few rewrites that redirect unstable.elemental.com to my localhost

[19:36:00.0000] <jcranmer>
what makes a boolean attribute false?

[19:36:01.0000] <jcranmer>
it says it's presence makes it true

[19:36:02.0000] <jcranmer>
but if it's present, it's value is either empty of the name of the attribute itself

[19:36:03.0000] <jcranmer>
so what would attr="false" return?

[19:36:04.0000] <jcranmer>
s/return/reflect/

[19:37:00.0000] <takkaria>
I believe the thing that makes a boolean attribute false is if it is not set

[19:39:00.0000] <jcranmer>
I just want to have confirmation of this before I mark a bug as INVA

[19:40:00.0000] <aboodman>
Hixie: It seems like right now workers shut down whenever there is no non-gc'd port that refers to them. Even if they have script in progress or things like XHR and timers pending. Is this right?

[19:41:00.0000] <Hixie>
aboodman: they set closing to true and run through the closing steps first, so it's gracefull-ish, but yes, fundamentally they're just killed when their owners go away

[19:41:01.0000] <Hixie>
jcranmer: its absence

[19:42:00.0000] <aboodman>
right but it is non-deterministic

[19:42:01.0000] <aboodman>
if the UA doesn't GC then the worker might not shut down

[19:42:02.0000] <Hixie>
yeah, i wasn't sure what to do abotu that

[19:43:00.0000] <Hixie>
i mean, it's non-deterministic anyway, since it's on another thread

[19:43:01.0000] <aboodman>
I think that workers should stay alive as long as:

[19:43:02.0000] <aboodman>
they have script running, they have messages queued for them, they have http requests outstanding, or they have timers ounstanding

[19:44:00.0000] <aboodman>
that would eliminate a lot of hard to understand bugs, when workers sometimes go away and sometimes don't with the same code

[19:44:01.0000] <Hixie>
well it has to be more than that

[19:45:00.0000] <Hixie>
otherwise any random page can just start a worker with a timer and it'll never go away

[19:45:01.0000] <aboodman>
navigation would of course always either nuke or suspend workers

[19:45:02.0000] <aboodman>
whatever is appropriate for the UA

[19:46:00.0000] <Hixie>
how is that not non-deterministic?

[19:46:01.0000] <Hixie>
when you navigate, the worker now has an indeterminate amount of processing time left

[19:46:02.0000] <aboodman>
if i'm a developer and i write some code that uses workers, but i make a mistake and don't hold the port that comes back

[19:46:03.0000] <aboodman>
the code inside that worker will stop running at some random point

[19:46:04.0000] <aboodman>
even without navigating the page

[19:46:05.0000] <othermaciej_>
workers should stay alive at least so long as someone can message them, even if no messages are queued

[19:47:00.0000] <othermaciej>
otherwise, any global data set inside the worker could be lost on a race condition of message delivery

[19:47:01.0000] <aboodman>
othermaciej_: that is in the spec now i think

[19:47:02.0000] <othermaciej>
yeah, I'm just disagreeing with your "as long as" above, at least as an exhaustive list

[19:47:03.0000] <Hixie>
aboodman: it doesn't stop running at a random point

[19:47:04.0000] <aboodman>
right, sorry, I forgot about "someone can message them"

[19:47:05.0000] <Hixie>
aboodman: it runs to completion

[19:47:06.0000] <othermaciej>
and I am not sure "someone can message them" can be determined short of GC

[19:48:00.0000] <othermaciej>
unless there is an explicit way to close a message port I suppose

[19:48:01.0000] <Hixie>
there is, but i doubt many people will use it, and i certainly don't think we should rely on them doing so

[19:49:00.0000] <aboodman>
does "run to completion" means that the current event that is being processed finishes, or does it mean that all queued events finish being processed

[19:49:01.0000] <othermaciej>
yeah, you can't realy on it

[19:50:00.0000] <Hixie>
aboodman: all queued events get processed (but no new events can be queued once closing is true)

[19:50:01.0000] <aboodman>
othermaciej: I agree that you can't determine "someone can message them" short of a GC, but I"m just saying that that alone should not determine the worker's lifetime.

[19:51:00.0000] <aboodman>
Hixie: ok, but httprequests still may or may not happen, right?

[19:51:01.0000] <aboodman>
same with timers I suppose

[19:51:02.0000] <othermaciej>
aboodman: I think a message with queued messages or pending timers or pending I/O should continue even if no one can (currently) message it

[19:51:03.0000] <othermaciej>
aboodman: I don't see how you would do it without either counting on GC, or explicit close

[19:51:04.0000] <Hixie>
if an xml http request is outstanding, or if there is a repeating interval or an outstanding timer waiting, those are discarded, yes

[19:51:05.0000] <Hixie>
because otherwise you could use that to stay alive forever

[19:52:00.0000] <othermaciej>
I am assuming here that the message port for the thread is a GC object and anyone who can reach that object can send a message

[19:52:01.0000] <Hixie>
othermaciej: yes

[19:52:02.0000] <othermaciej>
Hixie: I know that is true in the current spec, what I mean is I am assuming it even for hypothetical designs that tried to make shutdown time not depend on GC

[19:53:00.0000] <aboodman>
othermaciej: I agree with you, but I don't think the spec reflects this

[19:54:00.0000] <Hixie>
othermaciej: ah ok

[19:54:01.0000] <othermaciej>
"fire and forget" threads should be allowed IMO, if they are doing I/O or something

[19:55:00.0000] <aboodman>
Hixie: are you worried specifically about named workers staying alive forever? Because I don't see the problem with anonymous ones doing that. They will go away at page unload.

[19:55:01.0000] <Hixie>
right now the mechanism for them going away at page unload is the same mechanism as them going away any other time

[19:55:02.0000] <Hixie>
i suppose we could have a new mechanism

[19:55:03.0000] <Hixie>
that is different for that case

[19:56:00.0000] <aboodman>
it doesn't have to be ... unload could just be defined to nuke all (anonymous) workers

[19:56:01.0000] <Hixie>
that won't work

[19:56:02.0000] <Hixie>
workers can be created by one window and passed off to another

[19:57:00.0000] <aboodman>
Right. I didn't realize that you intended for those to stay alive.

[19:57:01.0000] <aboodman>
past their owning window closing.

[19:57:02.0000] <Hixie>
they are still "active" and "needed", in that you can still send messages to them

[19:57:03.0000] <Hixie>
i think it'd be confusing if they had a lifetime bounded by whatever random iframe or worker created them

[19:58:00.0000] <Hixie>
you want to be able to have a worker spawn off new workers to service requests for gadgets or whatever

[19:58:01.0000] <Hixie>
and have those live independent of hte original factory worker

[19:58:02.0000] <othermaciej>
I think maybe there should be different mechanisms for termination on navigation vs otherwise

[19:58:03.0000] <othermaciej>
though unsure of the detauls

[19:58:04.0000] <othermaciej>
*details

[19:58:05.0000] <aboodman>
I see the problem. I was thinking of anonymous workers as more firmly attached to a page.

[19:58:06.0000] <othermaciej>
for the scope of a particular page I think it should be possible to make a thread that will periodically message you but that you don't care to message ever

[19:58:07.0000] <othermaciej>
for instance

[19:59:00.0000] <othermaciej>
but such a thread should not (under ordinary circumstances) survive a page navigation

[19:59:01.0000] <Hixie>
othermaciej: hm, right now to do that you'd have to keep a reference in the page

[19:59:02.0000] <Hixie>
iirc

[19:59:03.0000] <aboodman>
othermaciej: if it can message you, you have a port for it, so it will stay alive.

[19:59:04.0000] <Hixie>
not if you set up a handler for the port and then drop it all on the floor

[20:00:00.0000] <othermaciej>
oh I see, the ports are all bidirectional

[20:00:01.0000] <Hixie>
oh wait

[20:00:02.0000] <Hixie>
no

[20:00:03.0000] <Hixie>
othermaciej: yes

[20:00:04.0000] <Hixie>
the other port itself will prevent the port from being GC'ed

[20:00:05.0000] <Hixie>
so what you describe will work

[20:00:06.0000] <aboodman>
right.

[20:00:07.0000] <othermaciej>
I forgot about that (though it leads to something else in the spec I don't like, the fact that ports are killed on transfer)

[20:01:00.0000] <othermaciej>
ports that are endpoints of the same pipe protect each other?

[20:01:01.0000] <Hixie>
yeah

[20:01:02.0000] <Hixie>
(s/pipe/channel/)

[20:01:03.0000] <othermaciej>
what about a completely fire-and-forget thread, where you ask it to do some I/O but no further messaging is expected either way?

[20:01:04.0000] <aboodman>
but you couldn't, for example, set up a worker that would synchronize a database with the network but never message the creator.

[20:01:05.0000] <aboodman>
it would seem to work for a little while, but it would go away after awhile.

[20:02:00.0000] <Hixie>
othermaciej: right now that thread will die before it gets its onreadystatechange, assuming that it uses async IO

[20:02:01.0000] <Hixie>
aboodman: yeah...

[20:02:02.0000] <Hixie>
not sure how to handle these two cases

[20:03:00.0000] <othermaciej>
it seems like a potentially useful use case

[20:03:01.0000] <Hixie>
i agree it'd be good to handle them

[20:03:02.0000] <aboodman>
i'm less concerned about the use case than it failing in weird ways

[20:03:03.0000] <othermaciej>
you could let self-effected liveness criteria not count at page navigation time or something

[20:04:00.0000] <Hixie>
actually...

[20:04:01.0000] <Hixie>
/me looks at the spec

[20:05:00.0000] <Hixie>
so...

[20:05:01.0000] <Hixie>
what we could do is add an implicit link from any worker to whoever created it

[20:06:00.0000] <Hixie>
so that it won't ever get killed if whoever created it is still there, so long as it has any outstanding timers, intervals, database callbacks, or xmlhttp request callbacks pending

[20:09:00.0000] <aboodman>
so the creator doesn't reference it in the same way that a port does... it is only used in the case where there are no live ports

[20:10:00.0000] <aboodman>
if there are no live ports, but there is outstanding IO or timers, and the creator is still there, then it keeps running

[20:11:00.0000] <Hixie>
right

[20:11:01.0000] <Hixie>
though that will make debugging a bitch

[20:11:02.0000] <Hixie>
"why is that worker still running? how do i disable it?

[20:11:03.0000] <Hixie>
"

[20:11:04.0000] <Hixie>
just because it happens to have setInterval(function () {}, 100);

[20:12:00.0000] <aboodman>
hm, yeah. weird. i'll think about it.

[20:13:00.0000] <othermaciej>
a number of DOM-related objects have "live while some message is pending, even if no one kept a reference" semantics

[20:13:01.0000] <othermaciej>
for instance this is true of Image and XMLHttpRequest

[20:13:02.0000] <othermaciej>
workers of course have a much richer way of making themselves have a pending action

[20:13:03.0000] <aboodman>
yeah the only thing that is weird with workers is that their lifespans can be longer than a page

[20:14:00.0000] <othermaciej>
that is true of Image and XMLHttpRequest too

[20:14:01.0000] <Hixie>
yeah

[20:15:00.0000] <aboodman>
how is it true of image and xhr?

[20:15:01.0000] <Hixie>
aboodman: feel free to send mail if you come up with some other idea (or even if you don't); i agree that we should do something about this either way

[20:15:02.0000] <othermaciej>
you can hand one to a different page

[20:15:03.0000] <aboodman>
wow, I didn't realize that.

[20:15:04.0000] <othermaciej>
(don't remember what happens if you then navigate the originating page)

[20:15:05.0000] <othermaciej>
you can pass references to any object in the DOM cross-frame (which amounts to cross-page)

[20:15:06.0000] <aboodman>
i mean it makes sense you can pass the object to another page, it's just an object

[20:16:00.0000] <aboodman>
i just never considered what should happen when you do this with an xhr

[20:16:01.0000] <Hixie>
heh

[20:16:02.0000] <aboodman>
it would be interesting to know what xhr does when you navigate the originating page

[20:18:00.0000] <othermaciej>
it may depend on which frame started the load rather than which constructed the object

[20:18:01.0000] <othermaciej>
not really sure though

[20:18:02.0000] <aboodman>
i need to run, i'll start a thread with this feedback on the list

[20:18:03.0000] <aboodman>
thanks both for your time

[20:18:04.0000] <Hixie>
cool

[20:18:05.0000] <Hixie>
np

[20:18:06.0000] <Hixie>
thanks for the feedback!

[20:44:00.0000] <Hixie>
Lachy: ok, there's a checkbox in the status section now which, if checked, will let you know within 30 seconds of me committing some new revision

[23:19:00.0000] <Hixie>
aboodman2, othermaciej, it would be useful to have your input on the issues sicking has raised recently, like pushState() and the workers stuff

[23:20:00.0000] <aboodman2>
you mean the message he just sent, or was there something earlier?

[23:20:01.0000] <aboodman2>
also, I'm trying to reply, but whatwg.org is not responding :-/.

[23:20:02.0000] <Hixie>
that one, and the thread on pushState() recently (which i just replied to -- i'll give you a link in a sec)

[23:20:03.0000] <Hixie>
uh yes the server seems non-responsive wtf

[23:21:00.0000] <Hixie>
the machine is working fine except apache is on strike

[23:21:01.0000] <aboodman2>
I didn't understand his comment about pulling in all of navigator and window. I was going to check the proposal again, because I don't remember that being true.

[23:22:00.0000] <Hixie>
i think he was speaking hypothetically

[23:22:01.0000] <Hixie>
why is apache not working

[23:23:00.0000] <othermaciej>
Hixie: I saw the pushState thread

[23:23:01.0000] <othermaciej>
Hixie: in general it kind of makes sense to require auxiliary data to be stuffed into the URL somehow, since that preserves bookmarkability and shareability of the URL as well as back/forward

[23:24:00.0000] <othermaciej>
Hixie: but perhaps there are use cases where back/forward is very useful but recording a particular configuration persistently is not, in which case some form of data would make sense

[23:24:01.0000] <othermaciej>
that's my off-the-cuff opinion

[23:24:02.0000] <Hixie>
i'm ok with dropping the data part and only using URL

[23:25:00.0000] <Hixie>
ok seriously wtf is with my server

[23:25:01.0000] <Hixie>
/me reboots it, because, well, what else can he do

[23:26:00.0000] <aboodman2>
Hixie

[23:26:01.0000] <aboodman2>
whoa, sorry, new IRC client, unfamiliar controls

[23:26:02.0000] <Hixie>
aboodman2

[23:26:03.0000] <Hixie>
:-)

[23:27:00.0000] <aboodman2>
Hixie: I havne't looked at the state management spec at all, so I don't think I have much to say about that thread right now.

[23:27:01.0000] <Hixie>
aboodman2: server's back up

[23:27:02.0000] <Hixie>
ok

[23:27:03.0000] <aboodman2>
I'll try and catch up on it, but right now my higher concerns are workers, database, and offline

[23:28:00.0000] <Hixie>
sure

[23:29:00.0000] <Hixie>
i'm just curious to get the opinion of other implementors, lest one implementor end up with undue influence

[23:29:01.0000] <Hixie>
nothing against jonas of course, i have the utmost respect for him

[23:42:00.0000] <aboodman2>
Hixie: what is the name of the interface that Window and WorkerWindow share? I cannot find it.

[23:43:00.0000] <aboodman2>
WindowWorker, sorry

[23:46:00.0000] <aboodman2>
nevermind, found it.

[23:48:00.0000] <Hixie>
sorry was afk

[00:04:00.0000] <mcarter>
I don't suppose anyone has any idea what is supposed to happen (if there is a standard) if an HTTP/1.1 XHR response is half over when there is a page navigation. Should it a) shut the connection immediately, or b) wait for the rest of the response (though now irrelevant), and then re-use the connection?

[01:55:00.0000] <Hixie>
r2000!

[02:20:00.0000] <Hixie>
i'm starting to think that julian wants me to _require_ that people use URIs, not just allow it

[02:20:01.0000] <Hixie>
(for class names)

[02:20:02.0000] <hsivonen>
Hixie: yes

[02:21:00.0000] <Philip`>
(For class names that are intended to have a shared meaning across multiple sites and tools, I presume)

[02:21:01.0000] <hsivonen>
the migration of the cc prefix from http://web.resource.org/cc/ to http://creativecommons.org/ns# shows that vocabularies outlive domain bikeshedding

[02:22:00.0000] <Philip`>
Does anyone use IP addresses in namespace identifiers?

[02:22:01.0000] <hsivonen>
Philip`: not in the wikimedia dataset

[02:23:00.0000] <hsivonen>
also, the most common metadata item is utterly pointless: http://purl.org/dc/elements/1.1/ dc format

[02:24:00.0000] <Philip`>
http://www.google.com/search?q=%22xmlns+http+1..255%22 - oh, they do

[02:24:01.0000] <hsivonen>
if you are looking for metadata inside an svg file, surely you should be able to infer that the format is SVG

[02:24:02.0000] <Philip`>
(By the way, why does Google bolden the first two numbers in the IP addresses in the search results?)

[02:24:03.0000] <hsivonen>
wow

[02:26:00.0000] <Philip`>
Class names should just be GUIDs

[02:31:00.0000] <Hixie>
if he wants everyone to use uris, then that seems a bit rude...

[02:32:00.0000] <hsivonen>
Hixie: if only some people use uris and others use anything, anything may accidentally collide with an URI

[02:33:00.0000] <Hixie>
yeah, that's likely </sarcasm>

[02:34:00.0000] <hsivonen>
easy fix: if colon, must be uri

[02:38:00.0000] <Hixie>
seriously, if he's actually worried that people will invent class names that coincidentally happen to look exactly like URIs in his domain, then he has his priorities wrong

[02:39:00.0000] <othermaciej>
look, the problem is clear

[02:40:00.0000] <othermaciej>
we can't let extensions to the Web be limited by a central registry

[02:40:01.0000] <hsivonen>
othermaciej: including TLDs?

[02:40:02.0000] <othermaciej>
that is why we must use a mechanism with *two* central registries, one for URI schemes, and one for domain names

[02:41:00.0000] <othermaciej>
because then it is decentralized

[02:56:00.0000] <Hixie>
the recent thread on the big-issue markup is surprisingly good at distinguishing people who understand the web and accessibility and those who don't

[03:04:00.0000] <gDashiva>
hsivonen: About the cc: prefix, are you saying the namespace was changed, but everything just uses cc: and didn't notice, or?

[03:22:00.0000] <hsivonen>
gDashiva: the prefix was retained. the local part of the vocabulary was retained. the URI changed, breaking processors that process namespaces or RDF correctly

[03:23:00.0000] <Philip`>
Just set up an HTTP redirect at the old URI, and it should all be fine

[03:23:01.0000] <hsivonen>
hah

[03:24:00.0000] <Philip`>
Hixie: "If it is no a number"

[03:24:01.0000] <Hixie>
fixed

[03:24:02.0000] <Hixie>
hsivonen: is v.nu expected to be down?

[03:25:00.0000] <hsivonen>
Hixie: no

[03:25:01.0000] <Hixie>
it's been returning 503 for a few hours now

[03:25:02.0000] <Hixie>
html5.validator.nu specifically

[03:25:03.0000] <Hixie>
(trying to validate the spec)

[03:26:00.0000] <hsivonen>
I'll move to a place where I can sit down to fix.

[03:26:01.0000] <Hixie>
btw for those of you who weren't around earlier, there is now a magic checkbox in the SotD section of the whatwg draft

[03:26:02.0000] <Hixie>
and we crossed the 2000 revision mark

[03:26:03.0000] <gDashiva>
magic checkbox, wow

[03:28:00.0000] <Philip`>
Sadly revision 2000 was just shuffling some text around and seemingly didn't introduce any new errors, so I can't complain about an R2K bug

[03:28:01.0000] <Hixie>
heh

[03:29:00.0000] <gDashiva>
Maybe he modified the commit-watchers mail

[03:29:01.0000] <hsivonen>
Hixie: it's back up now

[03:29:02.0000] <hsivonen>
I really need to get that SMS thing in place

[03:30:00.0000] <hsivonen>
Hixie: thanks for the heads up

[03:30:01.0000] <Hixie>
np, sorry for not telling you earlier

[03:31:00.0000] <Hixie>
i can have my system e-mail you if it notices a problem if you like

[03:32:00.0000] <hsivonen>
Hixie: email doesn't reach me when there's a situation that I'm not otherwise babysittng the server :-(

[03:32:01.0000] <Hixie>
k

[03:38:00.0000] <Lachy>
that distributed extensibility thread is getting really annoying. It seems to be taken for granted that a) clashes are inevitable and b) it causes big problems when they occur, without actually explaining what the problem is when 2 distinct sites use common class names for different purposes, intended for processing by their own individual tools

[03:38:01.0000] <Hixie>
yes it's especially annoying given that both a and b are flase

[03:38:02.0000] <Hixie>
false

[03:39:00.0000] <Lachy>
if their extensions are intended for use outside their own domain, then such extensions need community involvment. Otherwise, it really makes no difference

[03:39:01.0000] <Hixie>
what should happen to transparency when someone uses toDateURL() with a non-transparency-supporting image type?

[03:39:02.0000] <Hixie>
Lachy: feel free to take over from me here, i'm quite ready to move on to more productive things

[03:40:00.0000] <Lachy>
nah, I've avoided contributing to the whole thread on the list in the hope that it would die out on its own

[03:40:01.0000] <gDashiva>
Clearly hash functions would never work on the internet

[03:41:00.0000] <Lachy>
it should either be converted to the <canvas> element's background colour (if any), or some default colour like white

[03:41:01.0000] <Philip`>
Canvases rarely have a background-color so it seems weird and unexpected to depend on that

[03:41:02.0000] <Philip`>
s/a/a non-transparent/

[03:42:00.0000] <Hixie>
not to mention off-DOM canvasaes

[03:42:01.0000] <Hixie>
or canvii

[03:42:02.0000] <Lachy>
then in those cases, they would use the default colour

[03:43:00.0000] <Hixie>
i'm saying it should just drop the alpha channel

[03:43:01.0000] <Hixie>
which basically means using black

[03:43:02.0000] <Philip`>
It seems simplest to always effectively composite onto a solid black or white background

[03:43:03.0000] <Philip`>
and if people want some other background then they can draw it manually

[03:43:04.0000] <Philip`>
"drop the alpha channel" may confuse people who are using premultiplied alpha representations

[03:44:00.0000] <Hixie>
+  <p>For image types that do not support all the channels (in particular,

[03:44:01.0000] <Hixie>
+   that do not support an alpha channel) the unsupported channels must be

[03:44:02.0000] <Hixie>
+   ignored when creating the image.

[03:44:03.0000] <Hixie>
what should i say instead?

[03:44:04.0000] <Philip`>
Why try to cover the non-existent case where an image format is missing support for the R/G/B channels?

[03:45:00.0000] <Philip`>
Well, not really non-existent, since some greyscale formats exist, but then you'd want to do something cleverer rather than just dropping all the colour channels

[03:45:01.0000] <Lachy>
so if someone set the colour to rgba(255, 0, 0, 0), then dropping the alpha channel would make it go red?

[03:45:02.0000] <Hixie>
if someone set the colour to rgba(255, 0, 0, 0), it would probably be stored as 0,0,0,0

[03:45:03.0000] <Lachy>
ok

[03:46:00.0000] <Philip`>
Oh, that sounds confusing

[03:46:01.0000] <Lachy>
what if it was rgba(255, 0, 0, 0.1)?

[03:46:02.0000] <Hixie>
Philip` is the expert here

[03:46:03.0000] <Hixie>
i'm hoping he'll tell me what it should say :-)

[03:47:00.0000] <Hixie>
should i just explicitly say that you multiple the alpha channel with the rgb channels and use taht?

[03:47:01.0000] <Hixie>
multiply, that

[03:47:02.0000] <Philip`>
That sounds reasonable, though it might be easier to phrase it in terms of compositing the image onto a solid black background with source-in

[03:48:00.0000] <Philip`>
Uh

[03:48:01.0000] <Philip`>
source-over

[03:48:02.0000] <Philip`>
or it might not be, I don't know :-p

[03:49:00.0000] <Lachy>
Philip`, wouldn't compositing the image onto a solid white background give better results?

[03:49:01.0000] <Philip`>
(At least then rgba(255,0,0,0) will always come out as rgb(0,0,0), and it won't depend on the internal representation)

[03:49:02.0000] <Philip`>
Lachy: What does "better" mean?

[03:50:00.0000] <Lachy>
well, the resulting colours won't be dark

[03:50:01.0000] <Hixie>
Philip`: is this ok?:

[03:50:02.0000] <Hixie>
  <p>For image types that do not support an alpha channel, the image

[03:50:03.0000] <Hixie>
  must first be composited onto a solid black background using the

[03:50:04.0000] <Hixie>
  source-over operator, and then exported.</p>

[03:50:05.0000] <Lachy>
what do image editors do in this situation?

[03:52:00.0000] <Hixie>
actually:

[03:52:01.0000] <Hixie>
+  <p>For image types that do not support an alpha channel, the image

[03:52:02.0000] <Hixie>
+  must be composited onto a solid black background using the

[03:52:03.0000] <Hixie>
+  source-over operator, and the resulting image must be the one used

[03:52:04.0000] <Hixie>
+  to create the <code title="">data:</code> URL.</p>

[03:53:00.0000] <Philip`>
Lachy: If you draw solid shapes onto the canvas, then it won't matter what the background colour is; and if you draw transparent objects, then try to save it to a format that doesn't support transparency, it's never going to work properly and it'll give unexpected results regardless of what we do

[03:54:00.0000] <Lachy>
ok, but it just seems more intuitive to me to use a white background

[03:54:01.0000] <Philip`>
so I guess what matters is just what's unexpected in the fewest cases

[03:55:00.0000] <Philip`>
Lachy: The image editors that I've used seem to use either white or the colour in the background position in your colour palette thing

[03:55:01.0000] <Lachy>
yeah, that's what I thought

[03:55:02.0000] <Philip`>
I wouldn't object to using white instead of black

[03:56:00.0000] <Philip`>
Hixie: It should perhaps be clearer that the compositing is creating a new temporary image, and not affecting the original at all

[03:56:01.0000] <krijnh>
"So it seems to be pointless to continue this discussion." - "Could you elaborate?" - huh?

[03:57:00.0000] <Hixie>
Philip`: that's what the second block above (which i committed) is trying to be more explicit about

[03:58:00.0000] <gDashiva>
Is Julian on IRC?

[03:59:00.0000] <krijnh>
In #html-wg, yes

[03:59:01.0000] <Philip`>
(Oops, got to go for an hour)

[04:01:00.0000] <gDashiva>
/me wonders if the whole thread could be killed off by a small IRC chat

[04:02:00.0000] <krijnh>
I have that idea with almost every thread on public-html :)

[04:04:00.0000] <Lachy>
it does indeed seem like our discussions in here are more productive than public-html sometimes.

[04:04:01.0000] <krijnh>
But in here there's no concept of concensus ;)

[04:04:02.0000] <krijnh>
Ow, wait :)

[04:18:00.0000] <gDashiva>
krijnh: I disagree, there's consensus all the time

[04:18:01.0000] <Hixie>
the reload thing works really well in webkit

[04:20:00.0000] <krijnh>
gDashiva: Yeah, but that's because Hixie kicks people who don't agree with him :)

[04:20:01.0000] <Hixie>
hah

[04:20:02.0000] <Hixie>
i haven't kicked anyone like ever from this channel :-P

[04:20:03.0000] <krijnh>
Luckily such lines aren't logged by me :)

[04:20:04.0000] <gDashiva>
Even if that were true, it would still be consensus in the channel

[04:20:05.0000] <Hixie>
actually we lost ops years ago

[04:20:06.0000] <Lachy>
 /kick krijnh

[04:20:07.0000] <krijnh>
:+

[04:20:08.0000] <gDashiva>
That's how the w3c consensus works, isn't it? They just keep people with common sense locked out :P

[04:20:09.0000] <Hixie>
hey woah watch out he does our logs!

[04:21:00.0000] <krijnh>
Yeah, watch out for my powers!

[04:22:00.0000] <Lachy>
there's a way to get ops back into the channel. I think it involves registerring the channel somehow

[04:23:00.0000] <gDashiva>
You need op to register, though :)

[04:23:01.0000] <Hixie>
i'm sad that nobody loves my little update tracking thingy

[04:23:02.0000] <Hixie>
except me

[04:23:03.0000] <Lachy>
I haven't seen it work yet

[04:23:04.0000] <krijnh>
This kicking thing reminds me of #css on Quakenet, when CounterStrike:Source was just released :P

[04:23:05.0000] <gDashiva>
They can probably make an exception, but the usual wisdom is that if the channel members can't coordinate a channel clearing, they aren't coordinated enough to need ops

[04:23:06.0000] <Hixie>
Lachy: odd, what browser?

[04:24:00.0000] <Lachy>
probably cause I haven't been paying attention. what exactly does it do?

[04:24:01.0000] <Hixie>
if you check the checkbox, it should let you know within 30 seconds of the spec being updated

[04:25:00.0000] <Lachy>
oh, nice

[04:25:01.0000] <Hixie>
(checking the checkbox also forces an immediate check)

[04:25:02.0000] <Lachy>
can you make the automatic updates checkbox more visible?

[04:25:03.0000] <Hixie>
sure

[04:25:04.0000] <Hixie>
where should i put it

[04:25:05.0000] <Lachy>
maybe put it up with the login box in the top right of the page?

[04:25:06.0000] <Hixie>
ok

[04:26:00.0000] <Lachy>
or in the Version History list

[04:29:00.0000] <Lachy>
Hixie, why don't you combine revision.dat and revision-message.dat, so it doesn't take 2 requests to get the info that could be obtained with one

[04:30:00.0000] <Lachy>
just make revision.dat contain a string like:

[04:30:01.0000] <Lachy>
r2006 Fixed Fix validation errors.

[04:30:02.0000] <Lachy>
and then split on the first space

[04:30:03.0000] <Lachy>
s/Fixed //

[04:31:00.0000] <Hixie>
mostly to make the backend easier

[04:31:01.0000] <Hixie>
it's all managed from shell scripts

[04:32:00.0000] <Lachy>
ok.

[04:32:01.0000] <Lachy>
Now that there's an API, I might be able to look into providing a DOM diff feature

[04:37:00.0000] <Lachy>
Hixie, what's the point of appending the '?[todays date]' to the URL for revision.dat?

[04:39:00.0000] <Hixie>
poor man's no-cache

[04:39:01.0000] <Lachy>
oh, why don't you just use a real Cache-Control header?

[04:40:00.0000] <Hixie>
it was easier than working out the correct .htaccess fu :-)

[04:40:01.0000] <Lachy>
ok

[04:41:00.0000] <hsivonen>
Hixie: have you considered allowing stuff like <org.webkit.canvas>?

[04:41:01.0000] <Hixie>
it is allowed

[04:42:00.0000] <Hixie>
oh you mean as a tag name

[04:42:01.0000] <hsivonen>
in the tag name

[04:42:02.0000] <Hixie>
that would have a terrible accessibility story

[04:42:03.0000] <Hixie>
the whole point of using class="" is that it has a fallback element that at least somewhat provides some semantics

[04:42:04.0000] <zcorpan>
everyone would just use divs anyway

[04:42:05.0000] <Hixie>
e.g. <p class="note"> instead of <ch.hixie.spec.note>

[04:42:06.0000] <hsivonen>
how's that worse than <span class='org.webkit.canvas'>?

[04:43:00.0000] <Hixie>
for browser vendors, the extension strategy is to come to the working group, ask for input for a few days, then mint a new element like <canvas>

[04:43:01.0000] <hsivonen>
Hixie: I agree

[04:44:00.0000] <Hixie>
so we're not talking about browser vendors or things like <canvas> here

[04:45:00.0000] <hsivonen>
well, <org.purl.dc.3.2.phase-of-moon-at-document-creation>

[04:46:00.0000] <Hixie>
then yes, <span class="org.purl.dc.3.2.phase-of-moon-at-document-creation"> is better

[04:46:01.0000] <Lachy>
I don't really understand what kind of extensions they're asking for. If it's just for personal/organisational site-specific use, then there's no problem to solve. If it's for wider use in the community, then letting them mint their own extensions without communication is a bad idea

[04:46:02.0000] <hsivonen>
Hixie: why?

[04:46:03.0000] <Hixie>
hsivonen: because then UAs don't have to worry about whether the element is inline or block, they can just look at whether it's a <div> or a <span>

[04:47:00.0000] <hsivonen>
Lachy: as far as I can tell, the problem is having to ask Hixie or Tantek.

[04:47:01.0000] <Hixie>
and for cases where there is a more appropriate semantic, there is built-in fallback

[04:47:02.0000] <Hixie>
they don't have to ask tantek, they can mint their own class names as they wish

[04:47:03.0000] <hsivonen>
Hixie: isn't span-likeness the default?

[04:48:00.0000] <Hixie>
hsivonen: it could be, but then the moment you wanted a block-likeness, or an emphasis-likeness, or an aside-likeness (as i really want for class=note) then you'd have to change to a different syntax

[04:48:01.0000] <Hixie>
Lachy: updated the ui, is it better now?

[04:48:02.0000] <hsivonen>
Hixie: you'd just use Cascading Semantic Sheets

[04:49:00.0000] <Hixie>
hsivonen: do you mean literally cascading style sheets, or tongue-in-cheek a hypothetical cascading semantic sheets?

[04:49:01.0000] <Lachy>
Hixie, yes

[04:49:02.0000] <hsivonen>
Hixie: both. 

[04:49:03.0000] <hsivonen>
Hixie: display: block;

[04:50:00.0000] <Hixie>
hsivonen: for CSS, that doesn't help in the no-CSS case. for the latter, i don't understand how that would work in practice. i suggested it years ago before i realised how intractable the problem was.

[04:50:01.0000] <Hixie>
hsivonen: (in principle, we can't rely on CSS.)

[04:51:00.0000] <Hixie>
hsivonen: (and relying on CSS doesn't help turn the element into <aside> or <em> or whatever in media that i haven't written a sheet for)

[04:51:01.0000] <Hixie>
hsivonen: (which was in fact the very problem that came up today)

[04:51:02.0000] <hsivonen>
Hixie: I'm not seriously suggesting separation of semantics and structure

[04:52:00.0000] <hsivonen>
Hixie: but in practice, people (who don't consider accessibility of extensions) seems to be OK with including CSS for their custom stuff

[04:53:00.0000] <Hixie>
in practice, people (who don't consider accessibility) seem to be OK with doing everything with <font> and <table> and <img> and don't bother putting alt="" on their <img>s either

[04:53:01.0000] <Hixie>
but that doesn't mean we should allow any of that

[04:54:00.0000] <hsivonen>
/me is annoyed at dojotoolkit.org breaking Open Link in New Window

[04:55:00.0000] <hsivonen>
Hixie: not allowing it is centralized command

[04:55:01.0000] <Hixie>
?

[04:55:02.0000] <Hixie>
can you rephrase that? i don't understand what you mean

[04:56:00.0000] <hsivonen>
Hixie: saying that people aren't allowed to do stuff is not Decentralized.

[04:56:01.0000] <Hixie>
and?

[04:57:00.0000] <hsivonen>
Decentralized Good, Centralized Bad, it seems :-)

[04:57:01.0000] <Hixie>
well then let's disband the w3c, the whatwg, the ietf, iana, and give up on web standards

[04:58:00.0000] <Philip`>
That's actually quite a good plan

[04:58:01.0000] <Hixie>
not if we want any kind of interoperability

[05:04:00.0000] <jgraham>
Hixie: Is there any actual problem with disallowing classnames that look like URIs but do not correspond to the requirements for minting URIs (e.g. that they are owned by the person who owns the domain name in the case of tag: URIs)

[05:04:01.0000] <Hixie>
it won't make him happy, is the problem

[05:04:02.0000] <Hixie>
because despite his love affair with uris, he also thinks they're ugly

[05:04:03.0000] <hsivonen>
Hixie: you could do with what XForms does with inputtypes

[05:04:04.0000] <jgraham>
It would solve about half the issues he has raised

[05:04:05.0000] <hsivonen>
and WF2 by reference

[05:05:00.0000] <Hixie>
which is?

[05:05:01.0000] <hsivonen>
Hixie: if the string contains a colon, it must be an absolute URL

[05:05:02.0000] <hsivonen>
basically denying the colon to those who don't love URIs an identifiers

[05:05:03.0000] <gDashiva>
That would kill off namespaces too :)

[05:05:04.0000] <jgraham>
hsivonen: URLs are worse than URIs in this case I think

[05:06:00.0000] <hsivonen>
jgraham: well, absolute IRI then

[05:06:01.0000] <hsivonen>
because URI wouldn't be internationalized

[05:06:02.0000] <hsivonen>
and we wouldn't want to leave things uninternationalized

[05:06:03.0000] <jgraham>
Yeah, OK :)

[05:07:00.0000] <gDashiva>
I'm wondering, though. What is there to prevent me from creating an awesome vocabulary using some company I don't relate to for the namespace URL?

[05:07:01.0000] <hsivonen>
gDashiva: you don't have the Authority!

[05:08:00.0000] <jgraham>
gDashiva: There are MUST NOT conditions!

[05:08:01.0000] <gDashiva>
So Hixie could add "@class identifiers MUST NOT conflict with other identifiers made by people who worry about conflicting identifiers"

[05:10:00.0000] <krijnh>
(why don't you guys talk about this with Julian, in #html-wg? You're all there as well)

[05:12:00.0000] <gDashiva>
We need to establish consensus first

[05:14:00.0000] <Lachy>
gDashiva, do you mean we need to get everyone to agree to move the conversation to the other channel, before we can move it?

[05:16:00.0000] <gDashiva>
Lachy: Yes, otherwise it wouldn't be backwards compatible with all the chatters

[05:16:01.0000] <Lachy>
alright. If that's what we're going to do, then we'd better organise an official survey announce it, give everyone sufficient time to answer it, and then come back in a couple of weeks with the results :-)

[05:17:00.0000] <gDashiva>
/me waves to the log watchers.

[05:18:00.0000] <Lachy>
I'm surprised by the total lack of discussion about the alt attribute changes.

[05:18:01.0000] <gDashiva>
I'm surprised the video thread died out so fast

[05:18:02.0000] <hsivonen>
we could just try switching and see if connections to irc.w3.org stay up and whether there will be Formal Objections

[05:18:03.0000] <krijnh>
Lachy: please don't be! I'm happy the way it is :)

[05:18:04.0000] <Lachy>
we had a total of 2 emails giving feedback about the spec, and one responding to say it was fixed.

[05:19:00.0000] <Lachy>
I'm happy too. Just very surprised.

[05:19:01.0000] <Lachy>
oh, and we also had one blog entry discussing process issues, rather than any technical issues.

[05:19:02.0000] <gDashiva>
hsivonen: We all know nothing good can come out of such unilateral implementations without sufficient evaluation in a peer group

[05:22:00.0000] <Hixie>
video thread?

[05:23:00.0000] <Hixie>
hsivonen: switching what?

[05:23:01.0000] <hsivonen>
Hixie: switching to discussing on #html-wg

[05:24:00.0000] <Hixie>
ah

[05:24:01.0000] <Lachy>
http://standardssuck.org/grddl-bridging-the-interwebs

[05:25:00.0000] <gDashiva>
Lachy: I don't have headphones, are there transcripts available?

[05:25:01.0000] <Lachy>
not yet. But you can try lip reading.

[05:26:00.0000] <Lachy>
"If you look at microformats just right, it's really RDF" Hah! :-)

[05:26:01.0000] <gDashiva>
Isn't everything RDF if you squint enough?

[05:27:00.0000] <Lachy>
he's giving a Matrix analogy

[05:27:01.0000] <Lachy>
talking about red and blue pills

[05:27:02.0000] <gDashiva>
Oh

[05:27:03.0000] <gDashiva>
heh

[05:31:00.0000] <zcorpan>
http://www.w3.org/2002/09/wbs/32212/calls4html2008aug/ 

[05:32:00.0000] <hsivonen>
zcorpan: Has that questionnaire been announced to the HTML WG?

[05:33:00.0000] <zcorpan>
hsivonen: not afaik

[05:33:01.0000] <hsivonen>
zcorpan: is it for PFWG only?

[05:33:02.0000] <zcorpan>
hsivonen: not afaict

[05:36:00.0000] <zcorpan>
or maybe it is

[05:39:00.0000] <Lachy>
Decentralized Extensibility is to Namespaces what Intelligent Design is to Creationism" - hsivonen, do you mean Decentralised extensibility and namespaces are exactly the same thing?

[05:39:01.0000] <gDashiva>
No, that ID is a sneaky way to attempt to get acceptance for C

[05:39:02.0000] <Lachy>
because ID and creationism as the same thing, with a different name

[05:39:03.0000] <hsivonen>
Lachy: I mean that Decentralized Extensibility is a marketing fig leaf for Namespaces

[05:40:00.0000] <Lachy>
ok

[05:43:00.0000] <Hixie>
the red and blue pills is where i stopped watching that podcast

[05:43:01.0000] <Hixie>
i couldn't take it anymore

[05:43:02.0000] <Hixie>
too much truth that i couldn't handle, maybe, you decide :-)

[05:44:00.0000] <hsivonen>
http://annevankesteren.nl/2007/04/html-red-pill

[05:44:01.0000] <Lachy>
I haven't finished watching it either.

[05:44:02.0000] <gDashiva>
anne's post reminds me of final destination

[05:46:00.0000] <Hixie>
i'll be interested to see how sam responds to my last e-mail

[05:59:00.0000] <Lachy>
How can Julian claim that clashing namespace prefixes has never been a problem in practice, yet be so adamant about clashing class names being a problem?

[06:00:00.0000] <gDashiva>
Apparently there are lots of malicious authors who would gladly steal your class name and abuse it, but would never consider doing the same with a namespace because namespaces are holy

[06:02:00.0000] <hsivonen>
gDashiva: just like the XHTML namespace is holy and the XHTML2 WG has Authority over it despite browser vendors having the Power to pollute it with <canvas>.

[06:03:00.0000] <Hixie>
it's because nobody uses namespaces

[06:04:00.0000] <Hixie>
/me ducks

[06:05:00.0000] <gDashiva>
Hunt-and-peck = duck typing?

[06:07:00.0000] <Lachy>
nobody would use URI based class names either. They don't solve any real problems, that aren't solved by a simple organisational prefix, and even the benefits of such a prefix are questionable for private class names.

[06:08:00.0000] <gDashiva>
Hixie: What you suggested about only complicating the class=math classname, wouldn't that lead to the same copypaste problems namespaces is drowning us in?

[06:09:00.0000] <Hixie>
how so?

[06:09:01.0000] <Lachy>
so, I guess my problem is that I'm don't really understand the intended use cases they're trying to solve with whatever it is they're asking for, and it's not really that clear to me what they're asking for since URIs are supposedly ugly, yet they want a URI-based extension mechanism

[06:09:02.0000] <gDashiva>
Since the internal classnames would mean different things if you pasted them into a different container classname

[06:10:00.0000] <Lachy>
gDashiva, doesn't seem to have been a problem for microformats

[06:11:00.0000] <hsivonen>
Lachy: how do we really know what's a problem for microformats when parsing them is not documented, so we cannot evaluate parsing against real content

[06:11:01.0000] <Hixie>
yeah i really don't fully understand what is being requested either

[06:12:00.0000] <Hixie>
gDashiva: well that's like <circle> in svg not meaning anything useful outside of an <svg> container

[06:12:01.0000] <Hixie>
gDashiva: which isn't really a big problem

[06:12:02.0000] <Lachy>
maybe it would be better to wait and see if some sort of distributed extensibility design pattern emerges, and then once such a thing as proven itself, then we can spec it

[06:13:00.0000] <gDashiva>
Hixie: Yes. But if we entertain the idea of collisions, there could be a SillyVectorGraphics container and the cirle would be sillified when pasted there.

[06:13:01.0000] <Hixie>
Lachy: like, say, microformats? :-)

[06:13:02.0000] <Lachy>
so if a design pattern ever emerges that says to use reverse-dns domains, like class="org.example-foobar", and it becomes popular we could spec it

[06:13:03.0000] <Hixie>
gDashiva: yes, indeed.

[06:13:04.0000] <Lachy>
but, yeah, microformats seems to do well in that space anyway

[06:14:00.0000] <Hixie>
gDashiva: i don't see that as happening much though

[06:15:00.0000] <gDashiva>
Hixie: Agreed, but when the premise of disagreement is that collisions will happen, I thought it might be useful to see where that takes us :)

[06:17:00.0000] <Hixie>
gDashiva: it'll take us to the same place as people copying an <li> from an <ol> to a <ul>, or a <dt> from a <dl> to a <dialog>

[06:21:00.0000] <Hixie>
class="" is new? good to know

[06:21:01.0000] <Hixie>
(doesn't it predate xml, let alone xmlns?)

[06:23:00.0000] <Hixie>
ok, bed time (long past)

[06:24:00.0000] <Hixie>
nn

[06:33:00.0000] <gDashiva>
... did I disconnect?

[06:34:00.0000] <gDashiva>
Hope our logger has a stable connection :)

[06:34:01.0000] <krijnh>
zzz

[06:36:00.0000] <gDashiva>
krijnh: Is your tagline list public?

[06:36:01.0000] <krijnh>
gDashiva: no :)

[06:36:02.0000] <gDashiva>
So you want us to brute force it, then? :)

[06:36:03.0000] <krijnh>
Well, in a way it is

[06:37:00.0000] <krijnh>
What do you want?

[06:37:01.0000] <zcorpan>
krijnh: how about an ajaxy button for the next tagline?

[06:37:02.0000] <gDashiva>
Something like that

[06:37:03.0000] <zcorpan>
/me has reloaded the page to see all taglines :)

[06:37:04.0000] <krijnh>
Really?

[06:37:05.0000] <krijnh>
So that's why my server load is so high

[06:37:06.0000] <gDashiva>
zcorpan: Are they sequential?

[06:37:07.0000] <krijnh>
I'm on ADSL lite you know..

[06:37:08.0000] <zcorpan>
gDashiva: think so

[06:38:00.0000] <krijnh>
Are they?

[06:38:01.0000] <krijnh>
Cool feature

[06:38:02.0000] <zcorpan>
or maybe they aren't, don't remember

[06:38:03.0000] <krijnh>
There are 30 of them, collect 'm all!

[06:38:04.0000] <gDashiva>
krijnh: You should use semantic HTML to reduce download size

[06:39:00.0000] <krijnh>
Where can I download that?

[06:40:00.0000] <gDashiva>
Dunno

[06:40:01.0000] <Lachy>
hmm, I never noticed the tag line thing before

[06:40:02.0000] <krijnh>
Damn, it was developed especially for you! :(

[06:40:03.0000] <Lachy>
though I never go to the index page, since i have direct links to the latest logs of each channel in my history

[06:42:00.0000] <krijnh>
http://krijnhoetmer.nl/irc-logs/?gimme-a-hobby

[06:42:01.0000] <gDashiva>
That's a lot of style

[06:42:02.0000] <Lachy>
krijnh, are all these tag lines quotes from people on the maling list? Some of the sound familiar

[06:43:00.0000] <zcorpan>
krijnh: now make the tagline link to that 

[06:44:00.0000] <krijnh>
Lachy: Some of them are

[06:44:01.0000] <krijnh>
I just thought it was funny

[06:44:02.0000] <krijnh>
And wasting time is one of the main reasons I joined the HTML WG (in retrospect), so why not :)

[06:45:00.0000] <Lachy>
krijnh, can you add an API for others to add new tag lines?

[06:45:01.0000] <Lachy>
maybe if we type a message beginning with "tag:", it could be added to your list

[06:45:02.0000] <krijnh>
Yeah, just pm me

[06:46:00.0000] <zcorpan>
tag: [off] will this show up?

[06:46:01.0000] <krijnh>
Yes

[06:46:02.0000] <krijnh>
It's only hidden in #webapps

[06:47:00.0000] <zcorpan>
if i write that in #webapps?

[06:47:01.0000] <zcorpan>
or is the tag: feature only for #whatwg?

[06:47:02.0000] <krijnh>
It's no feature

[06:47:03.0000] <krijnh>
It's a made up thing by Lachy :)

[06:47:04.0000] <zcorpan>
bug?

[06:47:05.0000] <zcorpan>
well you must implement it!

[06:48:00.0000] <zcorpan>
or i won't use your logs

[06:48:01.0000] <zcorpan>
/me invites zakim

[06:49:00.0000] <krijnh>
:'(

[06:49:01.0000] <krijnh>
/me tries to come with something similar for Opera..

[06:50:00.0000] <zcorpan>
but you can't because opera has all features between the earth and the sun :)

[06:51:00.0000] <krijnh>
Indeed

[06:52:00.0000] <Lachy>
krijnh, has the [off] feature been used in #webapps at all for a legitimate purpose since the logs were officially activated?

[06:53:00.0000] <krijnh>
You shouldn't ask that question :)

[06:53:01.0000] <krijnh>
I think you know the answer already

[06:53:02.0000] <Lachy>
I actually don't. Though I could check my own logs, I guess

[06:54:00.0000] <zcorpan>
if you find something, could you quote it here? :P

[06:55:00.0000] <krijnh>
Cool, it has been

[06:55:01.0000] <gDashiva>
Lachy: It was used earlier today, by arve :)

[06:55:02.0000] <Lachy>
one said "[off] *sigh*"

[06:55:03.0000] <krijnh>
By an Opera employee :)

[06:56:00.0000] <zcorpan>
me?

[06:56:01.0000] <krijnh>
ARve

[06:57:00.0000] <Lachy>
I wouldn't really count that as legitimate use of it though, since it was designed to hide confidential information

[06:58:00.0000] <krijnh>
Jep

[06:59:00.0000] <gDashiva>
Have there been any more cases of 'oops, we are in a different channel' since that one time?

[08:08:00.0000] <virtuelv>
well, my sigh from earlier today was hardly relevant to what I was getting at

[08:24:00.0000] <hsivonen>
Hixie: what should I put into alt in the Image Report under the new spec text? alt="{image}"? That seems daft.

[08:36:00.0000] <Lachy>
hsivonen, alt={} might be reasonable for your case

[08:37:00.0000] <zcorpan>
jgraham: your outline tester doesn't work with entities

[08:37:01.0000] <Lachy>
is this for images from the website being validated, that are listed in image report table??

[08:38:00.0000] <zcorpan>
jgraham: other than the 5 xml entities

[08:38:01.0000] <hsivonen>
Lachy: yes

[08:41:00.0000] <zcorpan>
hsivonen: "An image in an e-mail or document intended for a specific person who is known to be able to view images" still says alt may be omitted

[08:43:00.0000] <Lachy>
I asked why that is still in the spec, but Hixie seemed to miss that question when he responded to my mail

[08:43:01.0000] <Lachy>
I don't see why emails should be treated differently now that there's the {...} syntax

[08:44:00.0000] <Lachy>
they could just use alt={} or some auto generated thing like alt="{attached image}"

[08:49:00.0000] <hsivonen>
my crystal ball tells me that alt='{attached image}' will be in the UI language of the sender--not in the language of the message

[08:49:01.0000] <hsivonen>
in practice that is

[08:50:00.0000] <Lachy>
in practice, my UI language is the language I write in. So then I guess alt={} would be better, since it allows the recipient to be told in whatever langauge they use

[08:53:00.0000] <Lachy>
Hixie, please respond to the last question at the end of this mail http://lists.w3.org/Archives/Public/public-html/2008Aug/0090.html

[09:04:00.0000] <takkaria>
hehe, request on the libxml2 mailing list for libxml2 to silently ignore NUL bytes in XML rather than error on it

[09:05:00.0000] <gsnedders>
That's against the spec1111!11!!

[09:05:01.0000] <hsivonen>
takkaria: database dump use case?

[09:10:00.0000] <takkaria>
hsivonen: no idea for the use case, just someone complaining that libxml2 errors

[09:10:01.0000] <takkaria>
I have this sneaking suspicion that they may end up putting an input layer before libxml2 that silently ignores NULs

[09:54:00.0000] <gDashiva>
I wonder if anyone will ever suggest [on], so you can do like We don't want to give up our [off] secrets [on] to people like you!

[09:54:01.0000] <takkaria>
[off] does this not get logged then?

[09:54:02.0000] <gDashiva>
In here it does

[09:54:03.0000] <gDashiva>
But #webapps has special treatment

[09:55:00.0000] <takkaria>
ah

[09:59:00.0000] <Lachy>
heh, [off] is getting more use in here where it doesn't work, than it is in #webapps :-)

[09:59:01.0000] <zcorpan>
[off] [off] [off] [off] [off] [off] [off] [off] [off] [off] [off] [off] [off]

[10:02:00.0000] <gDashiva>
4whoa

[10:02:01.0000] <jcranmer>
stress testing?

[10:02:02.0000] <gDashiva>
Netsplit maybe

[10:03:00.0000] <jcranmer>
11:54 -!- Netsplit brown.freenode.net <-> irc.freenode.net quits:

[10:03:01.0000] <jcranmer>
but that doesn't explain zcorpan's comment (AFAIK)

[10:04:00.0000] <zcorpan>
/me updates http://wiki.whatwg.org/wiki/FAQ#What_is_the_namespace_declaration.3F to match the spec

[10:04:01.0000] <zcorpan>
but it doesn't talk about foreign lands

[10:05:00.0000] <zcorpan>
[off] has the feature in here to make random people disconnect

[10:50:00.0000] <gsnedders>
Hixie: for no-num and no-toc, for header, should they be on the header element, or on the first highest h1–h6?

[11:00:00.0000] <zcorpan>
/me thinks Hixie is making an experiment with alt to see if accessibility experts are actually reading the spec or if they'll go "yep the spec is fine now that alt is always required"

[11:25:00.0000] <zcorpan>
Hixie: http://validator.whatwg.org/ has the old name in the link text

[13:01:00.0000] <davidmccabe>
Hixie: Hello. I'm considering writing up a proposal for drag-and-drop between browser windows. And I'm told you're the guy to talk to.

[13:01:01.0000] <davidmccabe>
I didn't find anything with google, but I thought I'd ask whether this has been proposed before to anyone's recollection. 

[13:04:00.0000] <davidmccabe>
To summarize what I'm imagining: You would be able to set certain properties on DOM elements that make them draggable or droppable, and a javascript object would be passed from one window to the other when a drop is completed. 

[13:04:01.0000] <davidmccabe>
The protocol could be modeled after those in Cocoa for instance. 

[13:16:00.0000] <roc_>
HTML5 has a drag-drop spec

[13:16:01.0000] <roc_>
take a look

[13:17:00.0000] <roc>
you can't pass JS objects as drag data (that would present some problems), but you can pass strings

[13:21:00.0000] <davidmccabe>
Hah.

[13:21:01.0000] <davidmccabe>
I grepped the spec; how did I miss it?

[13:23:00.0000] <Hixie>
hsivonen: alt="{image from site}" maybe, or alt="{small image}" vs alt="{large image}", or some other categorising information that actually showing the image would convey (like size, transparency, colour) but that you would never actually write next to the image if the image was visible

[13:24:00.0000] <Hixie>
zcorpan: old name of what?

[13:24:01.0000] <Hixie>
davidmccabe: http://www.whatwg.org/specs/web-apps/current-work/#dnd

[13:25:00.0000] <davidmccabe>
Found it; sorry to bother you.

[13:25:01.0000] <Lachy>
this is weird, there seems to be a distinct lack of excitement over the upcoming olympic games ceremony in this country. I thought people here would be much more excited, but so far I've only heard from people who aren't that interested in seeing it

[13:25:02.0000] <Hixie>
np

[13:25:03.0000] <Dashiva>
Lachy: What country?

[13:25:04.0000] <Lachy>
Norway

[13:25:05.0000] <Lachy>
where I'm stuck living for now

[13:26:00.0000] <Dashiva>
Well, that's easy. Norway is all about winter olympics :)

[13:26:01.0000] <gsnedders>
/me is in Sweden

[13:26:02.0000] <Lachy>
wtf? who cares about the winter olympics?!

[13:26:03.0000] <gsnedders>
Lachy: Me!

[13:26:04.0000] <Dashiva>
Norway does

[13:26:05.0000] <gsnedders>
They're better than the summer ones!

[13:26:06.0000] <Lachy>
it's not as if Australians are going to win many medals at that one

[13:26:07.0000] <gsnedders>
The skeleton takes _real_ bravery

[13:26:08.0000] <Dashiva>
Lachy: Take your situation and reverse it to get Norway :P

[13:27:00.0000] <Lachy>
but we're champions in the pool, so I've gotta make sure I see those events

[13:27:01.0000] <Lachy>
especially the 1500

[13:27:02.0000] <Lachy>
I hope Hackett wins that one again

[13:27:03.0000] <Dashiva>
Norway gets like a bronze in archery and a silver in solo rowing if we're lucky

[13:28:00.0000] <gsnedders>
Dashiva: You're from Norway? Ah.

[13:28:01.0000] <wilhelm>
There is a summer olymptics too? (c:

[13:28:02.0000] <Lachy>
wilhelm, yes!

[13:28:03.0000] <gsnedders>
/me knew you were from Scandinavia

[13:37:00.0000] <Dashiva>
gsnedders: Yes

[13:38:00.0000] <gsnedders>
Hixie: Answer me! ("for no-num and no-toc, for header, should they be on the header element, or on the first highest h1–h6?")

[13:40:00.0000] <Hixie>
oops, missed your question!

[13:41:00.0000] <Hixie>
you mean where should you look for it?

[13:41:01.0000] <Hixie>
i'd say h1-h6 element

[13:41:02.0000] <gsnedders>
Hixie: yeah

[13:41:03.0000] <Hixie>
for compat with bert's

[13:41:04.0000] <Hixie>
or just assume it for header descendants

[13:41:05.0000] <Hixie>
and don't look

[13:41:06.0000] <gsnedders>
Hixie: Nothing that uses Bert's uses <header> though, so it isn't really a compat issue

[13:41:07.0000] <Hixie>
it might become one if people migrate back :-)

[13:42:00.0000] <gsnedders>
Hixie: They won't!

[13:42:01.0000] <gsnedders>
/me grabs knive

[13:42:02.0000] <Hixie>
hah

[13:42:03.0000] <gsnedders>
*knife

[13:42:04.0000] <Hixie>
oh something came up in the public-html list that i was going to ask you to add to the specgen

[13:43:00.0000] <gsnedders>
Hixie: Well, it'd be useful to know what :)

[13:43:01.0000] <Hixie>
yeah i'm trying to work out how to phrase it

[13:44:00.0000] <Lachy>
gsnedders, adding "Note: ", "Warning: " and "**" to notes, warnings and issues

[13:45:00.0000] <Lachy>
replacing the current stylesheet approach

[13:45:01.0000] <Lachy>
that uses ::before and ::after

[13:45:02.0000] <Hixie>
basically yeah, but i have a specific markup style i'd like :-)

[13:46:00.0000] <gsnedders>
Hixie: Email, plz.

[13:46:01.0000] <Lachy>
/me suggests <strong class="ch.hixie.whatwg.spec.issue-marker">**</strong> :-)

[13:46:02.0000] <Hixie>
ok

[13:46:03.0000] <gsnedders>
(I'll probably lose it in IRC)

[13:46:04.0000] <Hixie>
e-mail addr?

[13:47:00.0000] <gsnedders>
geoffers⊙gc

[13:47:01.0000] <takkaria>
Lachy: not org.whatwg.spec.issue-marker?

[13:48:00.0000] <Lachy>
takkaria, I intentionally avoided using whatwg.org cause that would have been too obvious and not long enough

[13:48:01.0000] <Lachy>
gotta make sure it's globally unique

[13:49:00.0000] <gsnedders>
and it would conflict between w3c.org and whatwg.org

[13:49:01.0000] <Hixie>
yeah, i mean, you all have authority over whatwg.org too!

[13:49:02.0000] <takkaria>
hmm, maybe ch.hixie.whatwg.spec.html5.issue-marker, then? :P

[13:49:03.0000] <gsnedders>
Hixie: and power. Don't forget the power.

[13:49:04.0000] <Hixie>
none of us have any pwoer sadly

[13:50:00.0000] <Lachy>
takkaria, right. I forgot we needed to distinguish between webforms2 and webcontrols1

[13:50:01.0000] <gsnedders>
Oh, there are multiple defs of single terms in WF2

[13:57:00.0000] <Hixie>
gsnedders: you have mail

[13:57:01.0000] <Hixie>
don't worry abotu wf2

[13:58:00.0000] <gsnedders>
That's why I didn't mention it before :)

[13:58:01.0000] <Hixie>
:-)

[13:59:00.0000] <gsnedders>
That should be doable

[13:59:01.0000] <gsnedders>
/me shoves that into the 1.1 list

[13:59:02.0000] <gsnedders>
(1.0 is, for all intents and purposes, feature frozen)

[14:00:00.0000] <takkaria>
Hixie: after WF2 is integrated and the form stuff is specced, HTML5 is actually pretty much feature complete isn't it?

[14:00:01.0000] <Hixie>
pretty much

[14:00:02.0000] <Hixie>
rendering section and aria are still missing

[14:16:00.0000] <gsnedders>
/me hopes he hasn't missed anyone out from the ack section of the spec-gen docs

[14:17:00.0000] <gsnedders>
"Andrew Sidwell, Anne van Kesteren, Henri Sivonen, Ian Hickson, James Graham, Lachlan Hunt, Magnus Kristiansen, Michael(tm) Smith, and Philip Taylor" + special thanks to Bert

[14:27:00.0000] <gsnedders>
Anyone think they should be in that list, do bully me

[14:29:00.0000] <gsnedders>
I also think I ought to stop hotlinking the whatwg style sheet ;P

[14:29:01.0000] <Hixie>
why?

[14:29:02.0000] <Hixie>
:-)

[14:29:03.0000] <takkaria>
gsnedders: I have no idea why I'm in there, but fair enough. :)

[14:30:00.0000] <gsnedders>
Hixie: So the docs work without an internet connection

[14:32:00.0000] <Hixie>
ah

[14:32:01.0000] <Hixie>
well

[14:32:02.0000] <Hixie>
details

[14:34:00.0000] <gsnedders>
takkaria: I was doing it from memory, and I think you've been at least as helpful as Dashiva :)

[14:36:00.0000] <virtuelv>
does anyone know anything about http://www.netsurf-browser.org/  ?

[14:38:00.0000] <hasather_>
virtuelv: takkaria I think :)

[14:38:01.0000] <Dashiva>
gsnedders: I don't even recall being helpful

[14:39:00.0000] <gsnedders>
Dashiva: Helpful by stopping me from getting bored :)

[14:39:01.0000] <Dashiva>
So the bar for getting credit is pretty low then :)

[14:40:00.0000] <takkaria>
Rob Burns might also qualify

[14:41:00.0000] <gsnedders>
takkaria: he doesn't make me laugh, though

[14:42:00.0000] <gsnedders>
http://spec-gen.gsnedders.com/ — a website!

[14:43:00.0000] <virtuelv>
gsnedders: does that do all of what the css3 module post-processor does?

[14:44:00.0000] <gsnedders>
virtuelv: No. It does most of what people use, though. All of what Hixie uses, for example.

[14:44:01.0000] <virtuelv>
/me checks out a copy

[14:45:00.0000] <gsnedders>
virtuelv: http://stuff.gsnedders.com/spec-gen — examples

[14:45:01.0000] <takkaria>
gsnedders: the link to w3.org in section 3 is preceded by a ">"

[14:45:02.0000] <gsnedders>
takkaria: in the docs?

[14:46:00.0000] <gsnedders>
Oh, yeah

[14:50:00.0000] <gsnedders>
takkaria: fixed

[14:55:00.0000] <Hixie>
is there any way to create a DOM Document object in IE that doesn't involve an HTML parser?

[14:57:00.0000] <Hixie>
 w(new ActiveXObject("Msxml.DOMDocument").readyState);

[14:57:01.0000] <Hixie>
...returns 4.

[14:57:02.0000] <Hixie>
wtf.

[15:17:00.0000] <Hixie>
should we rename "irrelevant" to "hidden"? It's what we have on <command> for the same basic semantic...

[15:17:01.0000] <Hixie>
but I don't want people to just use it to hide things, it's more subtle than that...

[15:17:02.0000] <Hixie>
though i guess nobody will understand the subtleness...

[15:17:03.0000] <Hixie>
hmm.

[15:18:00.0000] <gsnedders>
Hixie: irrelevant, plz.

[15:18:01.0000] <gsnedders>
:P

[15:18:02.0000] <Hixie>
it's harder to type, and people don't get it

[15:19:00.0000] <gsnedders>
Hixie: then @useless?

[15:19:01.0000] <Lachy>
Hixie, hidden is better

[15:19:02.0000] <Hixie>
ooh, that's not a bad name

[15:19:03.0000] <Lachy>
or ignore=""

[15:20:00.0000] <Hixie>
it would be ignored="" but yeah, that might work too

[15:20:01.0000] <Hixie>
one example of something for which i think this attribute isn't appropriate is hiding one panel of a tabbed panel in a dialog when another panel is active

[15:21:00.0000] <Hixie>
because the fact that the controls are tabbed isn't intrinsic to the dialog, it's just a presentation

[15:21:01.0000] <Lachy>
it might be a little unintuitive for e.g. <script ignored="">, since AIUI, the script would still execute.

[15:22:00.0000] <webben>
Hixie: inactive or ignore would be better than irrelevant or hidden.

[15:22:01.0000] <webben>
irrelevant is just confusing; hidden is confusing presentational

[15:22:02.0000] <webben>
*confusingly

[15:22:03.0000] <gsnedders>
/me waves g'nite

[15:23:00.0000] <webben>
IIRC someone was arguing re inactive that people would misspell it.

[15:23:01.0000] <Lachy>
whatever you choose, change hidden="" for command to be the same

[15:23:02.0000] <Lachy>
or just drop hidden from command, and let it use the global attribute

[15:24:00.0000] <jgraham>
inactive is good hidden and ignore is bad

[15:24:01.0000] <Lachy>
why is ignore bad?

[15:24:02.0000] <webben>
Lachy: because parsers shouldn't ignore the markup for instance

[15:24:03.0000] <webben>
and because events can still be fired for it (IIRC)

[15:24:04.0000] <jgraham>
ignore sounds too mch like "don't look at the man behind the curtian"

[15:25:00.0000] <jgraham>
If it should be ignored why is it there? 

[15:25:01.0000] <Lachy>
webben, so? That's why we're not calling it dont-parse=""

[15:25:02.0000] <jgraham>
Whereas inactive seems like a good description of the semantic

[15:26:00.0000] <Lachy>
inactive might work

[15:27:00.0000] <webben>
hmm on the other hand "Elements in a section hidden by the irrelevant attribute are still active, e.g. scripts and form controls in such sections still render execute and submit respectively"

[15:27:01.0000] <webben>
note "active"

[15:28:00.0000] <webben>
from http://www.whatwg.org/specs/web-apps/current-work/#the-irrelevant

[15:32:00.0000] <Lachy>
Hixie, the new spec update system has already proven much more useful than twitter

[15:32:01.0000] <Lachy>
I tend to ignore most twitter messages, so I rarely saw the whatwg ones

[15:33:00.0000] <Lachy>
/me goes to stop following whatwg twitts

[15:37:00.0000] <Hixie>
inactive is too much like "disabled"

[15:38:00.0000] <Hixie>
Lachy: cool, glad to hear it!

[15:44:00.0000] <Lachy>
Hixie, so Web Controls is now officially dead?

[15:44:01.0000] <Lachy>
I was wondering what was going to happen with it

[15:45:00.0000] <Hixie>
it's not officially anything

[15:45:01.0000] <Hixie>
i don't know what will happen to it

[15:45:02.0000] <Hixie>
certainly ARIA has somewhat addressed one of the big use cases for web controls 1.0

[15:45:03.0000] <Hixie>
though imho in a less good way, but that's another story

[15:46:00.0000] <Hixie>
i think i'm gonna drop autosubmit from <menu>

[15:46:01.0000] <Hixie>
it doesn't seem to have a compelling use

[15:47:00.0000] <Lachy>
what was it designed for originally?

[15:47:01.0000] <Hixie>
i'm not sure

[15:49:00.0000] <Lachy>
oh, it was designed for menus made with <select> that automatically submit, which is commonly done with JavaScript

[15:49:01.0000] <Lachy>
I think

[15:49:02.0000] <Hixie>
that's what i thought but the way it was defined only affected radio and checkbox inputs, and the example in the source that used autosubmit with a <Select> had JS that woudl take care of the submission anyway for legacy UAs

[15:49:03.0000] <Hixie>
so..

[15:50:00.0000] <Lachy>
is the disabled-javascript case not worth considering?

[15:51:00.0000] <Hixie>
i don't think it's worth a whole extra attribute here

[15:53:00.0000] <Lachy>
I couldn't see any example in the spec using autosubmi

[15:53:01.0000] <Lachy>
was it hidden in a comment?

[15:53:02.0000] <Hixie>
yes

[16:27:00.0000] <Philip`>
/me finally gets his box working again

[16:28:00.0000] <Philip`>
It kind of stopped networking, and then I guess it didn't like having around a year of software updates without being rebooted so it didn't start up trivially :-(

[16:28:01.0000] <jcranmer>
O_O

[16:29:00.0000] <Philip`>
Hixie: Is there some way you could modify your script that calls the multipage generator, so that if e.g. hypothetically the server running the multipage generator was done for some hours, it wouldn't start seven simultaneous requests to generate the multipage spec once it came back up?

[16:29:01.0000] <Hixie>
yeah, that should be possible

[16:30:00.0000] <Philip`>
Hopefully it shouldn't ever be necessary, so don't worry if it's a non-zero amount of effort :-)

[16:32:00.0000] <Hixie>
done

[16:33:00.0000] <Hixie>
(just changed -t 0 to -t 1 in the wget command line)

[16:33:01.0000] <Philip`>
/me won't test the changes

[16:33:02.0000] <Philip`>
so I'll just assume they'll work, and so I'll say thanks


2008-08-06
[17:46:00.0000] <Lachy>
Hixie, in web workers, this sentence is grammatically incorrect: "The origin and effective script origin of scripts running in workers are both the origin of the absolute URL given by the value of the URL attribute of the worker's WindowWorker object."

[17:47:00.0000] <Lachy>
it says "are both", but then it only mentions one thing

[17:48:00.0000] <Hixie>
in english, verbs are conjugated relative to the subject, not the object

[17:48:01.0000] <Lachy>
what?

[17:49:00.0000] <Hixie>
two apples are a pear

[17:49:01.0000] <Hixie>
not two apples is a pear

[17:49:02.0000] <Hixie>
similarly, an apple is two pears

[17:49:03.0000] <Hixie>
not an apple are two pears

[17:50:00.0000] <Lachy>
my problem is with the word both, not the word are

[17:50:01.0000] <SadEagle>
May I suggest "Both the origin and effective script origin of scripts running in workers are the absolute URL given by the value of the URL attribute of the worker's WindowWorker object."

[17:50:02.0000] <Hixie>
the word both in that sentence is also relating to the subjects

[17:50:03.0000] <Hixie>
a and b are both letters

[17:50:04.0000] <Lachy>
then SadEagle's suggestion makes more sense

[17:51:00.0000] <SadEagle>
Yeah, but that's a syntactic ambiguity. My mind parsed is the same as Lachy's. 

[17:51:01.0000] <Hixie>
fair enough

[17:51:02.0000] <Hixie>
changed

[17:52:00.0000] <SadEagle>
Hixie: may I abuse you to clarify the definition of CSS2.1's clip property?

[17:52:01.0000] <Hixie>
you can try but i make no promises of knowing the answer

[17:53:00.0000] <SadEagle>
It just seems to be defined to be off top-right corner for RTL direction, and everyone seems to implement to alway sbe off the top-left corner... Unless direction:rtl somehow reversed the meaning of left and right borders...

[17:53:01.0000] <Hixie>
i have no idea how it is supposd to interact with directionality

[17:54:00.0000] <SadEagle>
fair enough, thanks for your time :-)

[18:06:00.0000] <Hixie>
ok sicking has convinced me to remove Window from workers

[18:06:01.0000] <Hixie>
what should we call the global object?

[18:18:00.0000] <Philip`>
Linuk

[18:25:00.0000] <Hixie>
he suggested WorkerGlobalScope, with that object only containing two members, "self" pointing to itself, and "utils" pointing to a separate object with APIs

[18:25:01.0000] <Hixie>
so you'd do utils.openDatabase() etc

[18:26:00.0000] <Hixie>
not sure if you'd need to do |new utils.XMLHttpRequest()| or if we'd still have constructors in the global scope

[18:26:01.0000] <Hixie>
he vanished before i could ask

[18:29:00.0000] <Lachy>
Hixie, in step 6 of "When a script invokes the import(url) method...", it says "or gets prematurely aborted by the "kill a worker" algorithm below." - s/below/above/

[18:30:00.0000] <Hixie>
thx

[18:38:00.0000] <Lachy>
Hixie, is URL_MISMATCH_ERR a new exception code that you invented for this, or is it defined somewhere other than DOM3Core?

[18:38:01.0000] <Hixie>
invented

[18:38:02.0000] <Lachy>
ok

[18:39:00.0000] <Lachy>
I guess that's the reason for the "Code 19" issue after it

[18:39:01.0000] <Hixie>
it's on the list in the wiki :-)

[18:39:02.0000] <Lachy>
which page?

[18:40:00.0000] <Hixie>
exception codes

[18:42:00.0000] <Lachy>
oh, nice. SECURITY_ERR is useful.

[18:42:01.0000] <Lachy>
but there doesn't appear to be any response to your proposal for it on public-webapi

[18:45:00.0000] <Hixie>
we so need someone to own DOM Core and actually write the spec

[18:47:00.0000] <Lachy>
is there anyone in webapps who's supposed to be working on it?

[18:47:01.0000] <Lachy>
or is zcorpan still intending to write DOM5 one day?

[18:47:02.0000] <Hixie>
dunno

[18:49:00.0000] <Lachy>
ok. well, if no-ones working on it by the time I get Selectors API to CR, then I might pick it up.

[18:50:00.0000] <Lachy>
also, I might have to incorporate :context/:scope directly into Selectors API. I got no response from the CSSWG saying whether or not the proposal has been accepted or rejected, and it really needs to get done before browsers ship with selectors api

[18:51:00.0000] <Lachy>
I should probably wait till anne gets back, since hes our csswg rep

[18:56:00.0000] <Hixie>
i've found that just doing things and letting people know is an effective incentive for them to get their act together

[18:56:01.0000] <Hixie>
ask for forgiveness, not permission, and all :-)

[18:59:00.0000] <jcranmer>
hmm, that's the first and, to date, only discussion I've participated in on the W3C mailing list

[19:01:00.0000] <jcranmer>
(debate over whether or not scoping stylesheets should refer to rerooting the selection tree or merely limiting the possible set of nodes to match)

[19:03:00.0000] <Lachy>
jcranmer, looks like you've participated in 2 other threads as well

[19:04:00.0000] <Lachy>
with just one message each, though

[19:04:01.0000] <jcranmer>
I made a response and didn't follow up

[19:04:02.0000] <jcranmer>
that doesn't count in my book

[19:04:03.0000] <Lachy>
ok, so participation requires at least 2 posts per thread?

[19:04:04.0000] <jcranmer>
well, maybe the later one counts

[19:05:00.0000] <jcranmer>
since I did rebutt a point

[19:05:01.0000] <jcranmer>
the other one was "oh, we've already done that, look here:"

[19:05:02.0000] <Lachy>
I like people who contribute once, say what they need and leave. It makes reading the thread so much easier and non-repetitive

[19:06:00.0000] <jcranmer>
I guess I prefer my discussions... back-and-forth

[19:06:01.0000] <jcranmer>
so long as both sides are understanding each other

[19:06:02.0000] <jcranmer>
if not, it's only a giant shouting match

[19:06:03.0000] <Lachy>
Hixie, the workers spec would be a lot easier for me to comprehend if the unwritten tutorial section was actually written demonstrated how the API was intended to be used

[19:07:00.0000] <Hixie>
yeah i intend to write that section tonight

[19:07:01.0000] <Lachy>
oh good. Then I should have waited a day before reviewing the spec :-)

[19:07:02.0000] <Hixie>
heh

[19:07:03.0000] <Hixie>
i just checked in big changes

[19:08:00.0000] <Lachy>
well, I aint reading it again yet it was hard enough the first time

[19:08:01.0000] <Hixie>
hah

[19:16:00.0000] <Hixie>
hmmmm

[19:17:00.0000] <Hixie>
one way to solve the problem that was being discussed the other day is to make the first port that a worker receives be automatically assigned to a variable in the global scope

[19:17:01.0000] <Hixie>
that way you can never hit the case where you GC straight away

[19:20:00.0000] <Hixie>
ports could also be simplified by having them queue events until such time as someone enables them manually

[23:59:00.0000] <hsivonen>
Hixie: ARIA has aria-hidden=true for irrelevant=''

[00:27:00.0000] <hsivonen>
I just realized I broke my promise not to use SVG in hypothetical examples.

[00:52:00.0000] <Hixie>
hsivonen: aria-hidden="" isn't accessible, as i understand it (since it doesn't do anything except for ATs)

[00:54:00.0000] <hsivonen>
Hixie: you are supposed to use it together with *[aria-hidden="true"] { display:none; }

[00:55:00.0000] <hsivonen>
Hixie: doesn't work with non-CSS non-AT UAs, though.

[00:55:01.0000] <hsivonen>
unless those UAs start triggering on ARIA

[00:56:00.0000] <Hixie>
CSS is optional, anything that relies on CSS is not accessible almost by definition :-)

[00:56:01.0000] <Hixie>
isn't triggering on ARIA specifically counter to ARIA?

[00:57:00.0000] <hsivonen>
Hixie: yes and yes

[00:57:01.0000] <hsivonen>
are there contemporary UAs that support JS but not CSS?

[02:01:00.0000] <Hixie>
heycam: can't we have some flag on the interface that says that HasProperty returns true for anything that NamedGetter would find, or something?

[02:01:01.0000] <Hixie>
i don't want to have to spec this all over the place if i can help it :-)

[02:02:00.0000] <heycam>
Hixie, but it's more than that.  Object.prototype.hasOwnProperty() returns true, too.

[02:02:01.0000] <heycam>
so they're real properties

[02:02:02.0000] <Hixie>
ok, well then a flag that says that then

[02:02:03.0000] <Hixie>
(i have no idea what that means :-) )

[02:02:04.0000] <Hixie>
/me looks up hasOwnProperty

[02:02:05.0000] <heycam>
so, ecma-262 defines Object.prototype.hasOwnProperty as just returning true if the object has such a property

[02:03:00.0000] <Hixie>
so we can define an interface attribute says specifies that the object has, in addition, properties for each of the things that namedgetter would return, or something

[02:03:01.0000] <Hixie>
or, alternatively:

[02:04:00.0000] <Hixie>
define a spec "hook" that i can refer to which automatically defines all of NameGetter, IndexGetter, etc

[02:04:01.0000] <heycam>
yeah i think a spec hook sounds better

[02:04:02.0000] <Hixie>
e.g. so that I can just write:

[02:05:00.0000] <Hixie>
"The BlaBla interface is a _Magical Land of Fairies_ in which the properties are the list of blablas, and for which the _Magical Fairy Setter Operation_ is foobarbaz."

[02:05:01.0000] <Hixie>
or whatever

[02:05:02.0000] <Hixie>
(not sure exactly what i would need to hook up each time)

[02:06:00.0000] <heycam>
i think you just need to define a set of (name, value) pairs, and call the hook with that

[02:06:01.0000] <Hixie>
or alternatively, have an attribute to enable this magic, and then say that the prose must define the _List Of Pixies_ and the _Pixie Getting_ and _Pixie Setting_ operations

[02:06:02.0000] <heycam>
yeah

[02:06:03.0000] <Hixie>
i think i need a list and an operation for setting, since that's usually non-trivial on my end

[02:06:04.0000] <Hixie>
and maybe deleting, if we decide we are ever going to support 'delete foo'

[02:07:00.0000] <heycam>
/me enqueues

[02:08:00.0000] <heycam>
will it be anything more than 'nodes in the document that match these criteria'?

[02:08:01.0000] <heycam>
anyway, bbl

[02:09:00.0000] <Hixie>
probably

[02:09:01.0000] <Hixie>
(e.g. localStorage)

[02:52:00.0000] <Hixie>
does anyone have an example of some mathematical sequence that would be useful to calculate, is simple to code, but which takes a disproportionally high amount of processor time to calculate?

[02:52:01.0000] <Philip`>
Prime numbers?

[02:52:02.0000] <Hixie>
and for which numbers can be obtained one after the other as computation continues?

[02:53:00.0000] <webben>
hsivonen: I think http://www.webbie.org.uk/ might be an example.

[02:53:01.0000] <Hixie>
prime numbers could work

[02:53:02.0000] <webben>
(it's a weird customization of IE used with Thunder: http://www.screenreader.net/ )

[02:53:03.0000] <webben>
hsivonen: (example of "contemporary UAs that support JS but not CSS" that is)

[02:54:00.0000] <Philip`>
http://www.research.att.com/~njas/sequences/ has a lot of sequences

[02:54:01.0000] <othermaciej>
Hixie: what counts as "disproportionately high"?

[02:54:02.0000] <webben>
hsivonen: also http://edbrowse.sourceforge.net/ though I'm not sure how actively developed that is.

[02:54:03.0000] <othermaciej>
the naiive way to code fibonacci is pretty slow

[02:54:04.0000] <Hixie>
othermaciej: like, something which would take a minute to give 10 numbers

[02:54:05.0000] <othermaciej>
ackerman is inherently slow

[02:55:00.0000] <Hixie>
i'm gonna try a very naive prime number search

[02:55:01.0000] <Hixie>
and see if that's slow enough for a worker demo

[02:55:02.0000] <othermaciej>
that won't take a minute to give 10 numbers

[02:55:03.0000] <Hixie>
it might if i start it high enough :-)

[02:55:04.0000] <othermaciej>
calling ack() with large-ish parameters would though

[02:55:05.0000] <Philip`>
There's no non-naive algorithm that will take a minute to give 10 numbers

[02:56:00.0000] <Philip`>
because you could rewrite it as "function f() { return [ 12376, 9162, 649, ... ] }"

[02:56:01.0000] <Philip`>
which is a less naive algorithm

[02:56:02.0000] <Philip`>
so I suppose you'll have to use some artificial pointless example instead

[02:57:00.0000] <othermaciej>
you can pass in a parameter asking what number

[02:57:01.0000] <othermaciej>
to ensure a lookup table would have to be impractically large

[02:58:00.0000] <othermaciej>
numerically estimating a definite integral might be slow enough

[02:58:01.0000] <othermaciej>
given the right integrand

[03:02:00.0000] <Hixie>
/me has just written the world's simplest yet the world's most advanced prime number search program ever

[03:02:01.0000] <Hixie>
so simple it's 13 lines of JS, so complex it won't work in any Web browser today!

[03:02:02.0000] <Hixie>
http://www.whatwg.org/demos/workers/compuation/

[03:03:00.0000] <Hixie>
renamed to http://www.whatwg.org/demos/workers/primes/

[03:06:00.0000] <hendry>
Hixie: doesn't work for me. first time i've seen syntax like    var worker = createWorker('worker.js');

[03:06:01.0000] <Philip`>
You should only loop up to i <= Math.sqrt(n) else it's unnecessarily inefficient

[03:07:00.0000] <Hixie>
Philip`: might that not distract from the point of the code?

[03:07:01.0000] <Hixie>
hendry: that's how advanced it is

[03:07:02.0000] <Hixie>
Philip`: (fixed)

[03:08:00.0000] <Philip`>
It would distract me less than the obvious easily-fixable inefficiency would :-)

[03:08:01.0000] <Hixie>
:-)

[03:11:00.0000] <Philip`>
Obviously the next step is to develop a multithreaded general number field sieve in JS

[03:12:00.0000] <Hixie>
if that would make a good example, please do so, i can fill in the worker-related bits

[03:12:01.0000] <mcarter>
Hixie, well, the chicken/egg question is answered -- real people are writing real code with calls to createWorker. so what are the browser vendors waiting for?

[03:12:02.0000] <jacobolus>
Philip`: because otherwise this is super-efficient code!

[03:12:03.0000] <Hixie>
mcarter: :-P

[03:13:00.0000] <jacobolus>
mcarter: you should tell them

[03:13:01.0000] <Philip`>
Hixie: I wasn't being serious :-p

[03:14:00.0000] <Hixie>
Philip`: aww :-(

[03:14:01.0000] <Philip`>
A C++ version at http://factor-by-gnfs.cvs.sourceforge.net/factor-by-gnfs/gnfs/ looks non-trivial

[03:14:02.0000] <mcarter>
jacobolus, if they don't know now, they should know on their own in the morning. At least some of them.

[03:14:03.0000] <iarwain>
Sorry for this off-topic question: What IRC logging software does this channel use? 

[03:14:04.0000] <jacobolus>
why in the morning?

[03:14:05.0000] <Hixie>
iarwain: krijnh rolledh is own, i believe

[03:15:00.0000] <iarwain>
Hixie: oh, ok. Thanks.

[03:30:00.0000] <Hixie>
http://www.whatwg.org/specs/web-workers/current-work/#tutorial

[03:30:01.0000] <Hixie>
first example

[03:30:02.0000] <Hixie>
is the style ok?

[03:32:00.0000] <Lachy>
/me is reading it now

[03:32:01.0000] <hsivonen>
webben: thanks. the first one of those two while a UA is not a browser, it seems

[03:33:00.0000] <webben>
hsivonen: Not sure quite what distinction you're drawing?

[03:33:01.0000] <webben>
what's the difference between using it and IE itself in that sense?

[03:35:00.0000] <hsivonen>
webben: oops. I skipped the word "browser" and saw "Accessible RSS news reader"

[03:36:00.0000] <hsivonen>
webben: it's for the blind, though, so it presumably is OK for it to key its presentation off ARIA

[03:37:00.0000] <webben>
hsivonen: It uses Trident so it's conceivably possible it would apply display:none; by not displaying that text; one would have to test that (or email the dev) to confirm I guess.

[03:37:01.0000] <Lachy>
Hixie, the port variable isn't mentioned anywhere else in the spec, AFAICS

[03:37:02.0000] <webben>
hsivonen: Yeah, I guess it will get ARIA with IE8.

[03:39:00.0000] <Hixie>
Lachy: fixed

[03:41:00.0000] <hsivonen>
webben: anyway, it seems that Hixie's "not accessible" scenario needs a visual UA that cannot be considered constituting AT in any way that doesn't support CSS and that supports scripting the DOM with JS

[03:42:00.0000] <Hixie>
e.g. firefox (with styles disabled)

[03:43:00.0000] <webben>
hmm ... "cannot be considered constituting AT" seems a rather high bar to set (especially since, as Hixie mentioned, rejecting publisher CSS can itself be an accessibility feature)

[03:44:00.0000] <webben>
I wonder if there's mobile browser that does that though.

[03:44:01.0000] <Hixie>
links also supports JS, as i understand it

[03:44:02.0000] <webben>
links doesn't (I /think/) ; but ELinks sort of does.

[03:45:00.0000] <hsivonen>
that must have been fun to hack in considering the level of documentation in the Links codebase

[03:45:01.0000] <webben>
but ELinks also supports CSS.

[03:45:02.0000] <Lachy>
Hixie, where is the MessagePort interface defined? Did it get renamed to something else?

[03:45:03.0000] <Hixie>
also, ARIA, if i'm not mistaken, is only supposed to be used to expose things to the system accessibility apis, it's not supposed to be natively interpreted by the UA. (but i'm not sure i really follow aria's goals and requirements)

[03:45:04.0000] <Hixie>
(so i could be wrong)

[03:45:05.0000] <Hixie>
Lachy: html5

[03:45:06.0000] <hsivonen>
is ELinks a rewrite or does it actually build on the original Links?

[03:46:00.0000] <hsivonen>
/me one considered hacking on Links as a weekend project and gave up after seeing the source

[03:46:01.0000] <webben>
Hixie: Not sure that's right actually.

[03:46:02.0000] <hsivonen>
s/one/once/

[03:46:03.0000] <webben>
hsivonen: dunno; ask in #elinks ;) ... I have a vague memory it's a partial rewrite.

[03:52:00.0000] <Lachy>
Hixie, why is the port property added seperately, instead of being defined in the interface and just set accordingly?

[03:53:00.0000] <Hixie>
because webidl doesn't yet have the concept of replaceable properties, and i want authors to be able to delete it or replace it or whatever, since it indirectly controls the lifetime of the worker

[03:53:01.0000] <Lachy>
ok

[04:07:00.0000] <Lachy>
hsivonen, given the new alt text requirements, how are you intending to adjust your validator? Will it now raise an error when alt is missing?

[04:27:00.0000] <hsivonen>
Lachy: I'm going to wait until accessibility advocates come back from vacation and the PFWG responds to the petitions of PFWG participants petitioning the PFWG as HTML WG participants and then Hixie responding to the response.

[04:31:00.0000] <Hixie>
can someone look over http://www.whatwg.org/demos/workers/stocks/ and tell me (a) what parts need explaining and (b) what parts have errors? The intro to this is at http://www.whatwg.org/specs/web-workers/current-work/#a-worker0

[04:31:01.0000] <gDashiva>
hsivonen: It's a bit disheartening when it sinks in that you're not even joking :)

[04:33:00.0000] <gDashiva>
Hixie: for..in on an array is kinda meh

[04:34:00.0000] <Hixie>
fixed

[04:34:01.0000] <gDashiva>
Not setting button.type that I can see

[04:35:00.0000] <gDashiva>
li.appendChild(a) should probably be (button)?

[04:35:01.0000] <Hixie>
doesn't it default to 'button'?

[04:35:02.0000] <Hixie>
oops, yes

[04:35:03.0000] <gDashiva>
Hixie: only in IE :)

[04:36:00.0000] <Hixie>
fixed and fixed

[04:37:00.0000] <Philip`>
I guess you might miss the first messages sent from the workers, since onmessage isn't set until a later script block after creating them

[04:37:01.0000] <Philip`>
which sounds suboptimal

[04:37:02.0000] <Hixie>
the messages are queued until you set onmessage (or call .start() on the port)

[04:38:00.0000] <Philip`>
Oh, okay

[04:39:00.0000] <Hixie>
though actually in this case it makes no difference since the workers don't send data until it is requested

[04:39:01.0000] <Hixie>
oh actually, no, you're right, it would make a difference

[04:39:02.0000] <Hixie>
since we postMessage() first

[04:39:03.0000] <Hixie>
nm

[04:39:04.0000] <Hixie>
but anyway, not a problem

[04:39:05.0000] <Hixie>
i'll comment on that in the prose

[04:40:00.0000] <gDashiva>
If stock.cgi is slow (or just slow net), you'll get time shift on the ticker timer

[04:40:01.0000] <Philip`>
http://www.whatwg.org/demos/workers/stocks/io.js tries to uses async XHR but returns the responseText without waiting for the response

[04:40:02.0000] <gDashiva>
/me shakes fist at Philip` 

[04:40:03.0000] <gDashiva>
I was gonna say that now :)

[04:40:04.0000] <Hixie>
time shift?

[04:40:05.0000] <Hixie>
oh, "true" is async? oops

[04:40:06.0000] <Hixie>
man who makes APIs with optional arguments that default to true

[04:40:07.0000] <gDashiva>
Hixie: You'll get updates at 10s, 20s+delay, 30s+2xdelay, etc

[04:41:00.0000] <Hixie>
gDashiva: that's ok, better than overloading the server :-)

[04:41:01.0000] <Hixie>
Philip`: fixed, thanks

[04:41:02.0000] <gDashiva>
kk

[04:42:00.0000] <Philip`>
http://www.whatwg.org/demos/workers/stocks/ticker.js puts a semicolon after the function statement, which is inconsistent with the rest of the code

[04:42:01.0000] <gDashiva>
No, it's an assignment

[04:42:02.0000] <Hixie>
oops fixed, thanks

[04:42:03.0000] <Hixie>
he means the first one

[04:42:04.0000] <gDashiva>
Oh, my bad

[04:44:00.0000] <gDashiva>
Now we just need a script to fake createWorker :)

[04:45:00.0000] <Philip`>
If there's a temporary network error then an exception will be thrown out of get(), and the ticker will permanently silently stop updating

[04:46:00.0000] <Hixie>
reload io.js

[04:46:01.0000] <Hixie>
does that work?

[04:47:00.0000] <Philip`>
I think so

[04:56:00.0000] <gDashiva>
500 when you access search.cgi without a query :)

[04:58:00.0000] <Philip`>
500 when you access it with a query, too

[04:58:01.0000] <Philip`>
Oh, no, it worked this time

[04:58:02.0000] <Hixie>
just finished writing them

[04:59:00.0000] <Philip`>
http://www.whatwg.org/demos/workers/stocks/search.cgi?%00 gives a peculiar error message

[04:59:01.0000] <Hixie>
heh

[05:00:00.0000] <Hixie>
appears to be an apache bug

[05:00:01.0000] <Hixie>
works ok at the command line

[05:00:02.0000] <zcorpan>
Hixie: isn't it try/catch, not try/except (in io.js)?

[05:00:03.0000] <Philip`>
Why does http://www.whatwg.org/demos/workers/stocks/search.cgi?GO return BGLH?

[05:01:00.0000] <Hixie>
because it matches "Goodger", which has stock ticker BGLH

[05:01:01.0000] <Hixie>
zcorpan: fixing

[05:01:02.0000] <Philip`>
Oh

[05:01:03.0000] <Hixie>
search with a blank string to see all the symbols it knows

[05:02:00.0000] <Hixie>
there's no way to determine what labels it is matching them against though

[05:02:01.0000] <Hixie>
short of a brute force attack

[05:02:02.0000] <hsivonen>
/me learns that the participant list of some W3C WGs is visible to Members only

[05:03:00.0000] <Philip`>
It seems kind of bad that http://www.whatwg.org/demos/workers/stocks/search.cgi?%47 doesn't work

[05:03:01.0000] <zcorpan>
http://www.whatwg.org/demos/workers/stocks/search.cgi?%00 gives 503

[05:04:00.0000] <Philip`>
zcorpan: Old :-p

[05:04:01.0000] <zcorpan>
oh

[05:04:02.0000] <Hixie>
Philip`: for %xx -- eh, whatever :-)

[05:11:00.0000] <hsivonen>
I wonder which ones are most common: unquoted, double-quoted or single-quoted attributes

[05:11:01.0000] <hsivonen>
my guess is double-quoted

[05:11:02.0000] <Philip`>
Double-quoted, then single-quoted, then unquoted

[05:12:00.0000] <hsivonen>
Philip`: thanks

[05:12:01.0000] <Philip`>
or maybe not

[05:12:02.0000] <Philip`>
s/maybe//

[05:14:00.0000] <Philip`>
http://canvex.lazyilluminati.com/misc/statestats.txt - state transitions from some collection of documents: BeforeAttributeValueState -> AttributeValueDoubleQuotedState: 1505620 BeforeAttributeValueState -> AttributeValueUnquotedState: 137469 BeforeAttributeValueState -> AttributeValueSingleQuotedState: 50487

[05:14:01.0000] <Philip`>
Argh, why does irssi forget to paste newlines?

[05:14:02.0000] <Philip`>
  BeforeAttributeValueState -> AttributeValueDoubleQuotedState: 1505620

[05:14:03.0000] <Philip`>
  BeforeAttributeValueState -> AttributeValueUnquotedState: 137469

[05:14:04.0000] <Philip`>
  BeforeAttributeValueState -> AttributeValueSingleQuotedState: 50487

[05:14:05.0000] <Philip`>
suggests double-quoted is most popular by far, and single-quoted is least

[05:15:00.0000] <hsivonen>
Philip`: thanks

[05:15:01.0000] <Philip`>
but I wouldn't necessarily recommend trusting me

[05:16:00.0000] <zcorpan>
Hixie: it would be nice with a close button on the spec updates box which makes the box disappear and not appear again until there's a later revision again

[05:16:01.0000] <Hixie>
remind me in a couple of days

[05:16:02.0000] <Hixie>
or just hit reload :-)

[05:17:00.0000] <zcorpan>
i've reloaded but it still says "You are reading r2025 but the latest revision is r2026"

[05:19:00.0000] <hsivonen>
It's fun how document published by Dublin Core Metadata Initiative start excessive visible metadata labeling

[05:20:00.0000] <hsivonen>
I would never have guessed that "Using Dublin Core" was the title if it hadn't "Title: " in front of it

[05:21:00.0000] <Hixie>
zcorpan: your browser is caching something

[05:23:00.0000] <Hixie>
ok, workers spec is stable for another 12 hours. bed time.

[05:30:00.0000] <zcorpan>
oh, 12h, maybe that'll be enough time to implement it

[05:32:00.0000] <hsivonen>
I wonder if HTML5 should define a mapping to Dublin Core from HTML5 and HTTP 1.1 native data

[06:08:00.0000] <Lachy>
hsivonen, why? Does anyone actually do anything useful with Dublin Core?

[06:09:00.0000] <hsivonen>
Lachy: DC is all the rage for archivists

[06:09:01.0000] <hsivonen>
Lachy: DC has useful parts like title

[06:10:00.0000] <hsivonen>
Lachy: DC also has utterly bogus parts like date

[06:10:01.0000] <hsivonen>
(date isn't a field. it's a datatype)

[06:11:00.0000] <hsivonen>
Lachy: (I'm not a big metadata believer. Participation in two goverment metadata projects made me a non-believer.)

[06:11:01.0000] <hsivonen>
government even

[06:12:00.0000] <hsivonen>
Lachy: we could define the obvious.

[06:12:01.0000] <Lachy>
I like metadata when it actually serves a useful purpose for users. Like embedding track title, artist and album cover art, etc. in music, videos.

[06:12:02.0000] <hsivonen>
like: <title> is the DC title

[06:12:03.0000] <hsivonen>
HTTP Content-Type is the DC format

[06:13:00.0000] <Lachy>
I just found most of the DC metadata was largely useless for documents on the web

[06:13:01.0000] <Lachy>
ok, mappings like those could work

[06:13:02.0000] <hsivonen>
Lachy: of course it is. there's always more value in writing software that looks at <title> than in writing software that look for DC.Title in meta

[07:01:00.0000] <gDashiva>
Anyone know the rationale for not allowing html entities in innerHTML on XHTML?

[07:02:00.0000] <annevk>
innerHTML simply uses an XML parser in that case without anything special

[07:03:00.0000] <gDashiva>
Yes, but why?

[07:04:00.0000] <annevk>
to not complicate things require support for things that are optional

[07:04:01.0000] <annevk>
or require*

[07:05:00.0000] <gDashiva>
Yet another roadblock to using xhtml

[07:07:00.0000] <Lachy>
gDashiva, how is missing entity refs a roadblock to using XHTML?

[07:08:00.0000] <Lachy>
they're not necessary. You can always use the numeric referece or JavaScript's \u#### syntax

[07:08:01.0000] <gDashiva>
Because innerHTML stops working where it used to work

[07:09:00.0000] <Lachy>
did any browser ever support entity refs in innerHTML for XHTML?

[07:09:01.0000] <gDashiva>
Opera does

[07:09:02.0000] <gDashiva>
I haven't tested safari

[07:09:03.0000] <Lachy>
even without the XHTML DOCTYPE in the document?

[07:10:00.0000] <gDashiva>
Yes

[07:11:00.0000] <gDashiva>
But if there's no doctype, it will error on entities in the initial markup

[07:13:00.0000] <takkaria>
why can't .innerHTML in XHTML use the html5 parser?

[07:14:00.0000] <hsivonen>
takkaria: Namespaces and tag inference

[07:15:00.0000] <annevk>
gDashiva, Opera doesn't use an XML parser for innerHTML afaict

[08:02:00.0000] <gsnedders>
/me is tempted to do some Pingback testing when he gets back home

[08:40:00.0000] <zcorpan>
http://simon.html5.org/test/html/dom/interfaces/HTMLElement/HTMLMediaElement/src/003.htm seems to freeze firefox

[08:42:00.0000] <Philip`>
Is it intentional that it claims to be testing "absent src=''" but actually has <source src="#">?

[08:46:00.0000] <mpilgrim>
Hixie: for class=note, class=issue screenreader problem -- try putting <span class="type">note</span> before each line, then hide it offscreen with span.type{display:block;position:absolute;top:-500px;width:1px;height:1px}

[08:47:00.0000] <mpilgrim>
jaws should read it but it won't show up in visual browsers

[08:47:01.0000] <mpilgrim>
adapted from http://code.google.com/p/doctype/wiki/ArticleAccessibility11

[08:48:00.0000] <mpilgrim>
hmm, that page needs a link back to webaim's skiplinks article

[08:51:00.0000] <Philip`>
http://krijnhoetmer.nl/irc-logs/whatwg/20080805#l-32

[08:52:00.0000] <mpilgrim>
well, great minds think alike

[08:52:01.0000] <zcorpan>
Philip`: yeah

[08:52:02.0000] <zcorpan>
Philip`: it doesn't have src on the <video>/<audio> element

[08:53:00.0000] <Philip`>
zcorpan: Oh, right

[09:09:00.0000] <csarven>
HTML5 appears to discontinue rel=section? In HTML4 @rel=section "Refers to a document serving as a section in a collection of documents." http://www.w3.org/TR/html401/types.html#type-links -- Would you say this is sort of talking about the common navigation items that lead to main sections of the site?

[09:10:00.0000] <jcranmer>
"JS and Flash must have the capacity to be slow... CSS... does not have to be slow"

[09:45:00.0000] <gsnedders>
Hixie: DuplicateTermException: dom-messageport-close (i.e., multiple dfns of it)

[09:52:00.0000] <annevk>
Hixie also made like 150 revisions

[10:01:00.0000] <gsnedders>
Hmm.

[10:01:01.0000] <gsnedders>
I still can't see why onreadystatechange isn't being called :\

[10:06:00.0000] <gsnedders>
/me plays around with null bytes in HTTP headers

[10:28:00.0000] <gsnedders>
readystatechange isn't called when async=false in Gecko, as far as I can see

[10:29:00.0000] <gsnedders>
annevk: you have any idea?

[10:29:01.0000] <gDashiva>
gsnedders: It's a rather meaningless venture

[10:29:02.0000] <gsnedders>
gDashiva: Like life itself.

[10:29:03.0000] <gDashiva>
By the time the event can be handled, the entire request is finished

[10:30:00.0000] <gsnedders>
gDashiva: It works in Saf and Op and IE!

[10:32:00.0000] <annevk>
it's a bug

[12:45:00.0000] <takkaria>
the guy who's posting to whatwg in an aggressive tone about javascript seem to be an <term ns="http://diveintomark.org/archives/2004/08/16/specs">asshole</term>

[12:45:01.0000] <takkaria>
judging by his website, anyway

[13:06:00.0000] <zcorpan>
hah, i was just thinking "hmm, nothing has happened on the whatwg blog for a while" and when i take a look, markp has posted a "This week in HTML 5" that hasn't yet reached my feed reader

[13:06:01.0000] <hsivonen>
now that Troy McLure is blogging about the new alt, I wonder if there will be comments

[13:09:00.0000] <Hixie>
hey, mpilgrim++

[13:09:01.0000] <Hixie>
i hope he does do that regularly!

[13:10:00.0000] <zcorpan>
hsivonen: pointer?

[13:10:01.0000] <annevk>
zcorpan, it's a reference to markp I believe

[13:11:00.0000] <zcorpan>
ah

[13:12:00.0000] <hsivonen>
http://blog.whatwg.org/omit-alt#comment-7771

[13:12:01.0000] <hsivonen>
(should be McClure)

[13:14:00.0000] <hsivonen>
I feel that Namespaces have hurt me personally

[13:15:00.0000] <hsivonen>
perhaps I should blog about what proportion of lines of code in my XML serializer is devoted to prefix allocation

[13:15:01.0000] <zcorpan>
can something obvious be done in xml5 to make it better?

[13:16:00.0000] <hsivonen>
zcorpan: not as far as I can tell

[13:20:00.0000] <hober>
how, err, eloquent: http://blog.whatwg.org/omit-alt#comment-8380

[13:22:00.0000] <Hixie>
i was going to edit it, but wordpress doesn't let me

[13:22:01.0000] <Hixie>
oh well

[13:27:00.0000] <Hixie>
ok i added a 'close' button, for whoever was asking for that

[13:28:00.0000] <annevk>
what did you want to do with the "fucker" comment?

[13:28:01.0000] <annevk>
/me can edit it

[13:28:02.0000] <Hixie>
nah it's ok

[13:29:00.0000] <Hixie>
i was going to change it to be very polite formal english and then leave a comment saying "edited for language"

[13:29:01.0000] <Hixie>
but it's a year old, who cares

[13:29:02.0000] <zcorpan>
Hixie: thanks!

[13:29:03.0000] <annevk>
fair enough

[13:29:04.0000] <Hixie>
for reference, how do you edit comments?

[13:32:00.0000] <zcorpan>
Hixie: though, it pops up again after 30s

[13:33:00.0000] <Hixie>
oh

[13:33:01.0000] <Hixie>
heh

[13:35:00.0000] <zcorpan>
/me wonders if the script could be made to work on the multipage version too, other than the first page

[13:36:00.0000] <Hixie>
man you people are never happy

[13:36:01.0000] <Hixie>
it probably could, speak to philip :-)

[13:37:00.0000] <zcorpan>
:)

[13:37:01.0000] <zcorpan>
Hixie: if you don't do anything, there's nothing to complain about :P

[13:38:00.0000] <zcorpan>
Hixie: when you do things there are always things that can be improved :)

[13:38:01.0000] <Hixie>
i'd complain if i did nothing :-P

[13:38:02.0000] <Hixie>
ok what simple demo can i do for shared workers

[13:38:03.0000] <Hixie>
i'm thinking maybe a party line demo

[13:41:00.0000] <annevk>
Hixie, log in, hit the edit button next to the comment

[13:41:01.0000] <gsnedders>
Hixie: DuplicateTermException: dom-messageport-close (i.e., multiple dfns of it)

[13:42:00.0000] <Hixie>
annevk: i didn't see an edit button

[13:44:00.0000] <annevk>
Hixie, next to the date?

[13:44:01.0000] <annevk>
(it's a link, not a button; sorry)

[13:44:02.0000] <Hixie>
oh yeah!

[13:48:00.0000] <zcorpan>
didn't the spec have "Big Issue:" markers before?

[13:48:01.0000] <zcorpan>
i seem to remember Lachy implemented them

[13:56:00.0000] <Hixie>
zcorpan: yes, went away when i wrote the annotation stuff

[13:57:00.0000] <Hixie>
sam just said it wasn't productive to address feedback on proposals

[13:57:01.0000] <Hixie>
that's... an interesting opinion

[13:58:00.0000] <hsivonen>
RDFa would be more importable to HTML5 if it used full URIs instead of CURIEs

[13:58:01.0000] <hsivonen>
that would remove the Namespace dependency and the qnames-in-content anti-pattern

[13:59:00.0000] <zcorpan>
hsivonen: i've been told that CURIEs don't depend on namespaces

[13:59:01.0000] <hsivonen>
zcorpan: how so?

[14:01:00.0000] <zcorpan>
wait let me dig it up

[14:09:00.0000] <zcorpan>
hmm can't find it in email, but i remember someone (rich?) saying that the xmlns mechanism is just one way to do CURIEs, you could equally well use any other mechanism like <meta> tags or other attributes

[14:10:00.0000] <zcorpan>
which left me even more confused

[14:10:01.0000] <hsivonen>
zcorpan: ah. no you need a mapping scope even if it isn't established by xmlns

[14:10:02.0000] <hsivonen>
zcorpan: it seems that you'd still get all the problems of mapping scopes

[14:11:00.0000] <zcorpan>
yes

[14:11:01.0000] <hsivonen>
Hixie: have you considered supporting RDFa with full URIs instead of CURIEs?

[14:13:00.0000] <zcorpan>
i don't understand what problem curies is trying to solve

[14:13:01.0000] <hsivonen>
zcorpan: they try the solve the problem of URIs being too long to be used as convenient identifiers

[14:14:00.0000] <hsivonen>
having the cake and eating it too

[14:14:01.0000] <zcorpan>
well apparently they weren't good enough for aria implementors :)

[14:15:00.0000] <zcorpan>
aria implementors wanted strings, not URIs

[14:15:01.0000] <Lachy>
Hixie, can you add one more link the spec update UI, pointing to: "http://html5.org/tools/web-apps-tracker?from=" + current_revision

[14:15:02.0000] <hsivonen>
I can't blame ARIA implementors for that

[14:15:03.0000] <Hixie>
hsivonen: to solve what problem?

[14:16:00.0000] <Hixie>
Lachy: :-P

[14:16:01.0000] <hsivonen>
Hixie: the problem of overlaying RDF-mappable metadata onto HTML with object literals as visible metadata without tantek's approval

[14:17:00.0000] <Hixie>
is that a rael problem?

[14:17:01.0000] <Hixie>
real, even

[14:17:02.0000] <hsivonen>
I don't know.

[14:18:00.0000] <Hixie>
/me moves on

[14:20:00.0000] <zcorpan>
i wonder if i should write a script that handles CURIEs correctly, just to see how ugly it gets

[14:21:00.0000] <hsivonen>
zcorpan: with which XML API?

[14:21:01.0000] <zcorpan>
hsivonen: no with javascript

[14:21:02.0000] <hsivonen>
zcorpan: I read DOM as the API :-)

[14:22:00.0000] <zcorpan>
fair enough :)

[14:22:01.0000] <hsivonen>
zcorpan: is it even *possible* to handle CURIEs correctly if the DOM tree has undergone arbitrary scripted mutations?

[14:22:02.0000] <hsivonen>
unless you do some bookkeeping before any mutations take place

[14:22:03.0000] <zcorpan>
well i didn't intend it to be performant

[14:23:00.0000] <hsivonen>
I mean bookkeeping like traversing the DOM tree and storing a snapshot of the namespace mapping scope locally at each element node before mutations

[14:25:00.0000] <Hixie>
Lachy: i added the link but it doesn't work, because the html5.org page is out of date (?)

[14:26:00.0000] <zcorpan>
Hixie: web-apps-tracker just tracks source, not whatwg-header

[14:26:01.0000] <Hixie>
oh, right, ok

[14:26:02.0000] <Hixie>
that would explain it

[14:26:03.0000] <Hixie>
so it'll work for the next revs :-)

[14:33:00.0000] <Lachy>
Hixie, thanks

[14:34:00.0000] <Lachy>
did someone else moderate the 2 comments in the blog's moderation queue already? I got mail about them being there, but they're gone now.

[14:37:00.0000] <hsivonen>
http://dubinko.info/blog/2008/07/28/erdf-11-proposal-discussion/

[14:37:01.0000] <hsivonen>
 A new profile string of:

[14:37:02.0000] <hsivonen>
"http://purl.org/NET/erdf11/profile"

[14:54:00.0000] <Lachy>
does eRDF prefixed class names with fixed prefixes, or with prefixes bound to a URI, like xmlns?

[14:55:00.0000] <hsivonen>
Lachy: bound to URI

[14:55:01.0000] <hsivonen>
in head

[14:56:00.0000] <Lachy>
ok. how unfortunate.

[14:57:00.0000] <Lachy>
it seems like using defined prefixes like foaf.depicts and cc.license would be sufficient

[14:57:01.0000] <hsivonen>
Lachy: how would that work for less known RDF vocabularies

[14:57:02.0000] <hsivonen>
?

[14:58:00.0000] <hsivonen>
assuming you want to be able to map to RDF

[14:58:01.0000] <jgraham>
But Lachy what it I wanted to use cc.licese to represent the license plate of my clasic car ?!!

[14:58:02.0000] <hober>
heh

[14:59:00.0000] <jgraham>
/me wonders who the end users consuming all this RDF are

[14:59:01.0000] <hsivonen>
how about eRDF5: like eRDF but with full URIs?

[15:00:00.0000] <Lachy>
hsivonen, jgraham, it would require communication with the community to settle on a prefix for a common vocabulary

[15:01:00.0000] <hsivonen>
jgraham: feed RDF into Naked Objects and you get generated UIs: http://www.nakedobjects.org/tutorial/index.shtml

[15:01:01.0000] <Lachy>
jgraham, the RDF model seems to be built mostly with publishing metadata in mind, rather than actually making use of what's published

[15:02:00.0000] <Lachy>
I'm yet to see any killer app for RDF metadata, unlike the many that exist for microformats

[15:03:00.0000] <hsivonen>
I haven't seen a killer app for microformats

[15:03:01.0000] <Lachy>
hsivonen, I've seen people extracting hcard information and storing it in a user's address book

[15:04:00.0000] <Lachy>
there was a tool somewhere that, given a URL to a page with hcard, would spit out a vcard file

[15:04:01.0000] <hsivonen>
Lachy: I'd call that an application but not a killer one

[15:05:00.0000] <Lachy>
there's also various calendar webapps.

[15:05:01.0000] <Lachy>
hsivonen, perhaps not, it's more than I've seen for RDF

[15:06:00.0000] <Lachy>
we might start seeing more soon. Firefox 3 has built in support for microformats, exposed through some extension api, so extension developers can make use of it

[15:16:00.0000] <Hixie>
/me agrees with hsivonen 

[15:17:00.0000] <Hixie>
the future of microformat type data as far as i can tell lies in heuristic analysers like those that detect addresses in e-mails in yahoo mail or google mail

[15:17:01.0000] <Hixie>
data that isn't explicitly structured, which the computer determines is of a particular type

[15:18:00.0000] <Hixie>
certainly so far we as a race have had better luck convincing computers to detect data than we have had convincing humans to mark it up

[15:19:00.0000] <roc>
depends on the data

[15:20:00.0000] <roc>
we haven't been successful at getting computers to extract programs from text, for example

[15:20:01.0000] <Hixie>
sure. i'm talking primarily about the kind of data a microformat or RDF would be used for.

[15:21:00.0000] <Hixie>
for things for which we typically use a whole format, e.g. documents, programs, images, etc, users seem willing to write data in a semi-structured way (usually using tools, though not always)

[15:27:00.0000] <othermaciej>
when it comes to user interface, even a fairly low rate of false-positive detection can be extremely irritating if it has a visible result

[15:27:01.0000] <Hixie>
indeed

[15:27:02.0000] <othermaciej>
thus, microformats are valuable as a higher-confidence indicator of particular structured data, perhaps trustworthy enough to decorate the UI

[15:28:00.0000] <othermaciej>
so far no one has turned this into a good UI

[15:28:01.0000] <Hixie>
so far it seems you get a better rate of detection if you rely on heuristics than if you rely on authors getting their semantics right

[15:28:02.0000] <othermaciej>
but it doesn't seem impossible

[15:28:03.0000] <othermaciej>
and many web sites have decent chunks of interesting microformat data

[15:28:04.0000] <othermaciej>
well, if microformats lead to a user-visible effect authors will be more likely to get them right

[15:28:05.0000] <hsivonen>
virtually all data detection in Mail.app is false positives

[15:29:00.0000] <Hixie>
othermaciej: <title> has a relatively high rate of bogus contents, to the point where google has to run heuristics on the contents of <title> to determine whether to show that or something else

[15:30:00.0000] <Hixie>
i guess maybe the right approach is more of a hybrid approach

[15:30:01.0000] <hsivonen>
Hixie: dc:title to rescue!

[15:30:02.0000] <othermaciej>
lol

[15:30:03.0000] <Hixie>
with the semantics being somewhat broad (e.g. <title>, as opposed to RDF triples) and the heuristics start from those hints and work from there

[15:30:04.0000] <Hixie>
s/start/starting/

[15:49:00.0000] <Lachy>
wow, the Aurora concept browser is horribly cluttered. http://labs.mozilla.com/2008/08/introducing-the-concept-series-call-for-participation/ (see the first video on that page)

[15:50:00.0000] <Lachy>
http://www.pcpro.co.uk/news/216897/mozilla-reveals-the-firefox-of-the-future.html

[15:51:00.0000] <Lachy>
it has some nice hypothetical features in it though, not sure how implementable they are though. But the UI is so bad, it makes it unusable even if they could be implemented

[15:53:00.0000] <Hixie>
the idea of concept browsers isn't the ui, it's the features :-)

[15:56:00.0000] <roc>
"Firefox of the future"? I don't think so :-)

[15:56:01.0000] <Lachy>
I realise, that, but they should try to focus on presenting the features, instead of destroying the whole idea with over engineered and flashy UI

[15:57:00.0000] <Hixie>
to be honest i had too much trouble with the "scenario"-type presentation to actually see the ui

[15:57:01.0000] <Hixie>
i don't understand why people always waffle so much instead of just saying their piece and shutting up

[15:57:02.0000] <Hixie>
(though i'm sure i waffle sometimes too)

[15:59:00.0000] <roc>
the fact is, bling sells

[16:00:00.0000] <Lachy>
the second video with the history search feature is ok. Being able to search and filter history by both keywords and approximately when I visited a page is something I've wanted for a long time

[16:01:00.0000] <Hixie>
i've been telling people that search browsers should have full-text-search through history of all pages ever viewed (and no expiring ever) for some time

[16:01:01.0000] <Hixie>
the number of times i want to search for something i saw recently but can't find any more is mindboggling

[16:02:00.0000] <Lachy>
sometimes, i visit a page, close it and then want to return to it a couple of hours later. But then just searching for keywords doesn't help when I have about 6 months of browsing history

[16:02:01.0000] <Lachy>
I set my history expiration to 999 days

[16:02:02.0000] <roc>
you need good heuristics for searching all that data though

[16:03:00.0000] <roc>
I hate to be parochial, but FF3's urlbar search is better than Opera's for that reason, even though it indexes less data

[16:04:00.0000] <Lachy>
my opera profile isn't kept around long enough for me to have a decent history, since I'm always testing with internal builds. So I can't really comment on the quality of Opera's history search

[16:05:00.0000] <Lachy>
I use Firefox for personal stuff and opera for testing, to keep things separate

[16:05:01.0000] <roc>
you can't use the same opera profile across different internal builds?

[16:06:00.0000] <aboodman>
hello playmobil

[16:06:01.0000] <roc>
My Opera upgrades seem to preserve my profile just fine

[16:06:02.0000] <Lachy>
I can, but internal builds have so many bugs, I like to keep a more stable profile for everyday use separate from the testing profile.

[16:06:03.0000] <playmobil>
Hey there aboodman!

[16:06:04.0000] <Lachy>
also, running 2 separate instances of Opera, even with separate profiles, is confusing

[16:06:05.0000] <Hixie>
/me commits another workers example

[16:07:00.0000] <Hixie>
the longest one so far

[16:07:01.0000] <Hixie>
can't wait for people to implement workers so i can test these things!

[16:07:02.0000] <aboodman>
that reminds me

[16:07:03.0000] <aboodman>
Hixie: I had an idea for the gc issue

[16:08:00.0000] <Lachy>
Hixie, is the google gears team intending to update that to use this API to provide support in existing browsers?

[16:08:01.0000] <Hixie>
a lot of the gc stuff got simplified away last night

[16:08:02.0000] <Hixie>
but go ahead

[16:08:03.0000] <aboodman>
I wanted to write it up but you had too many responses to m yfeedback and I got distracted

[16:08:04.0000] <Hixie>
Lachy: no idea, ask aboodman :-)

[16:08:05.0000] <aboodman>
Lachy: no idea, ask playmobil

[16:08:06.0000] <Lachy>
LOL

[16:09:00.0000] <aboodman>
Hixie: your latest idea I think still doesn't work consistently

[16:09:01.0000] <aboodman>
if gmail opens a named worker on instance 1, then instance 2

[16:09:02.0000] <aboodman>
then the user closes instance 1

[16:09:03.0000] <aboodman>
then the behavior of xhr is different than if he had closed instance 2 first

[16:09:04.0000] <aboodman>
i fully admit this is a crazy edge case, but it irritates me that its inconsistent

[16:10:00.0000] <Hixie>
only if gmail doesn't keep a reference to instance 2

[16:10:01.0000] <Hixie>
but yes

[16:10:02.0000] <Hixie>
i agree

[16:10:03.0000] <Hixie>
so

[16:10:04.0000] <aboodman>
what if instead, the list of things that i suggested keeps the worker live, but...

[16:10:05.0000] <Hixie>
i was thinking of not including the 'port' global for shared workers

[16:11:00.0000] <Hixie>
which would make it consistent, though bring back the odd behaviour for shared workers

[16:11:01.0000] <aboodman>
i don't like the port global anyway

[16:11:02.0000] <Hixie>
it makes the code so much easier, check out the examples

[16:11:03.0000] <Hixie>
but what was your idea?

[16:11:04.0000] <aboodman>
once all pages that ever had a MessagePort that refers to a worker unload, the worker is killed

[16:12:00.0000] <aboodman>
does that make sense at all?

[16:12:01.0000] <Hixie>
so basically make the lifetime be the maximum lifetime that it could have today, and prevent it from ever shutting down automatically before that?

[16:13:00.0000] <Hixie>
sure, i could buy that

[16:13:01.0000] <aboodman>
basically, workers are ref counted by the messages pending to them, and the messageports that are live, and the script running in them. this is what you have today.

[16:14:00.0000] <Lachy>
without the port global, what would be used instead of port.postMessage()?

[16:14:01.0000] <Hixie>
Lachy: see the example i just added

[16:14:02.0000] <aboodman>
additionally, things like httprequests that are pending can keep the worker alive past the point where it has no more references, but only if a page is alive that ever had a mesasgeport for that worker.

[16:14:03.0000] <Hixie>
oh, ok

[16:14:04.0000] <Hixie>
yeah i could buy that

[16:15:00.0000] <Hixie>
shall i go ahead and make those changes?

[16:15:01.0000] <aboodman>
i think it is better than what is there now

[16:16:00.0000] <Hixie>
agreed

[16:16:01.0000] <Lachy>
oh, ok. So listening for onconnect and getting a reference to the port from there.

[16:19:00.0000] <aboodman>
Hixie: you say that MessagePort implements EventTarget, but I don't see it in the interface description

[16:19:01.0000] <aboodman>
is there some convention that I am missing?

[16:20:00.0000] <Hixie>
search for "eventtarget"

[16:20:01.0000] <Hixie>
oops, i made the wrong object implement it

[16:20:02.0000] <Hixie>
my bad

[16:20:03.0000] <Hixie>
let me fix that

[16:20:04.0000] <Hixie>
as soon as i've done this lifetime change

[16:24:00.0000] <aboodman>
and as long as I have you

[16:24:01.0000] <aboodman>
utils--

[16:24:02.0000] <aboodman>
i will send mail with all this too

[16:25:00.0000] <Hixie>
utils was sicking's idea, basically he said we shouldn'e be poluting the global scope with new features over time as it will cause clashes

[16:25:01.0000] <Hixie>
(a lesson to learn from Window)

[16:25:02.0000] <Hixie>
hey so what happens if you have a Window

[16:25:03.0000] <Hixie>
and it creates a worker

[16:26:00.0000] <Hixie>
and that worker creates a worker

[16:26:01.0000] <Hixie>
and the nested worker then starts an XHR

[16:26:02.0000] <Hixie>
and everyone drops their references so all the ports get GC'ed

[16:28:00.0000] <aboodman>
so first of all, I think workers count as 'asynchronous things'

[16:28:01.0000] <aboodman>
(we need a better word)

[16:28:02.0000] <aboodman>
so a nested worker could keep its parent alive same as an xhr

[16:29:00.0000] <Hixie>
so if a worker creates a worker, those two workers can never die until the window is closed?

[16:29:01.0000] <Hixie>
that seems bad

[16:29:02.0000] <aboodman>
no they can die

[16:29:03.0000] <aboodman>
if they aren't doing anything and no ports or mesages reference them

[16:29:04.0000] <Hixie>
this is going to be the most complicated set of conditions i have ever written

[16:29:05.0000] <Hixie>
/me flexes his wrists

[16:29:06.0000] <aboodman>
agree

[16:30:00.0000] <aboodman>
stop me if it won't work, it's just an idea

[16:31:00.0000] <aboodman>
i haven't thought through nested workers

[16:31:01.0000] <aboodman>
nested workers are another thing i could live without

[16:31:02.0000] <Hixie>
i see no reason why it shouldn't work, it'll just be a bitch to write in english

[16:31:03.0000] <Hixie>
but that's what i do

[16:31:04.0000] <Hixie>
:-)

[16:31:05.0000] <roc>
I haven't really been paying attention, but how about saying that closing a window stops all workers it owns; any other early termination is just an optimization only

[16:31:06.0000] <roc>
*and* provide a method for transferring ownership explicitly

[16:31:07.0000] <Hixie>
roc: workers can be shared between windows, like XHR objects and so on

[16:31:08.0000] <Hixie>
oh

[16:31:09.0000] <Hixie>
hm

[16:32:00.0000] <Hixie>
that seems like it's pushing the complexity on the author

[16:32:01.0000] <roc>
complicated constraints are also complexity for the author

[16:32:02.0000] <Hixie>
not necessarily

[16:35:00.0000] <roc>
your call. Just thought I'd mention it as an option

[16:35:01.0000] <aboodman>
ideally, the issue of when workers shut down doesn't come up

[16:35:02.0000] <aboodman>
they just sort of do the right thing

[16:35:03.0000] <aboodman>
from an author's point of view

[16:36:00.0000] <aboodman>
and all the complexity is pushed onto UA developers

[16:38:00.0000] <roc>
of course, that's always the goal for every feature

[16:38:01.0000] <roc>
another tradeoff to keep in mind is that you don't want to create mysterious worker leaks

[16:42:00.0000] <aboodman>
yep

[16:42:01.0000] <aboodman>
i think my proposal achieves both of these (no mysterious leaks and 'does the right thing')

[16:42:02.0000] <aboodman>
complexity is another issue

[16:45:00.0000] <Hixie>
aboodman: ok http://www.whatwg.org/specs/web-workers/current-work/#the-workers

[16:45:01.0000] <Hixie>
text with a green underline is a reference to the main html5 spec

[16:46:00.0000] <Hixie>
i haven't set up cross-references cross-spec yet

[16:46:01.0000] <Hixie>
i got rid of the "front line" stuff too

[16:48:00.0000] <Hixie>
aboodman: fixed the eventtarget thing

[16:48:01.0000] <aboodman>
looking

[16:49:00.0000] <aboodman>
i just sanity tested the lifetime idea with michaeln

[16:49:01.0000] <aboodman>
he said it seemed reasonable

[16:49:02.0000] <aboodman>
i am not sure about nested worker

[16:49:03.0000] <aboodman>
do we really need workers creating workers?

[16:49:04.0000] <Hixie>
sure, why not?

[16:49:05.0000] <Hixie>
seems a bit weird to disallow it

[16:50:00.0000] <Hixie>
especially since you can always just create a worker and pass the channel port to the worker

[16:50:01.0000] <aboodman>
http://www.whatwg.org/specs/web-apps/current-work/multipage/comms.html#messageport0

[16:50:02.0000] <Hixie>
so it's not like you'd remove complexity by not allowing it

[16:50:03.0000] <aboodman>
i still don't see it

[16:50:04.0000] <Hixie>
oh MessagePort!

[16:50:05.0000] <Hixie>
i thought you meant WorkerGlobalScope

[16:50:06.0000] <aboodman>
same question for both

[16:52:00.0000] <Hixie>
hm yes, MessagePort isn't right

[16:52:01.0000] <Hixie>
/me fixes

[16:53:00.0000] <Hixie>
fixed, see just below the idl

[16:53:01.0000] <aboodman>
what should i look for in the worker document about lifetime

[16:53:02.0000] <Hixie>
http://www.whatwg.org/specs/web-workers/current-work/#the-workers

[16:54:00.0000] <Hixie>
section "2.4 The worker's ports

[16:54:01.0000] <Hixie>
"

[16:54:02.0000] <Hixie>
and then search for "Closing orphan workers" which uses those erms

[16:54:03.0000] <Hixie>
terms

[16:55:00.0000] <aboodman>
sorry , but I still see nothing about 'EventTarget' near 'interface MessagePort'

[16:56:00.0000] <aboodman>
dang, now i see it

[16:56:01.0000] <aboodman>
so, why don't these IDL files use " : EventTarget" ?

[16:56:02.0000] <aboodman>
that would be more clear

[16:56:03.0000] <Hixie>
they don't inherit from EventTarget

[16:57:00.0000] <aboodman>
err...

[16:57:01.0000] <Hixie>
it's just that that one object implements several interfaces

[16:57:02.0000] <aboodman>
why not say that MessagePort inherits EventTarget

[16:58:00.0000] <Hixie>
why would you say that? what about all the other interfaces that it might implement one day?

[16:58:01.0000] <Hixie>
e.g. the way Document implements Document, HTMLDocument, SVGDocument, EventTarget, etc

[16:58:02.0000] <Hixie>
Document inherits from Node

[16:58:03.0000] <Hixie>
because a Document is-a Node

[16:59:00.0000] <aboodman>
i see your question

[16:59:01.0000] <aboodman>
in many languages there is a way to say "implements"

[16:59:02.0000] <aboodman>
i'm requesting this for WebIDL

[16:59:03.0000] <Hixie>
that might be useful, yes

[16:59:04.0000] <Hixie>
we'd also need the opposite, "is implemented by"

[16:59:05.0000] <Hixie>
i requested both of these some time ago i believe :-)


2008-08-07
[17:05:00.0000] <Hixie>
(heycam: ^)

[17:05:01.0000] <Hixie>
bbiab

[17:08:00.0000] <aboodman>
hi michaeln 

[17:11:00.0000] <michaeln>
hello aboodman

[17:14:00.0000] <michaeln>
hello again aa

[17:44:00.0000] <Hixie>
aboodman: i can show that if we have message ports, it's possible to end up with a worker that's a nested worker even if we disallow worker creation from inside workers, so even if we remove the api we can't remove the complexity in the lifetime definitions

[17:45:00.0000] <Hixie>
(i just made a slideshow to prove it to myself)

[17:45:01.0000] <aboodman>
that makes sense

[17:45:02.0000] <aboodman>
the way you define workers, the whole idea of "nested" workers is nonsensical

[17:45:03.0000] <aboodman>
they are more of a graph

[17:46:00.0000] <aboodman>
the page that created a worker is not special

[17:48:00.0000] <Hixie>
yeah

[18:01:00.0000] <aboodman>
one way to think of it that is not that complicated...

[18:02:00.0000] <aboodman>
an httprequest can stay alive past when it has no references, in order to complete

[18:02:01.0000] <aboodman>
however those all get killed when a page unloads, no matter what

[18:02:02.0000] <Hixie>
yup

[18:02:03.0000] <Hixie>
that's basically what it says now

[18:03:00.0000] <aboodman>
so it makes sense that an httprequest can keep a worker alive in the same way

[18:03:01.0000] <aboodman>
and a page can kill the worker (and the httprequest) in the same way when it unloads

[18:06:00.0000] <Dashiva>
/me worries about the miconceptions documented thread

[18:07:00.0000] <Dashiva>
It seems like everyone agrees, yet there's so much back-and-forth. Oh, process.

[18:07:01.0000] <Hixie>
i'm mostly waiting on heycam to update his spec

[18:51:00.0000] <roc>
hmmm

[18:51:01.0000] <aboodman>
do we have a name for all the apis you can reach from JS in a browser?

[18:51:02.0000] <roc>
"On setting, the following algorithm must be run:

[18:51:03.0000] <roc>
   1.      If the root element is an svg element in the "http://www.w3.org/2000/svg" namespace, and the user agent supports SVG, then the setter must defer to the setter for the DOM attribute of the same name on the SVGDocument interface."

[18:51:04.0000] <roc>
but 'title' is readonly in SVGDocument

[18:51:05.0000] <aboodman>
DOM is not technically correct, "the web platform" sounds very pretentious, and might be misinterpreted....

[19:04:00.0000] <Hixie>
aboodman: i call them DOM APIs

[19:05:00.0000] <Hixie>
roc: then it would raise an exception i guess :-)

[19:05:01.0000] <aboodman>
Hixie: we need a better phrase, since everyone is going to say 'but they aren't a part of DOM'

[19:18:00.0000] <Hixie>
*shrug*

[19:18:01.0000] <Hixie>
if that's what people are complaining about, i'm happy

[19:19:00.0000] <Hixie>
sicking was all on me yesterday telling me that fixing the spec as the highest priority ever, but i haven't heard back from him about the chaned

[19:19:01.0000] <Hixie>
changes

[19:21:00.0000] <Hixie>
aboodman: btw one problem with making the offline cache an xml file is that it makes processing it a lot harder -- you can't just walk through the file one line at a time, now you have to check its conformance first and then you have to check namespaces and tag nams everywhere and so on

[19:21:01.0000] <Hixie>
names

[19:23:00.0000] <aboodman>
Hixie: I understand it pulls in a lot of dependencies, but from this developer's point of view, it seems conceptually simpler

[19:23:01.0000] <aboodman>
because XML is something that is already known that is being used to build something new

[19:23:02.0000] <Hixie>
it'd be so much buggier

[19:23:03.0000] <aboodman>
whereas this text thing is all new

[19:24:00.0000] <aboodman>
do linebreaks matter?

[19:24:01.0000] <aboodman>
what about two line breaks?

[19:24:02.0000] <aboodman>
what about \r

[19:24:03.0000] <aboodman>
what about tabs vs spaces

[19:24:04.0000] <aboodman>
etc etc

[19:24:05.0000] <aboodman>
all these have expected answers for xml

[19:24:06.0000] <aboodman>
(again, my opinion, and I do understand that xml has tons of baggage)

[19:24:07.0000] <Hixie>
xml has the same problem, except now you need to ask what tag name to use, what elements are allowed where and in what order, what attributes are allowed

[19:25:00.0000] <aboodman>
i hate arguing about this kind of thing because you can go round and round forever

[19:25:01.0000] <Hixie>
not to mention other questions they won't think to ask like what elements need to be in what namespaces, what happens if there are unexpected elements, can PIs be used, what encodings are allowed

[19:25:02.0000] <Hixie>
seems like moving to xml doesn't reduce the number of questions, it just makes them harder to understand

[19:26:00.0000] <aboodman>
i really hesitate to wade into this pool, but...

[19:26:01.0000] <aboodman>
don't schemas sort of handle this?

[19:27:00.0000] <Hixie>
schemas just add even more questions

[19:27:01.0000] <aboodman>
UAs don't necessarily have to process the schema, but it would make defining the expected tags more formal

[19:27:02.0000] <Hixie>
defining the structure isn't a problem

[19:27:03.0000] <Hixie>
i mean, the current structure is defined

[19:27:04.0000] <Hixie>
and it's not even xml

[19:28:00.0000] <aboodman>
i understand, and yet, again, as a developer, i'm telling you that the text format makes me queazy

[19:28:01.0000] <aboodman>
dunno what to tell you

[19:29:00.0000] <Hixie>
(the current definition is about two screen lengths, doing it in xml would double or triple that, i'd imagine)

[19:29:01.0000] <aboodman>
i can't believe that

[19:29:02.0000] <aboodman>
wouldn't it just be:

[19:29:03.0000] <aboodman>
the manifest is an XML file validating against <this schema>

[19:30:00.0000] <Hixie>
authors don't understand schemas, the schema would be many screens long, and the schema languages that exist can't define everything anyway

[19:31:00.0000] <michaeln1>
can you make the text file extensible?

[19:31:01.0000] <aboodman>
the authors don't necessarily have to understand the schema, the spec isn't primarily for authors anyway

[19:31:02.0000] <Hixie>
(for example, how do you, in a schema, say that the "Opportunistic caching namespaces must have the same origin as the manifest itself." or whatever?)

[19:31:03.0000] <Hixie>
michaeln1: the text file is extensible

[19:31:04.0000] <aboodman>
for authors, there could just be a sample file

[19:31:05.0000] <aboodman>
<shrug>

[19:31:06.0000] <roc>
if browsers don't validate against the schema, it's worthless

[19:31:07.0000] <aboodman>
sorry

[19:31:08.0000] <aboodman>
<shrug/>

[19:32:00.0000] <Hixie>
aboodman: i can provide a sample file for this too, if you want?

[19:32:01.0000] <Hixie>
s/?//

[19:32:02.0000] <roc>
but if they have to, icky ick

[19:32:03.0000] <aboodman>
that would definitely be helpful

[19:32:04.0000] <aboodman>
(the sample)

[19:32:05.0000] <Hixie>
will add that to my TODO

[19:32:06.0000] <aboodman>
but my natural reaction to text formats is always going to be:

[19:32:07.0000] <Hixie>
(once i'm done with workers)

[19:32:08.0000] <aboodman>
can i put comments in?

[19:32:09.0000] <aboodman>
can there be multiple blank lines?

[19:33:00.0000] <aboodman>
do carriage returns matter?

[19:33:01.0000] <aboodman>
etc

[19:33:02.0000] <aboodman>
where i feel like they are known answers with xml

[19:33:03.0000] <aboodman>
i will leave it at that, because i don't want to argue about it anymore

[19:33:04.0000] <Hixie>
sure but once you've done enough xml you'll find you have similar questions about xml vocabularies :-)

[19:33:05.0000] <Hixie>
not comments, but e.g., "what happens if this element contains text?"

[19:33:06.0000] <Hixie>
anyway

[19:33:07.0000] <Hixie>
i'll add an example

[19:34:00.0000] <Hixie>
that covers your questions

[19:34:01.0000] <michaeln1>
at least then your discussing the semantics rather than the line discipline

[19:34:02.0000] <Hixie>
michaeln1: ok, "do spaces matter at the start of this attribute?"

[19:36:00.0000] <michaeln1>
point taken

[19:37:00.0000] <aboodman>
Hixie: fair enough

[19:38:00.0000] <michaeln1>
how can i add an experimental category in the current scheme and retain compatibility with impls that don't know about my experimental category?

[19:39:00.0000] <Hixie>
you can't

[19:40:00.0000] <Hixie>
at least not with urls being on their own one per line like the other sections

[19:40:01.0000] <Hixie>
i can change the parser model to allow it if you want

[19:40:02.0000] <michaeln1>
same question for adding experimental attributes to entry lines?

[19:41:00.0000] <michaeln1>
could be done with creative use of comments... but ugly

[19:42:00.0000] <Hixie>
add the attribute on the line before the url, with a syntax that isn't a valid url, e.g. ":attr:"

[19:43:00.0000] <michaeln1>
xml is cleaner on this dimension

[19:43:01.0000] <Hixie>
cleaner is a matter of opinion

[19:43:02.0000] <Hixie>
it certainly has many many more axes to screw things up in :-)

[19:44:00.0000] <Hixie>
in practice, people implementing xml-based processors tend to ignore what the specs require, and just do what looks right, and you find that your extensibility model isn't nearly as extensible as it looks, because all the UAs fail in different ways when hitting the extensions

[20:06:00.0000] <aboodman>
Hixie: Are MessagePort's 'load' and 'error' events used for anything other than workers loading or failing to load

[20:06:01.0000] <Hixie>
no

[20:06:02.0000] <Hixie>
i believe the spec says so somewhere

[20:07:00.0000] <Hixie>
search for "nothing in this" in html5

[20:11:00.0000] <aboodman>
ok new feedback on the way

[20:11:01.0000] <Hixie>
cool

[20:11:02.0000] <aboodman>
it might be worthwhile for you and I to get together and hash out anything remaining

[20:11:03.0000] <aboodman>
at least so we are on the same page

[20:12:00.0000] <Hixie>
sure

[20:12:01.0000] <Hixie>
i think we're much closer to each other than we are to sicking, which is more my concern to be honest

[20:14:00.0000] <Hixie>
so i don't really understand the proposals you're making re worker objects, etc

[20:15:00.0000] <aboodman>
right, and it take soooo long to type

[20:15:01.0000] <Hixie>
is there any way you could rewrite the examples in the intro section to use what you're proposing?

[20:15:02.0000] <aboodman>
that is why I like the idea of dinner

[20:15:03.0000] <Hixie>
hmm, dinner

[20:23:00.0000] <Hixie>
i just wrote the worlds most complicated implementation of a numeric constant. but anyway.

[20:27:00.0000] <Hixie>
http://www.whatwg.org/specs/web-workers/current-work/#delegation

[20:36:00.0000] <Lachy>
Hixie, does that just return the value 10000000 when it's all complete?

[21:50:00.0000] <Hixie>
Lachy: yeah

[22:05:00.0000] <Hixie>
roc: re the mutation thing, i think the answer is that the mutation events spec is being updated to say that such events would be delayed past the end of the algorithm

[22:05:01.0000] <roc>
hmm

[22:05:02.0000] <roc>
who's doing that?

[22:05:03.0000] <Hixie>
shepazu, i believe

[22:06:00.0000] <Hixie>
roc: (also, if you want these issues dealt with sooner rather than later, let me know and i'll prioritise them)

[22:06:01.0000] <roc>
there's some sort of generic magic that says spec-algorithms happen atomically?

[22:06:02.0000] <Hixie>
probably not, i might have to add some text to the html5 spec once the dom events spec is updated to give me a hook to hold onto

[22:07:00.0000] <roc>
exciting

[22:07:01.0000] <Hixie>
i haven't really looked at how mutation events affect html5

[22:07:02.0000] <roc>
even more exciting

[22:09:00.0000] <roc>
it's not super urgent but implementors will have to make decisions about these things ... indeed, already have

[22:09:01.0000] <Hixie>
same applies to most of the feedback the spec has :-)

[22:10:00.0000] <roc>
it's reasonable to hope that we can change mutation event behaviour later without breaking things

[22:11:00.0000] <Hixie>
yeah

[02:24:00.0000] <Philip`>
"With multicore CPUs becoming prevalent, ..." - that's not going to make much sense towards the end of the expected lifespan of HTML5

[03:33:00.0000] <Philip`>
http://wiki.whatwg.org/wiki/User:SmithP and http://wiki.whatwg.org/wiki/User_talk:SmithP look like spam

[03:33:01.0000] <Philip`>
as does http://wiki.whatwg.org/wiki/User_talk:Flower

[07:36:00.0000] <annevk>
so with seamless <iframe> stuff does html:root match only the outermost <html> or also the one inside the <iframe>?

[07:36:01.0000] <annevk>
it would be nice if it only matched the outermost <html>

[07:37:00.0000] <annevk>
at least for the purpose of style sheets that would "go in" the <iframe>

[07:54:00.0000] <zcorpan>
annevk: it would match both

[07:54:01.0000] <jgraham>
FWIW the people on reddit have a reasonable point

[07:58:00.0000] <Lachy>
the comments on reddit are getting more and more ridiculous. AFACT, complaints seem to be based on the use of curly braces being somewhat ugly.

[08:00:00.0000] <hsivonen>
Lachy: they *are* ugly in UAs that don't hide them

[08:01:00.0000] <Philip`>
It's not the braces that are ugly, it's the embedding of one syntax into another syntax

[08:01:01.0000] <Philip`>
(as I understand it)

[08:01:02.0000] <Philip`>
particularly since that means tools that work on the outer syntax (e.g. CSS selectors) won't understand the new inner syntax

[08:06:00.0000] <zcorpan>
img[alt^="{"][alt$="}"] { ... }

[08:06:01.0000] <Lachy>
hsivonen, yes, I know. I didn't say they weren't :-)

[08:07:00.0000] <Philip`>
and particularly since the normal alt syntax (freeform text) and the new improper-alt alt syntax (freeform text surrounded by braces) overlap, which means if e.g. you have a UI for entering image alt text, and then someone types in "{Google Inc}", you don't really know what to do with it

[08:08:00.0000] <Lachy>
Philip`, the use of {...} in existing alt attributes turned out to be very low

[08:09:00.0000] <zcorpan>
i guess you could insert a zero-width space before or after

[08:09:01.0000] <Lachy>
so the chances that someone would want the alt to be {Google Inc} and want it to be legitimate alt text is slim.

[08:09:02.0000] <hsivonen>
Lachy: it's still something people should program an if case for if they are thorough

[08:09:03.0000] <Philip`>
Lachy: Conscientious developers will want to handle whatever input they're given in a sensible way, and not just assume that it doesn't matter if things get messed up in a small fraction of cases

[08:12:00.0000] <Lachy>
I don't understand. Why would authoring tools want to handle {...} differently from others anyway, other than automatically generating one like that if the author provides nothing.

[08:12:01.0000] <jgraham>
Lachy: Authoring tools should flag alt={...} as an error

[08:13:00.0000] <Lachy>
hmm, maybe a warning

[08:13:01.0000] <jgraham>
In something like Dreamweaver it is an error surely

[08:14:00.0000] <Lachy>
not necessarily. People use dreamweaver to write templates, which are later incoporated into other backend systems all the time. So the image they use in dreamweaver may just be a placeholder for user generated content

[08:14:01.0000] <Philip`>
If you're writing a LaTeX-to-HTML converter that converts equations into images and puts the LaTeX source in the alt attribute, then it's quite possible that the equation would be surrounded by {...}, and you wouldn't want that to be misinterpreted as meaning the alt text wasn't equivalent to the image

[08:15:00.0000] <Lachy>
would latex markup really be an appropriate alternate text?

[08:15:01.0000] <jgraham>
Lachy: That's another reason to dislike {} because AIUI that is quite a common syntax for templating

[08:15:02.0000] <jgraham>
Lachy: Yes

[08:15:03.0000] <Lachy>
why?

[08:16:00.0000] <Lachy>
we have mathml now

[08:16:01.0000] <jgraham>
Because if you have an image of an equation then many people will either be able to read the LaTeX directly or find a tool to render it

[08:16:02.0000] <Philip`>
We can't use MathML now

[08:17:00.0000] <jgraham>
(I agreew MathML is the "right" solution though)

[08:17:01.0000] <Philip`>
and even if we could, it's more reliable to use a LaTeX-to-PNG renderer than to use a LaTeX-to-MathML converter and then hope the user has the right browser and fonts

[08:17:02.0000] <Lachy>
templating languages like JSP use ${foo}

[08:18:00.0000] <Philip`>
so it'll still be an issue for the next decade or so, and in that period we want the equation-images to have the best possible alt text, which in this case is the LaTeX source because the converter doesn't have access to anything better

[08:18:01.0000] <Lachy>
what does wikipedia use as alt text when it renders math as images?

[08:19:00.0000] <Philip`>
<img class="tex" alt="\left(\beta mc^2 + \sum_{k = 1}^3 \alpha_k p_k \, c\right) \psi (\mathbf{x},t) = i \hbar \frac{\partial\psi}{\partial t}(\mathbf{x},t) " src="http://upload.wikimedia.org/math/b/9/0/b90f28f1bb825692930ce71234d02a84.png" />

[08:19:01.0000] <jgraham>
Philip` is too fast for me :)

[08:19:02.0000] <Lachy>
so that doesn't surround the value with {...}

[08:20:00.0000] <Lachy>
are there any cases where it would?

[08:20:01.0000] <jgraham>
Lachy: I think so

[08:20:02.0000] <jgraham>
(certianly you could)

[08:21:00.0000] <Lachy>
well, are there any more appropriate alternative characters that could be used?

[08:21:01.0000] <jcranmer>
I must ask, what equation is that for?

[08:22:00.0000] <jgraham>
jcranmer: It's the schrodinger equation isn't it? 

[08:22:01.0000] <jgraham>
/me is just reading the LaTeX

[08:22:02.0000] <jgraham>
(or something like that)

[08:22:03.0000] <Philip`>
<math>{x \over y}</math> gives <img class="tex" alt="{x \over y}" src="http://upload.wikimedia.org/math/9/0/8/908124d354754d7eb6ee962b21e1b288.png" />

[08:22:04.0000] <Philip`>
and that's a fairly plausible input

[08:22:05.0000] <jcranmer>
not in a format that I recognize quickly

[08:22:06.0000] <Philip`>
I don't think the braces are ever necessary, but there's usually no reason to not use them

[08:23:00.0000] <Philip`>
It's The Dirac equation in the form originally proposed by Dirac, obviously :-p

[08:23:01.0000] <jgraham>
Philip`: I wonder if they would be necessary with tensors where you can have raised or lowered indicies out the front

[08:24:00.0000] <Philip`>
The Dirac equation is superficially similar to the Schrödinger equation for a free particle:, so that's an easy mistake to make

[08:24:01.0000] <jcranmer>
/me points out that he hasn't passed diffeq

[08:24:02.0000] <jcranmer>
and we didn't even get to PDEs...

[08:25:00.0000] <jgraham>
Well I noticed the mc^2 term, thought oh relatvistic QM, but didn't join the dots...

[08:26:00.0000] <Philip`>
http://en.wikipedia.org/wiki/Special_relativity

[08:26:01.0000] <Philip`>
Oops

[08:26:02.0000] <Philip`>
That doesn't end in }

[08:26:03.0000] <Philip`>
but it starts in one, so that's halfway there

[08:27:00.0000] <Philip`>
If you had "{x^y}^z = x^{y^z}" then that's of the form "{...}"

[08:27:01.0000] <Philip`>
http://fr.wikipedia.org/wiki/Carr%C3%A9_magique_%28math%C3%A9matiques%29

[08:27:02.0000] <Philip`>
{n(n^2+1) \over 2}

[08:27:03.0000] <Philip`>
Aha, that's one

[08:30:00.0000] <Lachy>
ok. So are false positives like that really going to cause any major problems? Since the UA is required to expose the value anyway, and arguably, latex markup isn't really a great alternative text

[08:31:00.0000] <hsivonen>
Lachy: it seems that it's the best format for blind readers that you can stick into an attribute

[08:31:01.0000] <hsivonen>
(for blind readers who do math)

[08:32:00.0000] <Philip`>
They are going to cause problems for developers who don't want to cause minor problems, and therefore have to implement some nasty hack (like adding a space to the value) to make sure their output is not conflicting with HTML5's special markers, and who lose trust in the development of HTML since it's introducing new unobvious issues like this that never existed

[08:33:00.0000] <Philip`>
(and so they can't tell whether HTML6 is going to introduce more special markers that break even more of their tools)

[08:35:00.0000] <hsivonen>
I don't like {}, but I'm too tired to bikeshed it before the accessibility experts do

[08:36:00.0000] <jgraham>
I think Philip` is right. As far as I can tell adding random microsyntaxes to existing attributes has demonstratable problems whereas adding a new attribute only has hypothetical problems

[08:37:00.0000] <annevk>
I don't really like {} either

[08:40:00.0000] <annevk>
on the other hand, it's easy enough for me to add alt={photo} to things

[08:43:00.0000] <Philip`>
hsivonen: {...} vs [...] vs (...) vs _... etc would be somewhat like complaining about the colour of the bikeshed, but complaining about the basic concept of introducing some special internal syntax is more like complaining that the bikeshed has a halon fire suppression system and have to be careful to not accidentally set it off while you're in there, which is a more worthwhile complaint

[08:46:00.0000] <zcorpan>
i think it should be ^...^

[08:48:00.0000] <zcorpan>
alt="{:-{)}"

[08:48:01.0000] <zcorpan>
smiley for Mustache and Beard

[08:49:00.0000] <Philip`>
Use an attribute named ":-(" to indicate that the author is unhappy because they can't provide good alt text, so you'd write <img src=947618120746.jpg alt=Photo :-( >

[08:50:00.0000] <zcorpan>
who'll summarize all these problems and send to the list?

[08:50:01.0000] <zcorpan>
/me doesn't volunteer, has to work on video

[08:53:00.0000] <Philip`>
/me was going to just send email complaining how overloading syntax that breaks a small fraction of content still breaks all tools that might possibly deal with that content

[09:01:00.0000] <Lachy>
/me will write the mail, if no-one else is

[09:02:00.0000] <Philip`>
/me will send his one soon, but has probably forgotten most of the points that were discussed

[09:04:00.0000] <Lachy>
Philip`, ok. then I don't have to.

[09:35:00.0000] <takkaria>
hsivonen: yeah, the script could be public... it's a makefile

[09:43:00.0000] <takkaria>
someone couldn't help me out by telling me why the document "<meta charset='utf-16le'>" should be detected as windows-1251 according to the html5 spec?

[09:44:00.0000] <takkaria>
oh, sorry, I'm getting my testcases mixed up

[09:45:00.0000] <annevk>
utf-16 in <meta> means utf-8

[09:45:01.0000] <annevk>
not sure about utf16le

[09:46:00.0000] <jmb>
annevk: I'd expect it to be the same, seeing as it uses an ASCII-only algorithm to determine what it is

[09:51:00.0000] <zcorpan>
"If charset is a UTF-16 encoding, change it to UTF-8."

[09:52:00.0000] <annevk>
we need to change the charset stuff into some fixed list someday

[09:52:01.0000] <annevk>
UAs MUST support utf8, utf16, windows1252, ...

[09:52:02.0000] <annevk>
(writing them in their short form for no apparent reason)

[10:40:00.0000] <Philip`>
It's quite weird to read something that says "The most common Internet user protocols are:  - Telnet (remote login)  - FTP (file transfer)  - SMTP (electronic mail delivery)" and doesn't mention HTTP at all

[10:41:00.0000] <gDashiva>
I use telnet all the time to connect to my router

[10:44:00.0000] <Lachy>
gDashiva, what do you connect to your router with telnet for?

[10:46:00.0000] <Philip`>
(This document is from 1995 but it's based on one from 1994 which sounds like it's based on work from 1991, which I guess is why HTTP wasn't considered important)

[10:54:00.0000] <gDashiva>
Lachy: To use the CLI

[10:55:00.0000] <Lachy>
what for though? To configure the router or something?

[10:55:01.0000] <gDashiva>
Yes

[10:55:02.0000] <Lachy>
doesn't it have a GUI interface or web interface for that?

[10:55:03.0000] <Philip`>
My router supports SSH too

[10:55:04.0000] <Philip`>
(Unfortunately it doesn't support routing to IP addresses ending in .0)

[10:56:00.0000] <Philip`>
*SSH to access the CLI

[10:57:00.0000] <gDashiva>
Lachy: A really slow and horrible one (sometimes with JS errors for extra fun), and it doesn't support all the features

[10:57:01.0000] <gDashiva>
And that very web interface needs to be disabled for my web server anyhow

[10:57:02.0000] <gDashiva>
Otherwise it eats all the http requests :)

[11:00:00.0000] <Lachy>
gDashiva, how would it, if the URL is pointing to an IP address of another machine on the network?

[11:00:01.0000] <Lachy>
I assume the router wouldn't try to respond to external HTTP requests though?

[11:01:00.0000] <gDashiva>
You speak as if there's more than one IP address involved.

[11:01:01.0000] <Philip`>
Are you running the web server on the router?

[11:02:00.0000] <gDashiva>
No, I'm disabling the web server on the router.

[11:02:01.0000] <Philip`>
Surely you could just tell the router to redirect external connections to port 80 onto any internal IP with any port, and configure the web server accordingly, and then 192.168.0.1:80 would still be the router configuration interface

[11:03:00.0000] <gDashiva>
Sure, if it was a sensible router

[11:04:00.0000] <gDashiva>
As it is, it took me a lot of magic just to make it resolve internal requests to the external IP properly

[11:05:00.0000] <Philip`>
Hmm, I just add "192.168.0.123 mydomain.co.uk" to /etc/hosts and comment out that line whenever I move outside the LAN

[11:06:00.0000] <Lachy>
gDashiva, I don't understand how you've got it set up, or under which conditions it intercepts HTTP requests

[11:07:00.0000] <gDashiva>
Lachy: There's a router with the only public IP, and several machines behind the router. Pretty standard setup, isn't it?

[11:08:00.0000] <Lachy>
ok, sure. But which side of the router to HTTP requests need to come from to get intercepted by it?

[11:08:01.0000] <gDashiva>
Which side doesn't matter, and it serves on both its internal IP and the external IP

[11:09:00.0000] <Lachy>
so you have a web server running on a computer in your local network?

[11:09:01.0000] <gDashiva>
Yes

[11:09:02.0000] <Philip`>
Why not connect to computer in your local network using its local IP?

[11:09:03.0000] <Philip`>
s//the/

[11:10:00.0000] <gDashiva>
Because that hides all kinds of things that could be wrong for external access

[11:11:00.0000] <gDashiva>
Like the always-enjoyable event where the network decides it wants to die just long enough for the dhcp lease to expire, prompting a new external IP

[11:11:01.0000] <Lachy>
so if your router local IP is 10.0.0.1 and your web server's local IP is 10.0.0.2, and you want to connect to the web server from any other local machine, why doesn't http://10.0.0.2/ work without disabling the router's web interface?

[11:11:02.0000] <Philip`>
Hmm, fun

[11:11:03.0000] <gDashiva>
Lachy: It works, but see above

[11:12:00.0000] <Lachy>
/me *confused*

[11:12:01.0000] <Lachy>
I have roughly the same setup. But I just can't see how a router would interfere with that under any circumstances

[11:13:00.0000] <gDashiva>
If I access via 10.0.0.2 I won't notice that the domain name no longer points to the right IP, or that the router has started eating random packets, or that I messed up the vhost config and it only responds to 10.0.0.2 now, etc

[11:13:01.0000] <Philip`>
You could set up a monitoring service running on an external machine

[11:14:00.0000] <Lachy>
oh, so you want to be able to access your local web server via http://[your-external-ip]/ ?

[11:14:01.0000] <gDashiva>
Lachy: Well, usually with a bounce via DNS before the IP, but yes :)

[11:15:00.0000] <Lachy>
ok, so you're using a dynamic DNS service to provide a domain name?  That makes a little more sense. But I still don't get why the router would ever intercept the requests at all, if you'd set up port forwarding in it.

[11:16:00.0000] <gDashiva>
Because it sees "Hey, port 80, I do stuff with that" before it starts forwarding

[11:16:01.0000] <Philip`>
Can you configure it to use a different port?

[11:16:02.0000] <Lachy>
would the router ever respond with the web interface to an external request, if it were enabled?

[11:16:03.0000] <Philip`>
Can you open the firmware in a hex editor and hack it to use a different port?

[11:17:00.0000] <gDashiva>
Lachy: That's disabled by default, but it's possible

[11:17:01.0000] <gDashiva>
Philip`: Probably. Or I could just disable it and have it working.

[11:17:02.0000] <Lachy>
ok. what kind of router is it?

[11:17:03.0000] <Lachy>
(just so I know to avoid that POS in the future)

[11:17:04.0000] <Philip`>
gDashiva: That's no fun

[11:18:00.0000] <gDashiva>
I don't recall. Probably a speedtouch.

[11:18:01.0000] <Philip`>
You could open its web interface to see where it says the model nu... oh, wait

[11:19:00.0000] <gDashiva>
Philip`: I could VNC home and telnet from there, but I won't :)

[11:19:01.0000] <Philip`>
(Mine is called "ADSL Router", and there's no other visible identification anywhere)

[11:19:02.0000] <gDashiva>
But speaking of home, it's time for going there

[11:19:03.0000] <Lachy>
gDashiva, if you enable the web interface, there's probably some information about that in there somewhere

[11:19:04.0000] <gDashiva>
Lachy: Hah

[11:20:00.0000] <gDashiva>
Information would risk being useful

[11:21:00.0000] <gDashiva>
The port forwarding setup is a perfect example. Seven input boxes with names like "trigger port" and "port range"

[11:21:01.0000] <Lachy>
so what commands does it support via CLI? Is it some sort of linux based thing, or some proprietary thing?

[11:22:00.0000] <gDashiva>
It's a tree-based command system

[11:22:01.0000] <Lachy>
/me googles that

[11:22:02.0000] <Philip`>
Aha, apparently my router uses an ATMOS CLI

[11:22:03.0000] <gDashiva>
Like, you can do service; host; add name=httpd; or you can do :service host add name=httpd

[11:23:00.0000] <Lachy>
was there some documentation for that provided with the router?

[11:23:01.0000] <Philip`>
(which leads to obvious jokes about Sontarans and suchlike)

[11:23:02.0000] <gDashiva>
Of course not. The ISP has some very basic FAQs with some images, but that's all.

[11:23:03.0000] <Lachy>
ok, so you basically figured most of it out yourself?

[11:24:00.0000] <Lachy>
I have an old belkin ADSL router that I'm not using any more. I wonder if it supported any CLI

[11:24:01.0000] <gDashiva>
Lots of guessing based on command names, lots of help and some googling

[11:24:02.0000] <gDashiva>
The CLI manual is out there

[11:24:03.0000] <gDashiva>
But it only specifies the parameters to each command, not what the command actually does

[13:45:00.0000] <Hixie>
Philip`: i would imagine a latex replacement text would sound horrific with a screen reader

[13:47:00.0000] <webben>
Hixie: Dunno about that. blindmathers seem to use latex; I suspect they're used to what it sounds like; just like blind programmers are used to what code sounds like.

[13:47:01.0000] <webben>
not great for people new to math though!

[13:47:02.0000] <webben>
or ideal for anyone

[13:49:00.0000] <hsivonen>
does T.V. Raman's LaTeX to voice software integrate with any browser? Emacs W3 perhaps?

[13:51:00.0000] <Philip`>
Hixie: alt text isn't just for blind people

[13:52:00.0000] <Hixie>
no but it has to be appropriate for all the people it is for, including the blind people

[13:52:01.0000] <Philip`>
and it seems likely anyone who cares about maths would understand LaTeX in preference to any other encoding of equations

[13:52:02.0000] <hsivonen>
Hixie: afaict, there's no such alt format for all people

[13:53:00.0000] <hsivonen>
Hixie: what Philip` said

[13:53:01.0000] <Hixie>
i replied to the e-mail

[13:53:02.0000] <gsnedders>
What about my French relatives?

[13:53:03.0000] <Hixie>
providing more suitable alternative text

[13:54:00.0000] <webben>
problem is, as I said, "anyone who cares about maths" is not the only target for maths in markup

[13:54:01.0000] <webben>
also schoolkids, for instance, who might distinctly not care about maths ;)

[13:55:00.0000] <gsnedders>
Maths… eww.

[13:55:01.0000] <gsnedders>
:P

[14:44:00.0000] <Philip`>
Hmph, I was going to reply to the email but everyone else beat me to it :-(

[14:45:00.0000] <Philip`>
"(a \oplus b) \oplus c = a \oplus (b \oplus c)" is something I might write, and I don't see any way to encode it in words that is easier for anyone and isn't far more confusing for a user who might just have a graphical browser with images turned off

[14:46:00.0000] <Philip`>
(particularly since the target audience for documents written using LaTeX syntax is usually people familiar with maths and hence with LaTeX)

[14:46:01.0000] <hsivonen>
xml-dev gets mirrored in crazy many places. I post there and get multiple entries in my vanity feeds

[14:56:00.0000] <Hixie>
Philip`: i have no idea what that even means, which is for me pretty good evidence that latex is terrible alt text :-)

[14:57:00.0000] <Hixie>
but in any case

[14:57:01.0000] <Hixie>
i don't especially like the {...} thing either

[14:57:02.0000] <Philip`>
Hixie: It's evidence that you're not the target audience for such things :-)

[14:57:03.0000] <Hixie>
but right now i know of nothing better

[14:57:04.0000] <Hixie>
Philip`: i read wikipedia math pages

[15:10:00.0000] <Lachy>
Hixie, if the {...} syntax doesn't work out, you could try the kind="..." attribute I suggested in here quite a while ago

[15:10:01.0000] <Hixie>
a separate attribute has two problems

[15:10:02.0000] <Hixie>
1. it doesn't make alt="" required

[15:10:03.0000] <Hixie>
2. it doesn't have a good fallback story

[15:11:00.0000] <hsivonen>
Hixie: it seems to be that a boolean attribute behave-as-if-there-where-braces="" addresses both issues

[15:12:00.0000] <jgraham>
Hixie: 1 isn't a good reason for anything per-se

[15:12:01.0000] <Lachy>
ok, to address #2, we would need a value in either the alt or title attributes

[15:12:02.0000] <hsivonen>
Hixie: if now alt="{}" is ok, why wouldn't alt="" behave-as-if-there-where-braces="" be as good?

[15:12:03.0000] <jgraham>
I think what hsivonen says makes sense

[15:12:04.0000] <Hixie>
hsivonen: nobody has come up with a good name for such an attribute

[15:13:00.0000] <Hixie>
(that was what i initially proposed)

[15:13:01.0000] <hsivonen>
Hixie: also, the fallback story with behave-as-if-there-where-braces="" is even better (no brace clutter)

[15:13:02.0000] <hsivonen>
alt-is-type=''

[15:13:03.0000] <Hixie>
hsivonen: the brace clutter is a good thing

[15:13:04.0000] <jgraham>
alt-autogenerated

[15:13:05.0000] <hsivonen>
alt-is-not-alt

[15:13:06.0000] <Hixie>
like i said

[15:13:07.0000] <Hixie>
nobody has come up with a good name for such an attribute

[15:14:00.0000] <jgraham>
Hixie: It seems like a bad reason to reject it in favour of a solution with more substantial problems

[15:14:01.0000] <Dashiva>
alternate alternate text is a mess no matter how you put it :)

[15:14:02.0000] <Hixie>
if we come up with a bad name like those, or any that i have proposed, we'll just have people copy that attribute all over the place thinking it's the cool new thing

[15:14:03.0000] <Hixie>
jgraham: i'm not at all convinced that {} has more problems

[15:14:04.0000] <Hixie>
jgraham: the only problem i'm aware of is the latex thing, and i don't even agree that that is valid alt text

[15:14:05.0000] <hsivonen>
Hixie: why wouldn't some people think braces are the new black?

[15:14:06.0000] <jgraham>
Hixie: It breaks wikipedia (for some value of breaks)

[15:15:00.0000] <Hixie>
jgraham: wikipedia is already broken for all those cases

[15:15:01.0000] <hsivonen>
Hixie: there's also the collision problem Philip` pointed out

[15:15:02.0000] <Hixie>
hsivonen: because they wouldn't recognise them as anything special, they'd just think that was what the author happened to use -- it doesn't look like syntax

[15:15:03.0000] <jgraham>
Hixie: I disagree. I think LaTeX is the best fallback for maths content

[15:15:04.0000] <Hixie>
collision problem?

[15:15:05.0000] <hsivonen>
if I have a text field for alt, I need to write code that does *something* if the user enters a { as the first character

[15:15:06.0000] <Hixie>
jgraham: I disagree. I think it's horrible. :-)

[15:16:00.0000] <jgraham>
I also think that alt-autogenerated makes it look uncool

[15:16:01.0000] <Hixie>
hsivonen: why?

[15:16:02.0000] <Hixie>
jgraham: that name is a lie

[15:16:03.0000] <jgraham>
Hixie: I would be interested to hear the opinion of an AT-dependant maths user

[15:16:04.0000] <hsivonen>
Hixie: because that's how careful developers think

[15:17:00.0000] <jgraham>
Hixie: why is the name a lie

[15:17:01.0000] <hsivonen>
alt-unspecific

[15:17:02.0000] <Hixie>
jgraham: because it might not be autogenerated, and even if it was, it might not be appropriate to set the attribute

[15:18:00.0000] <hsivonen>
alt-quality=low

[15:18:01.0000] <Lachy>
hsivonen, would there be an alt-quality=high?

[15:18:02.0000] <hsivonen>
alt-type=category

[15:18:03.0000] <hsivonen>
Lachy: there could be

[15:19:00.0000] <jgraham>
no-text-equivalent

[15:19:01.0000] <Lachy>
then it becomes too subjective. Some authors will think their alt text is high quality, even if it's actually quite low (or vice versa)

[15:19:02.0000] <Dashiva>
jgraham: That's just no-alt with longer text :)

[15:20:00.0000] <jgraham>
(it doesn't really matter if the attribute has a long name since the typical case is automated systems not hand editing)

[15:20:01.0000] <jgraham>
Dashiva: I think it makes more sense than <img alt noalt>

[15:34:00.0000] <clotman>
Is the Web Forms 2.0 draft still being updated?

[15:37:00.0000] <Lachy>
clotman, it will be integrated into HTML5 soon

[15:37:01.0000] <clotman>
thanks for the info

[16:50:00.0000] <annevk>
Hixie, Web Workers is in quirks mode?

[16:51:00.0000] <Hixie>
yeah bert's script removes the doctype

[16:53:00.0000] <annevk>
ah, the original is in HTML5

[16:59:00.0000] <annevk>
/me reaches the thread about renaming <img>


2008-08-08
[17:00:00.0000] <annevk>
/me wonders why this person keeps insisting on changing it to <image> while it has been pointed out repeatedly in the past that it won't work

[17:40:00.0000] <mcarter>
in section 7.2.3, about parsing SSE, when is it determined what the newline character is? can it change line by line?

[17:42:00.0000] <annevk>
never, yes

[17:43:00.0000] <Hixie>
any line can be terminated by 0x0D, 0x0A, 0x0D0A, or the end of the file

[17:43:01.0000] <annevk>
just like it works for text/html resources

[17:46:00.0000] <mcarter>
so how do you tell if a particular \r counts as the end of a line. do you see if data: comes immediately after?

[17:48:00.0000] <mcarter>
i mean, i guess not all lines are data fields (i was just thinking about this in terms of the data)

[17:49:00.0000] <annevk>
for \r you check if a \n follows it

[17:49:01.0000] <annevk>
iirc you could first split on lines and then do the rest of the parsing

[17:50:00.0000] <mcarter>
the way the spec reads, the stream "data: hello\r\ndata: world\r\n\r\n" could be interpreted as either a single event "hello\nworld" or two events "hello" and "world"

[17:52:00.0000] <Hixie>
when you get a 0x0D, you look to see if the next byte is 0x0A

[17:52:01.0000] <Hixie>
if it is, you treat them as a single newline

[17:54:00.0000] <annevk>
section 7.2.4 is actually pretty clear about processing, no?

[17:54:01.0000] <mcarter>
so then is this payload allowed in SSE: "hello\rdata: world"

[17:56:00.0000] <annevk>
I don't think so

[17:56:01.0000] <annevk>
newlines are \n

[17:57:00.0000] <mcarter>
but this payload is allowed: "hello\rworld" ?

[17:58:00.0000] <mcarter>
or would that cause an event with data = "hello" and ignore the "world field" ?

[17:59:00.0000] <annevk>
the latter

[18:00:00.0000] <annevk>
"data: hello\rworld" becomes "data: hello", "world" per 7.2.4

[18:00:01.0000] <mcarter>
so there's no way to send the "\r" character at all over sse?

[18:00:02.0000] <annevk>
right

[18:00:03.0000] <annevk>
afaict

[18:02:00.0000] <mcarter>
do you know what the rationale is for not allowing the '\r' character?

[18:03:00.0000] <annevk>
it seems like a side effect of allowing it to be a newline character

[18:05:00.0000] <mcarter>
well, we had to switch away from SSE for orbited (we implemented it in gecko/webkit with xhr streaming) because we weren't able to send protocols like IRC over it (which use '\r')

[18:05:01.0000] <mcarter>
and its a shame that we wouldn't be able to use native SSE either, without some kind of extra encoding

[18:06:00.0000] <annevk>
IRC uses \r? lame

[18:09:00.0000] <mcarter>
i sort of feel like its the perogative of 20 year old protocols to use whatever framing they use

[18:12:00.0000] <annevk>
I suppose

[18:18:00.0000] <mcarter>
Hixie, is it your intention to disallow "\r" in the payload, or was it an oversight? (I'm getting ready to send some feedback in about it)

[18:18:01.0000] <Hixie>
yes

[18:18:02.0000] <Hixie>
\r in general isn't supported on the web

[18:18:03.0000] <Hixie>
use \n

[18:20:00.0000] <mcarter>
Hixie, but then SSE becomes useless for tunneling other text-based protocols

[18:21:00.0000] <mcarter>
other text-based protocols that require \r, anyway

[18:21:01.0000] <Hixie>
yes

[18:21:02.0000] <Hixie>
are there any such protocols where you couldn't translate \r to \n on the server?

[18:25:00.0000] <mcarter>
No. I can encode any protocol in base64 or hexidecmal or a string of "0" and "1" -- its just not ideal

[18:27:00.0000] <Hixie>
converting \r to \n seems like a non-issue

[18:28:00.0000] <Hixie>
it's not like you can use this without some level of conversion anyway

[18:28:01.0000] <mcarter>
it means though that \n needs to be escaped somehow as well

[18:29:00.0000] <mcarter>
I think supporting \r as a line ending as catering to a much smaller issue that allowing \r in the protocol would be

[18:31:00.0000] <mcarter>
I also think that silent stripping out any data after a \n and throwing it away is a pretty big problem

[18:31:01.0000] <mcarter>
* i mean, any data after a \r

[18:32:00.0000] <Hixie>
what i mean is, why can't you just convert all \rs to \ns, and all \ns to \ns, losing the distinction

[18:35:00.0000] <mcarter>
in the case where the data your sending has the characteristic that  \r\n is used as framing and \n is allowed in the payload

[18:42:00.0000] <Hixie>
are there any such cases?

[18:42:01.0000] <Hixie>
seems like that kind of protocol would be awful

[18:45:00.0000] <Hixie>
i mean we also can't handle cases where the data contains malformed UTF-8

[18:45:01.0000] <Hixie>
it's not really intended for tunneling existing protocols

[18:49:00.0000] <franksalim>
#actionscript

[18:49:01.0000] <franksalim>
ha

[18:49:02.0000] <franksalim>
whoops

[18:49:03.0000] <franksalim>
/me slits wrists

[18:57:00.0000] <franksalim>
mcarter, i don't even think that is the main problem with the line breaks in SSE

[18:59:00.0000] <franksalim>
\r\n is ambiguous...it could be a single break or two breaks (in the unlikely case that the server changed break character mid stream)

[18:59:01.0000] <franksalim>
in my understanding

[19:00:00.0000] <annevk>
that's just the server guys not reading the spec, doesn't make the spec ambigious

[19:02:00.0000] <franksalim>
annevk, i thought the issue was that any line could be terminated by a CR or a LF

[19:02:01.0000] <franksalim>
the 'look to see if the next character' thing doesn't really work if you get the CR and the LF in two read events

[19:03:00.0000] <franksalim>
do you just not dispatch a CR terminated event until there is another byte

[19:04:00.0000] <annevk>
right

[19:04:01.0000] <franksalim>
annevk, what if that was the last event...maybe another event doesn't arrive for an hour

[19:04:02.0000] <annevk>
well, you can simply dispatch it actually

[19:05:00.0000] <annevk>
you just need to ignore the next byte if it's LF

[19:06:00.0000] <franksalim>
scratch that

[19:06:01.0000] <franksalim>
consider the case where you receive data:something\r

[19:07:00.0000] <franksalim>
and then \n a short time later. do you dispatch it? or do you consider the \r\n a single newline?

[19:07:01.0000] <annevk>
see the spec

[19:08:00.0000] <franksalim>
annevk, i have seen the spec. i think there is an ambiguity

[19:08:01.0000] <Hixie>
you dispatch it as "something"

[19:08:02.0000] <Hixie>
and then when you get the \n later, you ignore the \n

[19:08:03.0000] <Hixie>
you don't ever need to look ahead

[19:08:04.0000] <Hixie>
you only ever need to look back

[19:08:05.0000] <franksalim>
you dispatch it, even though it is not followed by a blank line?

[19:09:00.0000] <franksalim>
"If the line is empty (a blank line) Dispatch the event, as defined below. "

[19:09:01.0000] <annevk>
\r is also a valid newline character

[19:09:02.0000] <franksalim>
so data:something\r should not be dispatched as something

[19:09:03.0000] <Hixie>
i don't really understand the question

[19:09:04.0000] <Hixie>
could you give some more background?

[19:10:00.0000] <annevk>
anyways, bedtime here

[19:10:01.0000] <franksalim>
annevk, goodnight

[19:10:02.0000] <franksalim>
Hixie, i will try

[19:11:00.0000] <franksalim>
data:something\r is not dispatched until a blank line is seen

[19:11:01.0000] <franksalim>
so data:something\r\r would be dispatched

[19:11:02.0000] <franksalim>
(as "something")

[19:12:00.0000] <franksalim>
are we on the same page so far?

[19:13:00.0000] <franksalim>
(you can tell me i'm crazy, i won't be offended)

[19:17:00.0000] <franksalim>
annevk, perhaps another time

[19:19:00.0000] <Hixie>
i don't know what you mean by "dispatched"

[19:19:01.0000] <Hixie>
are we talking server or client?

[19:19:02.0000] <Hixie>
where is the data coming from?

[19:19:03.0000] <franksalim>
client. by dispatched i mean dispatched as a messageevent

[19:20:00.0000] <Hixie>
so the client receives "data:something\r".

[19:20:01.0000] <Hixie>
ok

[19:20:02.0000] <franksalim>
yes. that is the case i am worried about

[19:20:03.0000] <franksalim>
the client receives "data:something\r". the next byte is "\n"

[19:21:00.0000] <Hixie>
so what's the problem?

[19:21:01.0000] <franksalim>
it is not clear if that "\n" was part of a "\r\n" or was a blank line

[19:21:02.0000] <Hixie>
if a \n follows a \r it is part of the \r\n sequence.

[19:21:03.0000] <Hixie>
always.

[19:21:04.0000] <franksalim>
since "\r\n", "\n", and "\r" are all considered valid line breaks

[19:23:00.0000] <franksalim>
is that implied somehow in the pseudo-BNF?

[19:23:01.0000] <Hixie>
the bnf is non-normative

[19:23:02.0000] <Hixie>
the next section says: The stream must then be parsed by reading everything line by line, with a U+000D CARRIAGE RETURN U+000A LINE FEED (CRLF) character pair, a single U+000A LINE FEED (LF) character, a single U+000D CARRIAGE RETURN (CR) character, and the end of the file being the four ways in which a line can end

[19:24:00.0000] <Hixie>
i can make this more explicit if you have reason to believe that browsers are going to implement thing incorrectly if you like

[19:29:00.0000] <franksalim>
i think the odds of a server using \r and \n interchangeably (which would break things) are low enough as not not need clarification, especially if you do not think it is worth the time

[19:29:01.0000] <franksalim>
*as to not

[19:36:00.0000] <franksalim>
actually, i just read the paragraph on appropriate buffering, and i think that takes care of it. i think that _definitely_ takes care of it

[19:36:01.0000] <franksalim>
Hixie, sorry

[19:37:00.0000] <Hixie>
there's a paragraph on buffering? :-)

[19:37:01.0000] <franksalim>
Hixie, yes. it might make me feel a little less foolish if that went in recently

[19:38:00.0000] <franksalim>
haha. this is my finest hour

[19:39:00.0000] <Hixie>
i haven't touched that section in a while :-)

[19:39:01.0000] <Hixie>
which paragraph do you mean?

[19:41:00.0000] <franksalim>
Since connections established to remote servers for such resources are expected to be long-lived, UAs should ensure that appropriate buffering is used. In particular, while line buffering may be safe if lines are defined to end with a single U+000A LINE FEED character, block buffering or line buffering with different expected line endings can cause delays in event dispatch. 

[19:42:00.0000] <franksalim>
it's a pretty good paragraph

[19:43:00.0000] <Hixie>
ah ok

[19:43:01.0000] <Hixie>
cool

[00:39:00.0000] <aboodman3>
hm, sorry about all the aboodmans. I don't know how to operate irc.

[00:40:00.0000] <Hixie>
no worries

[00:40:01.0000] <Hixie>
aboodmans are always welcome

[00:40:02.0000] <aboodman3>
Hixie: has anyone implemented MessageChannels?

[00:41:00.0000] <Hixie>
not to my knowledge, it was added only recently

[00:52:00.0000] <aboodman3>
when a port becomes unentangled, do messages start getting queued for it?

[00:53:00.0000] <Hixie>
no it just returns false

[00:53:01.0000] <Hixie>
when a port is unentangled that it's, it can never be entangled again

[00:54:00.0000] <aboodman3>
ok

[03:15:00.0000] <annevk>
Hixie, you wrote port(); rather than port; in one of the Workers IDL blocks

[03:43:00.0000] <Philip`>
http://wiki.whatwg.org/wiki/User_talk:Flowerwish - spam

[03:46:00.0000] <Lachy_>
Philip`, deleted

[04:08:00.0000] <hsivonen>
looks like marcos is going to have a rough ride practicing extensibility of the URI system: http://lists.w3.org/Archives/Public/www-tag/2008Aug/0068.html

[04:10:00.0000] <hsivonen>
if URI schemes supported Unilateral Extensibility, he could just proceed without consulting with any authority

[04:10:01.0000] <annevk>
seems he sort of misses the point talking about media types though

[04:11:00.0000] <hsivonen>
annevk: he meing stuart or marcos?

[04:11:01.0000] <annevk>
widget: is about what address a file has *within* a package

[04:11:02.0000] <annevk>
Stuart

[04:11:03.0000] <hsivonen>
/meing/being/

[04:11:04.0000] <annevk>
with the constrain that the address needs to be consistent regardless of where the package is

[04:19:00.0000] <annevk>
more URI scheme fun: http://omniplex.blogspot.com/2008/08/leader-of-pack.html

[04:22:00.0000] <hsivonen>
boo. del.icio.us is now delicious.com. no fun

[04:24:00.0000] <Philip`>
"We’ve seen a zillion different confusions and misspellings of “del.icio.us” over the years (for example, “de.licio.us”, “del.icio.us.com”, and “del.licio.us”), so moving to delicious.com will make it easier for people to find the site and share it with their friends." - that sounds fairly reasonable

[04:24:01.0000] <Philip`>
although they could have solved part of the problem by registering names like licio.us themselves

[04:25:00.0000] <annevk>
haha, markp about his blog.whatwg.org post "i admire my ability to get paid for this"

[04:27:00.0000] <hsivonen>
I admire markp's ability to restrain himself from posting to the WHATWG list. I mean, no doubt he has opinions about this stuff.

[04:28:00.0000] <Lachy_>
I upgraded the blog to wordpress 2.6, and wrote a shell script to simply the whole process in the future

[04:28:01.0000] <Lachy_>
let me know if you see any problems with the upgrade

[04:28:02.0000] <othermaciej>
I predict at least one problem

[04:28:03.0000] <othermaciej>
which is that you will still be using wordpress

[04:29:00.0000] <Lachy_>
othermaciej, do you have an alternative proposal?

[04:29:01.0000] <Philip`>
All the cool people write their own blogging software

[04:29:02.0000] <othermaciej>
no

[04:29:03.0000] <othermaciej>
but it is fun to bitch about wordpress

[04:29:04.0000] <Lachy>
one day, when I find the time, I will write my own blog

[04:30:00.0000] <Lachy>
othermaciej, agreed :-)

[04:30:01.0000] <othermaciej>
the sloppy markup, the hacky implementation, the shoddy security

[04:30:02.0000] <othermaciej>
no end of fun

[04:30:03.0000] <hsivonen>
I have wanted to write a blogging system and an online photo management system

[04:30:04.0000] <hsivonen>
then I figured that Flickr has the photo thing solved

[04:31:00.0000] <Philip`>
Blogging really ought to be a solved problem by now

[04:31:01.0000] <hsivonen>
but for the blogging system, I'm still using flat files under Apache and a Jython script in cron

[04:32:00.0000] <hsivonen>
because I wouldn't be practicing what I preach if I used a PHP system like WP

[04:32:01.0000] <hsivonen>
also, I become extremely unhappy when I have to fix someone else's PHP

[04:32:02.0000] <krijnh>
Rolling your own blogging thing definitely is fun :)

[04:32:03.0000] <hsivonen>
and I'm sure using WP would put me into that situation

[04:32:04.0000] <annevk>
I'm pretty happy with flickr though, although I want it to be easier to add tags to photos

[04:33:00.0000] <annevk>
for instance, modifying tags of photos on overview pages would be most welcome

[04:33:01.0000] <hsivonen>
the problem with the Jython script is that it such a legacy setup that I can't update it of move my blog to a different machine without breaking it

[04:34:00.0000] <hsivonen>
also, my CPython cron jobs depend on PyGenx, which doesn't have a Ubuntu package, so I can't move those scripts, either

[04:34:01.0000] <hsivonen>
yay for legacy

[04:38:00.0000] <annevk>
can't you export your content to a new system somehow?

[04:38:01.0000] <hsivonen>
annevk: content, yes. trivially. cron jobs, not so trivially

[04:39:00.0000] <hsivonen>
I'd love to have a pure-Python implementation of PyGenx

[04:40:00.0000] <hsivonen>
although now I haveextra disk space on the Ubuntu server, so I guess I could go ahead and install a compiler along with alla the dev packages

[04:41:00.0000] <hsivonen>
or devel rather

[04:43:00.0000] <hsivonen>
not being a bozo with feeds comes at a high price

[04:47:00.0000] <hsivonen>
someone should write something like Movable Type (generating flat files for Apache) with a TurboGears-based admin front end

[05:03:00.0000] <Lachy>
I originally planned to write my own blogging system in PHP, but now I want to write my own blogging system in python using html5lib

[05:04:00.0000] <gDashiva>
And soon you'll want to write it in xhtml2 using xforms

[05:04:01.0000] <Lachy>
gDashiva, no

[05:04:02.0000] <gDashiva>
The revolution is coming

[05:05:00.0000] <Lachy>
it will have XHTML on the backend for templates and authoring, and then serialise it as HTML

[05:06:00.0000] <zcorpan>
why not HTML for authoring? or what do you mean by authoring?

[05:06:01.0000] <Lachy>
has anyone written an HTML5 serialiser in python yet?

[05:06:02.0000] <zcorpan>
i think kid or genshi has one

[05:07:00.0000] <Lachy>
zcorpan, because I like being able to write markup in XHTML cause if I make a well-formedness errror, then I notice and fix it straight away.

[05:07:01.0000] <Lachy>
it makes me less reliant on on a validator for syntax checking

[05:08:00.0000] <virtuelv>
re the alt={foo} syntax: How does it impact existing server-side tools/templating languages?

[05:09:00.0000] <annevk>
it makes you reliant on XML though

[05:09:01.0000] <Lachy>
virtuelv, are there any such languages that use braces on their own? I know JSP expression language using ${...}, so that won't be affected.

[05:11:00.0000] <Lachy>
annevk, is that a problem?

[05:12:00.0000] <virtuelv>
Lachy: movable type has/had {} in templates

[05:13:00.0000] <annevk>
dunno

[05:13:01.0000] <Lachy>
virtuelv, ok. Then that might be a problem.

[05:14:00.0000] <virtuelv>
Django uses it

[05:15:00.0000] <virtuelv>
{% block title %}

[05:15:01.0000] <virtuelv>
http://www.djangoproject.com/documentation/templates/

[05:15:02.0000] <Lachy>
virtuelv, movable type uses {{...}}

[05:15:03.0000] <virtuelv>
IIRC it didn't initially

[05:15:04.0000] <Lachy>
know of any documentation that shows that?

[05:17:00.0000] <virtuelv>
Lachy: http://bradchoate.com/weblog/2003/07/19/simple-template

[05:21:00.0000] <virtuelv>
my memory seemed a bit off, though

[05:21:01.0000] <Lachy>
virtuelv, that's using them with $ signs, so there doesn't appear to be a problem

[05:21:02.0000] <virtuelv>
my main concern would be {% something %} though

[05:22:00.0000] <Lachy>
I don't see how that would be a problem

[05:22:01.0000] <virtuelv>
also survey, http://www.whenpenguinsattack.com/2006/07/19/php-template-engine-roundup/

[05:22:02.0000] <virtuelv>
Lachy: because you introduce ambiguity

[05:23:00.0000] <Lachy>
how is it ambiguous? {...} is different from {%...%}

[05:23:01.0000] <virtuelv>
yes, but {%foo%} is not in any way prohibited as alt-text now, is it?

[05:25:00.0000] <Lachy>
it's highly unlikely that anyone would legitimately use the % signs like that

[05:25:01.0000] <Lachy>
for alt text

[05:25:02.0000] <virtuelv>
Lachy: or {{foo}}?

[05:26:00.0000] <Lachy>
that's also unlikely to be a legitimate alt text

[05:27:00.0000] <zcorpan>
{{:-{)}}

[05:27:01.0000] <Lachy>
besides, if someone is using such a templating language, wants alt="{{foo}}" as alt text, then they would have to use some escaping mechanism in their templating langugae

[05:27:02.0000] <Lachy>
zcorpan, how does that clash with templating languages?

[05:28:00.0000] <zcorpan>
Lachy: it doesn't but it could be legitemate alt text

[05:28:01.0000] <Lachy>
well, if they want that as alt text, then it's a problem regardless of what HTML5 says.

[05:28:02.0000] <zcorpan>
why?

[05:28:03.0000] <Lachy>
because {{ ... }} clashes with some templating languages

[05:28:04.0000] <virtuelv>
here's a fun one

[05:29:00.0000] <virtuelv>
{a, b, c}

[05:29:01.0000] <virtuelv>
denoting the set a, b, c

[05:29:02.0000] <Lachy>
virtuelv, is that something from a templating language, or intended to be alt text?

[05:30:00.0000] <Lachy>
because I'm only interested in something that a) could be used as legitimate alt text, and b) directly clashes with a templating language.

[05:30:01.0000] <virtuelv>
Lachy: intended to be alt text

[05:30:02.0000] <virtuelv>
have a look at http://en.wikipedia.org/wiki/Set

[05:31:00.0000] <Lachy>
ok, that's a separate problem from the templating language problems we were talking about.

[05:31:01.0000] <virtuelv>
yes

[05:32:00.0000] <virtuelv>
templating languages is but one of the problem with microsyntaxes

[05:32:01.0000] <virtuelv>
WP's alt text seems to be latex-centric

[05:32:02.0000] <Lachy>
I'm not yet convinced it is a real problem though, since you've yet to demonstrate a real clash

[05:33:00.0000] <virtuelv>
but I'd have written {1,3} is a proper subset of {1,2,3,4}

[05:33:01.0000] <Lachy>
the latex problem may be a real problem. I'm not arguing either way on that particular issue

[05:34:00.0000] <virtuelv>
latex in itself is not the problem

[05:34:01.0000] <virtuelv>
the problem is that curly brackets (or any other brackets) have history and prior use outside of HTML5

[05:34:02.0000] <Lachy>
I didn't say it was. I was just referring to it as the latex problem because latex examples are the most common that have been raised

[05:35:00.0000] <virtuelv>
http://en.wikipedia.org/wiki/Curly_bracket#Uses_of_.E2.80.9C.7B.E2.80.9D_and_.E2.80.9C.7D.E2.80.9D

[05:35:01.0000] <Lachy>
I'm just not convinced the templating language issue is a real problem

[05:37:00.0000] <virtuelv>
Lachy: you will not know until you have actually surveyed them

[05:38:00.0000] <Lachy>
virtuelv, I know. That's why I've been looking at every template language you mentioned to see if it really does clash

[05:38:01.0000] <Lachy>
so far, i have seen one that does

[05:39:00.0000] <Philip`>
All characters have history and prior use outside of HTML5

[05:39:01.0000] <Lachy>
oh, here's one. smarty uses: {include file="footer.tpl"}

[05:40:00.0000] <Lachy>
http://www.smarty.net/crashcourse.php

[05:40:01.0000] <Lachy>
here's one that could really clash: <tr bgcolor="{cycle values="#aaaaaa,#bbbbbb"}"> (if that was <img alt> instead)

[05:41:00.0000] <Philip`>
E4X does stuff like 'var x = <img src={imgsrc} alt={imgalt}>;', if I remember correctly

[05:41:01.0000] <Philip`>
but that's not ambiguous since it's proper XML

[05:46:00.0000] <Philip`>
It would be easy if UAs weren't expected to implement special processing for not-good-alt images, because then alt="{Photo}" and alt="{...latex etc...}" could both be conforming and it wouldn't matter than a UA couldn't tell them apart

[05:46:01.0000] <Philip`>
particularly since in practice a UA won't be able to tell them apart anyway, because everyone is going to ignore whatever the spec says

[05:47:00.0000] <virtuelv>
the alt situation is, I believe, fracked up beyond repair

[05:47:01.0000] <virtuelv>
s/ck/kk/

[05:48:00.0000] <virtuelv>
<img noalt>

[05:53:00.0000] <Lachy>
virtuelv, noalt by itself doesn't address the problem, and when used with alt, then it's a bad name

[06:41:00.0000] <annevk>
jgraham, you forgot that no-text-equivalent has a bad fallback scenario

[06:42:00.0000] <Philip`>
Why is it bad?

[06:44:00.0000] <Philip`>
I think most current UAs announce that images are images, rather than treating them like <span>...alt text...</alt>, so it won't have the problem that they'll be indistinguishable from plain text (which is the problem that the new feature is trying to solve)

[06:44:01.0000] <annevk>
that's probably true, indeed

[06:55:00.0000] <virtuelv>
besides, I don't think the huge amount of DSC_[0-9]{4,}.jpg's on the web are going to get sensible alt's anytime soon

[06:56:00.0000] <virtuelv>
{DSC_0201.jpg} is no more meaningful than [IMAGE]

[06:56:01.0000] <jcranmer>
[bright blue swath of pixels in the upper half of the image, dull green swath of pixels in the lower half. Various hues of pixels scattered throughout] ?

[06:56:02.0000] <annevk>
a lot of my flickr images have meaningless titles

[06:57:00.0000] <annevk>
I don't really expect this to change

[06:57:01.0000] <virtuelv>
trying to fix the alt situation is like trying to explain every joke you ever tell

[06:58:00.0000] <Philip`>
That would be difficult - usually I don't even know whether I'm joking or not

[06:58:01.0000] <virtuelv>
yet, alt is trying to describe everything

[06:59:00.0000] <virtuelv>
If I show you a 7-11 cup with the "Oslo" text visible and three coins, you'll get the joke if it was meant for you to get

[06:59:01.0000] <virtuelv>
I couldn't possibly explain it to you without ruining it

[07:00:00.0000] <virtuelv>
s/joke/in-joke/

[07:00:01.0000] <jcranmer>
[7-11 cup with "Oslo" text visible and three coins], that should be good ALT? :-)

[07:00:02.0000] <virtuelv>
jcranmer: no

[07:02:00.0000] <virtuelv>
The proper alt-text would be "Oslo needs to do something about it's begging problem"

[07:03:00.0000] <virtuelv>
7-11 are standard-issue beggar cups

[07:03:01.0000] <jcranmer>
/me was expecting one of this rigged game explanations

[07:03:02.0000] <jcranmer>
"find under which coin the cup is under"

[07:05:00.0000] <virtuelv>
http://xkcd.com/3/

[07:06:00.0000] <virtuelv>
{Island (sketch)} or {Cartoon} or "I'm lonely"?

[07:09:00.0000] <virtuelv>
imo, {Cartoon} would best be replaced by class=Cartoon

[07:09:01.0000] <virtuelv>
s/class/whatever/

[07:09:02.0000] <virtuelv>
it's not alternative

[07:09:03.0000] <virtuelv>
it represents a classification

[07:11:00.0000] <Philip`>
http://www.mathtalk.com/demos/Demos/MathTalkforVisuallyImpaired.avi looks kind of painful to use

[07:14:00.0000] <jcranmer>
I hope your happy

[07:14:01.0000] <jcranmer>
s/r/'re/

[07:14:02.0000] <jcranmer>
I've now spent the past 10 minutes browsing XKCD archives

[07:14:03.0000] <virtuelv>
gawd, that looks painful

[07:18:00.0000] <jcranmer>
painful would be an understatement

[07:22:00.0000] <Philip`>
I would guess it's more bearable when you're just using voice output, not voice input

[07:22:01.0000] <Philip`>
(and using a normal keyboard)

[07:40:00.0000] <Philip`>
It took me way too long to find a joke in LaTeX

[07:46:00.0000] <Philip`>
I'm sure I saw one ages ago in someone's signature on Slashdot, but I can't remember it at all :-/

[07:48:00.0000] <hsivonen>
hendry: is this on your radar: http://www.w3.org/TR/2008/WD-ct-guidelines-20080801/ ?

[07:48:01.0000] <Philip`>
Oh, I think it was some variant of http://www.xs4all.nl/~jcdverha/scijokes/1_11.html#1

[07:53:00.0000] <hsivonen>
/me predicts: people who think Content Transformation sucks will start putting Cache-Control: no-transform in their headers

[07:53:01.0000] <hsivonen>
/me predicts: next proxy vendors start ignoring the header if they aren't already

[07:54:00.0000] <hsivonen>
solution: people use browsers that suck less with carriers who offer pure IP routing

[07:58:00.0000] <annevk>
oh yes, near free tickets (~400 EUR) to the Kilimanjaro thanks to miles

[07:58:01.0000] <annevk>
apparently they're good for something

[07:58:02.0000] <annevk>
can even get business class on the way back

[08:00:00.0000] <hsivonen>
is the Kilimanjaro area tourist-friendly these days?

[08:00:01.0000] <annevk>
afaik, yes

[08:09:00.0000] <annevk>
collegues climbed it last summer, I read on some blog of one of the former Mozilla XForms guys that he climbed it too recently; that's enough for me

[08:10:00.0000] <hsivonen>
I see that Kilimanjaro is entirely on the Tanzanian side. Somehow I had thought the border with Kenya ran through it. (I need to update my geography knowledge.)

[08:10:01.0000] <hsivonen>
/me is assuming the Kenya is not good for tourism these days, but what do I know

[08:12:00.0000] <hsivonen>
/me notices http://www.panoramio.com/ 

[08:12:01.0000] <hsivonen>
it seems it would be better for users if Google Maps pulled geotagged photos from the whole Web--including Flickr

[08:13:00.0000] <annevk>
oh, dunno about Kenya either; last thing I read was not good

[08:14:00.0000] <annevk>
Kenya has Mount Kenya which is the second highest mountain of Africa iirc

[08:16:00.0000] <hsivonen>
wow. Google Maps now has Wikipedia intergration awesomeness

[08:16:01.0000] <hsivonen>
how new is that?

[08:16:02.0000] <mpt>
hsivonen, both Wikipedia and (I assume) have spam control. The Web of geotagged photos in general does not. :-)

[08:17:00.0000] <mpt>
...(I assume) Panoramio have...

[08:17:01.0000] <hsivonen>
mpt: well, Google does have pretty good spam control on its own for Image Search

[08:18:00.0000] <hsivonen>
well, maybe they focus on filtering out porn rather than spam per se

[08:19:00.0000] <mpt>
I think spam's probably easier to filter out of an image search than out of a map, because you've given the image search engine some indication of what you're searching for

[08:20:00.0000] <hsivonen>
so far I haven't noticed spammy coordinate data on Flickr, but I haven't exactly looked, either

[08:20:01.0000] <mpt>
whereas if you're looking at a map, who knows why

[08:20:02.0000] <annevk>
where do you see this Wikipedia integration in Google maps?

[08:21:00.0000] <hsivonen>
annevk: there's a rectangle labeled More left to Map/Satellite/Terrain

[08:22:00.0000] <hsivonen>
it seems to me that Flickr as a data source is just owned by a competitor

[08:22:01.0000] <hsivonen>
but as far as searching *the Web* goes, getting data from there as well would be cool

[08:23:00.0000] <annevk>
neat

[08:23:01.0000] <annevk>
yeah, flickr integration would be nice

[08:24:00.0000] <annevk>
or random photo data source, even

[08:27:00.0000] <hsivonen>
/me is amused that the Content Transformation Landscape document enumerates Flash and SVGT under "Non Web Applications"

[08:54:00.0000] <hendry>
hsivonen: certainly is, thanks. have not read it carefully. i sent word to luca wurfl, and he sent in some good comments

[09:00:00.0000] <hsivonen>
hendry: is your employer in the business of creating/selling tranforming proxies?

[09:02:00.0000] <hendry>
hsivonen: no

[09:03:00.0000] <hendry>
hsivonen: how is validator.nu traffic? are many people using the gnu error format output?

[09:03:01.0000] <hsivonen>
hendry: I don't have up-to-date hard data, but my impression is that gnu and json are more popular than xml

[09:05:00.0000] <hsivonen>
lol. XML is so unpopular I might as well remove it

[09:06:00.0000] <hsivonen>
HTML is the most popular output format (obviously)

[09:07:00.0000] <hsivonen>
then from more popular to less popular: gnu, text, json, xml, xhtml

[09:07:01.0000] <hendry>
hsivonen: cool :)

[09:08:00.0000] <hsivonen>
those were for the HTML5 facet

[09:08:01.0000] <hsivonen>
note that Hixie uses out=gnu a lot

[09:08:02.0000] <hsivonen>
for the generic facet:

[09:09:00.0000] <hsivonen>
html, json, xml, xhtml, gnu, text

[09:09:01.0000] <hsivonen>
looks like people who use the html5 facet don't love XML

[09:09:02.0000] <hsivonen>
but XML seems worth keeping because it's used on the generic side

[09:10:00.0000] <hsivonen>
my sed skills, suck. otherwise, I'd calculate some per-unique-IP numbers

[09:11:00.0000] <hsivonen>
s/skills,/skills/

[09:12:00.0000] <hsivonen>
/me tries perl instead

[09:18:00.0000] <hsivonen>
unique IP order for generic: html, xml, xhtml, gnu, json, text

[09:19:00.0000] <hsivonen>
unique IP order for HTML5: html, text, gnu, json, xml, xhtml

[09:22:00.0000] <Lachy>
watching the opening ceremony, I really don't understand why all the countries came out in a seemingly random order, with Australia finally coming out second last.

[09:23:00.0000] <hsivonen>
interestingly, the HTML5 facet is used more than the generic facet from browsers, but the generic facet is more popular than the HTML5 facet as a Web service

[09:23:01.0000] <hsivonen>
in terms of unique client IPs

[09:23:02.0000] <Lachy>
at first, it was on in chinese before switching to a norwegian stream, so I had no chance of undetstanding anything.

[09:28:00.0000] <gDashiva>
hsivonen: Maybe people don't like writing long URLs in their code :)

[09:28:01.0000] <billmason>
Lachy: opening ceremony marching order system -- http://en.wikipedia.org/wiki/2008_Summer_Olympics_Opening_Ceremony#Marching_order

[09:32:00.0000] <annevk>
anyone tips for best compact camera of the year?

[09:32:01.0000] <annevk>
ideally it fits in my pocket, has > 10M pixels, has > 5 optical zoom, and bonus points for GPS

[09:33:00.0000] <takkaria>
what price range are you looking at?

[09:34:00.0000] <hsivonen>
annevk: if there's any compact camera other than phones with built-in GPS, I want to know, too

[09:34:01.0000] <annevk>
takkaria, no price range as of yet :)

[09:35:00.0000] <hsivonen>
hmm. this Wikipedia Ogg thing is going to be huge when Firefox 3.1 ships

[09:35:01.0000] <annevk>
I guess my maximum for a compact would be EUR 500-1000 but that seems quite excessive for such a thing

[09:36:00.0000] <takkaria>
annevk: the canon g9 is considered one of the best compacts, though no GPS

[09:38:00.0000] <Philip`>
hsivonen: I can't remember ever having read a Wikipedia article with video in it, so I'm not sure how many people will ever notice the video support

[09:38:01.0000] <annevk>
takkaria, interesting, I saw that one in store today

[09:38:02.0000] <annevk>
hsivonen, I just found http://www.dpreview.com/news/0808/08080702nikonp6000.asp

[09:39:00.0000] <hsivonen>
annevk: thanks

[09:39:01.0000] <annevk>
from yesterday apparently :)

[09:39:02.0000] <hsivonen>
/me wonders if camera GPS startup sucks as badly as car navigator GPS startup

[09:40:00.0000] <annevk>
I believe all GPS sucks badly :(

[09:40:01.0000] <annevk>
at least, the devices I've seen so far did when trying to find out the initial position

[09:40:02.0000] <takkaria>
hmm, I'd wait for a review to appear on dpreview before buying one-- I don't believe nikon have been leading the compact market at all in recent years

[09:41:00.0000] <annevk>
takkaria, so the main problem with the G9 is that it's not all that compact

[09:41:01.0000] <takkaria>
ah.  I'm a DSLR person, so lots of things are compact for me. :)

[09:42:00.0000] <annevk>
heh

[09:43:00.0000] <annevk>
the G9 comes with 32MiB by default

[09:43:01.0000] <annevk>
/me sighs

[09:43:02.0000] <Philip`>
/me sees that news.bbc.co.uk has (Flash) video on its front page, which is the first time he's ever seen that

[09:43:03.0000] <hsivonen>
it sucks environmentally that stuff ships with throw-away memory

[09:44:00.0000] <takkaria>
Philip`: oo, fun

[09:44:01.0000] <annevk>
hsivonen, good point

[09:44:02.0000] <Philip`>
Live video, even

[09:44:03.0000] <hsivonen>
last month, I my opportunity cost and accounting cost for legally disposing of useless MacBook RAM far exceeded the price I was able to sell it for

[09:46:00.0000] <Lachy>
wow, 15 strokes just to write the the first letter of Australia in chinese is insanely complicated.

[09:46:01.0000] <annevk>
the store owner told me http://www.dpreview.com/reviews/specs/Panasonic/panasonic_dmctz5.asp was pretty good

[09:46:02.0000] <hsivonen>
s/I my/my/

[09:47:00.0000] <hsivonen>
my text input really sucks these days

[09:47:01.0000] <takkaria>
annevk: ah, I have a friend with one of the Lumix line, and it does seem to be pretty good.  has fairly decent low-light performance if I remember right

[09:51:00.0000] <annevk>
ok

[09:57:00.0000] <annevk>
http://www.dpreview.com/reviews/specs/Fujifilm/fujifilm_F100fd.asp also looks reasonable (successor to my current camera it seems)

[10:03:00.0000] <hendry>
whoa that opening ceremony was amazing

[10:09:00.0000] <Philip`>
That flame has got to be environmentally unfriendly

[10:17:00.0000] <takkaria>
hmm, google's cache pages have changed their look

[12:36:00.0000] <jgraham>
annevk: You probably don't want >10 MP; the laws of physics say it gives worse picture quality on compacts than smaller numbers of MP

[12:37:00.0000] <jgraham>
(but is still useful for marketing)

[12:42:00.0000] <Philip`>
Most cameras seem to let you select the image size, like choosing 640x480 or 1280x960 or 4000x3000 or whatever they do. How does that work?

[12:42:01.0000] <Philip`>
like, do they only use a small fraction of the CCD or whatever it's called?

[12:43:00.0000] <Philip`>
or do they do some clever combination of adjacent pixel detectors to get more detail in those fewer output pixels?

[12:43:01.0000] <Philip`>
or do they save a maximum-resolution image as JPEG and then resize it before storing it, or something?

[12:44:00.0000] <jgraham>
I guess they resize before storing but I don't actually know (I'm pretty sure they don't use less of the CCD)

[12:44:01.0000] <Hixie>
is mikesmith back from holiday yet?

[13:12:00.0000] <Hixie>
the alt text debate no longer seems to include any accessibility "experts"

[13:28:00.0000] <Philip`>
Now it's just people who don't even pretend to know what they're talking about :-)

[13:32:00.0000] <hsivonen>
/me thinks Raman knows what he is talking about

[14:23:00.0000] <Philip`>
hsivonen: Well, at least I don't know what I'm talking about :-)

[15:04:00.0000] <annevk>
jgraham, so my 5MP might be ok?

[15:04:01.0000] <annevk>
why is buying something so simple so tricky :/

[15:11:00.0000] <hsivonen>
annevk: I thought buying one would mean just buying the latest Ixus, but there's a choice even between Ixus models

[15:12:00.0000] <hsivonen>
clearly, it would all be easier if Steve offered us just one camera

[15:13:00.0000] <annevk>
true, that would help

[15:13:01.0000] <jgraham>
annevk: There are other reaons it might not be OK but adding more pixels won't make much difference 

[15:13:02.0000] <annevk>
though I heard the iPhone camera sucks

[15:13:03.0000] <annevk>
jgraham, ok, grmbl

[15:15:00.0000] <jgraham>
annevk: If I were shopping for a compact camera I'd look for a) good low light performance b) image stabilisation of some sort or another c) reasonable optical zoom (bearing in mind that wide angles are just as useful as close ups) 

[15:15:01.0000] <jgraham>
And good reviews of image quality

[15:17:00.0000] <jgraham>
(in I would also look at price, non proprietry storage format, ergonomics and perhaps battery life)

[15:17:01.0000] <jgraham>
s/in//

[15:17:02.0000] <hsivonen>
annevk: http://luminous-landscape.com/whatsnew/#264

[15:19:00.0000] <annevk>
thanks jgraham, hsivonen

[15:19:01.0000] <annevk>
jgraham, any camera in mind btw? all sounds good to me :)

[15:19:02.0000] <jgraham>
I guess RAW is nice to hve but it doen't seem essential in a compact (like if you care enough to spend time postprocessing the photos you probably care enough to lug a DSLR about)

[15:20:00.0000] <annevk>
I don't really want to do postprocessing other than adding tags

[15:22:00.0000] <jgraham>
annevk: Not really. We bought a Canon A710IS for my g/f  about 18 months ago mainly because it was the cheapest camera with image stabalisation. It has been pretty good but I guess not perfect

[15:24:00.0000] <annevk>
does it perform well with low light?

[15:25:00.0000] <jgraham>
OK but not spectacular. IS helps but I think some of the other people have CCDs that work better undr those conditions. I guess if you check dpreview it will give some clues

[15:26:00.0000] <annevk>
hmm, the Sony models come with 4GiB of internal memory

[15:29:00.0000] <jgraham>
Do Sony still use their proprietry memory stick things?

[15:32:00.0000] <annevk>
I believe so, yes

[15:57:00.0000] <annevk>
hsivonen, reading that site, sounds bad :/

[15:59:00.0000] <annevk>
I might try to buy http://www.dpreview.com/reviews/panasonictz5/ as it seems reasonably good, except at high ISO values but apparently all compacts aren't very good in that area

[15:59:01.0000] <annevk>
excluding extras it's about ~350EUR which seems acceptable

[16:55:00.0000] <annevk>
hsivonen, http://www.w3.org/QA/2008/08/markup_validator_updated.html

[16:58:00.0000] <annevk>
hmm, they removed the XML media type warning for mislabeled XHTML


2008-08-09
[17:33:00.0000] <Philip`>
Hmm, I need to go on holiday instead of continuing the alt discussion - what a dreadful shame :-(

[17:39:00.0000] <Philip`>
http://www.bbc.co.uk/iplayer/ - <meta name="DCTERMS.modified" content="2076-02-12T12:00:00Z" /> - yay for metadata

[17:43:00.0000] <Philip`>
(And DCTERMS.modified doesn't even exist, according to the declared schema)

[17:50:00.0000] <eseidel>
/me wonders if anyone is interested in a theoretical discussion about <canvas> patterns? :)

[17:50:01.0000] <eseidel>
http://www.whatwg.org/specs/web-apps/current-work/#createpattern

[17:51:00.0000] <eseidel>
Safari's current behavior is slightly differnet from the spec, in that if an image is still loading off the network when createPattern is called with an HTMLImageElement parameter, the pattern will draw the image when it's finally loaded

[17:51:01.0000] <eseidel>
it will even draw it in incremental states :)

[17:51:02.0000] <eseidel>
/me should really email the list

[18:00:00.0000] <Philip`>
eseidel: That sounds unpleasantly unpredictable

[18:00:01.0000] <eseidel>
Philip`: yeah, I'm gonna change iet to make a copy

[18:00:02.0000] <eseidel>
Philip`, Hixie: but i sent a note to the whatwg list all the same

[18:00:03.0000] <eseidel>
this will now force web authors to listen for image load evnets

[18:00:04.0000] <eseidel>
which they might be doing anyway

[18:00:05.0000] <Philip`>
I think the spec just says it should throw an exception on createPattern if it's not complete

[18:01:00.0000] <eseidel>
If the image argument is an HTMLImageElement object whose complete attribute is false, then the implementation must raise an INVALID_STATE_ERR exception.

[18:01:01.0000] <eseidel>
oh

[18:01:02.0000] <eseidel>
it does

[18:01:03.0000] <eseidel>
nm then

[18:01:04.0000] <Philip`>
"If the image argument is an HTMLImageElement object whose complete attribute is false, then the implementation must raise an INVALID_STATE_ERR exception."

[18:04:00.0000] <Philip`>
http://philip.html5.org/tests/canvas/suite/tests/results.html#2d.pattern.image.incomplete possibly isn't entirely helpful since it assumes new Image().complete == false (since the spec says so), which is untrue in reality

[01:43:00.0000] <heycam>
hsivonen, yt?

[02:30:00.0000] <hsivonen>
heycam: I am now.

[03:25:00.0000] <heycam>
hsivonen, was gonna ask: how come http://s.validator.nu/xhtml10/text.rnc removes xml:space from <pre>?

[03:26:00.0000] <heycam>
/me afk again

[03:31:00.0000] <hsivonen>
heycam: hmm. must be a bug

[03:33:00.0000] <hsivonen>
heycam: It's now a common attribute, isn't it?

[03:34:00.0000] <hsivonen>
hmm. it's not a common attribute. sigh

[03:50:00.0000] <jacobolus>
annevk: i'm a bit late here, but if you want a compact camera w/ a small lens, there is no way 10 Mpx will actually be useful (the extra data will be some combination of noise and blur)

[03:55:00.0000] <mcarter>
do any of the opera SSE implementations have a way to set the retry interval, like the current spec does?

[04:01:00.0000] <annevk>
mcarter, no

[04:01:01.0000] <annevk>
jacobolus, ok, thanks

[04:01:02.0000] <annevk>
jacobolus, I will most likely buy this model: http://www.dpreview.com/reviews/specs/Panasonic/panasonic_dmctz5.asp

[04:02:00.0000] <mcarter>
annevk, does it (the application/x-dom-event-stream SSE) auto-reconnect when a response completes?

[04:03:00.0000] <annevk>
mcarter, it should, yes

[04:03:01.0000] <mcarter>
annevk, i don't suppose you recall if it does so immediately or not?

[04:03:02.0000] <jacobolus>
mcarter: I don't think so

[04:03:03.0000] <jacobolus>
mcarter: try it! :)

[04:04:00.0000] <annevk>
mcarter, what jacobolus said

[04:08:00.0000] <mcarter>
so it seems to reconnect automatically after a 5 second delay

[04:09:00.0000] <jacobolus>
really? did it do that in 8.x? huh.

[04:10:00.0000] <mcarter>
annevk, do you recall if there's a way to attach a trigger for when the stream completes, if i wanted to, for instance, remove the stream when the response completes?

[04:13:00.0000] <hsivonen>
heycam: hmm. My local copy works. For some reason, the deployment copy isn't pulling stuff from svn correctly or something, so you see a partial & broken "fix"

[04:13:01.0000] <hsivonen>
heycam: I'll investigate

[04:23:00.0000] <hsivonen>
heycam: fix deployed

[04:23:01.0000] <annevk>
mcarter, maybe use a special event that removes the stream?

[04:24:00.0000] <mcarter>
annevk, yeah, that works. But in the case of a network breakage it falls apart

[04:24:01.0000] <annevk>
mcarter, as in, the page registers a special listener for the "removestream" event and when it gets that from the server it kills the connection

[04:24:02.0000] <annevk>
mcarter, true

[04:24:03.0000] <hsivonen>
heycam: thanks

[06:37:00.0000] <hsivonen>
we should have a standard form for explaining why Web spec feature suggestions don't work like the Slashdot spam form (http://tech.slashdot.org/comments.pl?sid=623883&cid=24314727)

[08:05:00.0000] <heycam>
hsivonen, thanks

[09:54:00.0000] <takkaria>
I wonder if IE's authoratitive=true MIME header will disable sniffing on the charset too

[15:44:00.0000] <Hixie>
hsivonen: that form is pretty close already

[16:10:00.0000] <jcranmer>
how does the W3C moderation system work? First post is sent to moderator, and, if accepted, all subsequent posts go through?

[16:58:00.0000] <Lachy>
jcranmer, no. For their first post, they're sent a confirmation email asking them to manually agree for it and all subsequent mails to be archived. They probably have some sort of spam filter in place too. They do have list moderators, but most things are automated


2008-08-10
[17:33:00.0000] <Hixie>
the arguments saying that latex is more useful than anything else are realy arguing that it isn't alternative text but is better than the image itself

[17:41:00.0000] <Lachy>
Hixie, I don't think so.

[17:43:00.0000] <jcranmer>
I think they're working along the lines that "anyone who knows anything more complex than exponentiation knows (La)TeX", something which seems dubious to me

[17:44:00.0000] <jcranmer>
ideally, it could be read aloud, but conveying

[17:46:00.0000] <jcranmer>
\lim_{n \to \inf} \left[ \sum_{k=0}^\inf \frac{n^k}{\left(k+1\right)^{k+1}} \right]^{\frac{1}{n}}

[17:46:01.0000] <jcranmer>
is probably not fun

[17:47:00.0000] <jcranmer>
"The limit as n goes to infinity of the sum from k equals 0 to infinity of n to the k over k plus 1 to the k + 1 power all to the 1 over n power"

[17:54:00.0000] <Lachy>
both of those are complicated. That latter is even a little ambiguous in some parts

[17:56:00.0000] <Lachy>
e.g. does "k plus 1 to the k + 1 power" mean: k + 1^(k+1) or  (k + 1)^(k+1)

[17:57:00.0000] <jcranmer>
that's the problem with spoken stuff: it's hard to get parenthetics rights

[17:57:01.0000] <Hixie>
if i was reading those, the former would be "k plus one to the k plus oneth power", the latter would be "k plus one, all raised to the k plus oneth power".

[17:58:00.0000] <Hixie>
jcranmer: that's a problem with english, not just spoken word

[17:58:01.0000] <Lachy>
so we should all start speaking in TeX instead :-)

[17:59:00.0000] <jgraham>
FWIW T.V. Ramen's LaTeX->speech translator is supposed to set the inonation and pauses to effectively convey the meaning. You can't do that with pure english

[17:59:01.0000] <Hixie>
those could also be read as "k plus one to the open bracket k plus one close bracket" and "open breack k plus one close breacket to the open breack k plus one close bracket"

[17:59:02.0000] <Hixie>
wow i can't spell bracket

[17:59:03.0000] <Hixie>
wtf

[17:59:04.0000] <jgraham>
s/inonation/intonation/

[18:00:00.0000] <jgraham>
s/pure english/pure english conveyed as ascii text/

[18:00:01.0000] <Hixie>
you know, if the fallback is going to be some format language, maybe <object data=eq.png><object data=eq.latex>...</object></object> is a better solution

[18:00:02.0000] <Lachy>
you can't spell "break" either (unless you meant bracket there too)

[18:01:00.0000] <Hixie>
i meant bracket each time

[18:01:01.0000] <Lachy>
ok

[18:01:02.0000] <Hixie>
i really can't type it apparently :-)

[18:01:03.0000] <jcranmer>
:-)

[18:01:04.0000] <jcranmer>
+1 on the nesting object idea

[18:02:00.0000] <jgraham>
I don't see why this is a big problem. Latex is an ascii (or at least text-based) equivalent of a formula. When that formula is contained in an image the LaTeX is therefore an equivalent of that image

[18:02:01.0000] <Lachy>
why don't we just add a way to use latex natively in HTML?

[18:03:00.0000] <jgraham>
Lachy: Because that would be a big mess

[18:03:01.0000] <jcranmer>
I thought that was called MathML? ;-)

[18:03:02.0000] <jcranmer>
/me knows the two aren't really alike

[18:03:03.0000] <Hixie>
hat jcranmer said

[18:03:04.0000] <jgraham>
However I already suggested that we add a TeX-like syntax that maps to MathML

[18:03:05.0000] <Hixie>
mathml is our solution for math

[18:03:06.0000] <Lachy>
most mathematicians seem to use latex though anyway

[18:03:07.0000] <Hixie>
we already have a syntax

[18:03:08.0000] <jgraham>
Hixie rejected it

[18:04:00.0000] <jgraham>
Lachy: This is because TeX doesn't suck and MathML does

[18:04:01.0000] <jcranmer>
Lachy: there are numerous LaTeX->MathML converters

[18:04:02.0000] <jgraham>
(well and also because TeX has meaningful deployment)

[18:04:03.0000] <jcranmer>
and numerous MathML->TeX converters

[18:04:04.0000] <Lachy>
so why was mathml created in the first place, instead of just defining how to add latex to HTML and other XML languages?

[18:05:00.0000] <jgraham>
jcranmer: There are numerous (*subset of* LaTeX)->(MathML) convertors

[18:05:01.0000] <jcranmer>
I presume if you stick a lot of \kern + co in there, it might not translate

[18:06:00.0000] <jgraham>
Lachy: Well you should ask David Carslile (sp?) but I think a) people wanted to get rid of some TeX warts b) people thought they needed semantics c) XML is cool

[18:07:00.0000] <jgraham>
jcranmer: Or anything that relies on something you \def yourself

[18:07:01.0000] <jcranmer>
jgraham: a smart converter should be able to de-macro \def's

[18:08:00.0000] <jgraham>
jcranmer: By "smart convertor" you mean one that actually runs the TeX right?

[18:09:00.0000] <jcranmer>
that is ideal

[18:10:00.0000] <jgraham>
I think you would still have some difficulty because to do a good job at creating MathML you need to retain some element of what the user typed

[18:10:01.0000] <jgraham>
(to have a shot at the semantics)

[18:12:00.0000] <jgraham>
even then getting things like operator vs identifier right is hard

[18:13:00.0000] <jgraham>
it seems much more sensible to define a reduced subset of LaTeX that maps well onto MathML in the same way PDF is a subset of postscript

[18:22:00.0000] <jcranmer>
quick, OT question: if you were to create a utility that reposts spamified messages from a newsgroup of certain users, would you expect that the tools you use are able to properly support =?utf-8?Q?...?= in MIME headers?

[03:08:00.0000] <hsivonen>
hmm. people want a category for markp's posts as if there were too many others posts to clutter their feed readers

[03:29:00.0000] <Lachy>
hsivonen, yeah, I thought it was a bit odd too. But it's not hard to do

[05:19:00.0000] <Lachy>
Hixie, I contest your claim that "the main characters [or Stargate] aren't immortal".  IIRC, Daniel has died 4 times, Jack, Carter and Teal'c have died once, and yet each time they have been resurrected. And this doesn't include alternate realities and timelines. :-)

[05:20:00.0000] <Lachy>
s/[or/[of/

[07:14:00.0000] <jacobolus>
hsivonen, Lachy: you could alternately make a thing at Yahoo Pipes that filters out posts other than MarkP's

[07:30:00.0000] <jacobolus>
e.g. http://pipes.yahoo.com/pipes/pipe.run?_id=47fd026d4015483a48d8cd5656c0636e&_render=rss

[07:37:00.0000] <jacobolus>
better: http://pipes.yahoo.com/jrus/this_week_in_html5?_render=rss

[09:32:00.0000] <hsivonen>
/me guesses that someone will implement a pure-JS DOM for Web Worker XHR

[09:56:00.0000] <jgraham>
hsivonen: On the podcast thing he id recently John Reisig was talking about something like having access to insulated DOM-fragments in workers and a browser-level protocol for passing those fragments through channels to the main thread

[09:56:01.0000] <jgraham>
Sorry too many i's in Resig there...

[09:57:00.0000] <takkaria>
jgraham: there's some code to bind hubbub to libxml2 now; I don't know quite how much use that will be to you, but thought you would be interested

[10:00:00.0000] <Lachy>
jacobolus, what rules did you use to make that?

[10:01:00.0000] <Lachy>
jacobolus, btw, I suspect making a category would be easier. I just need to know what to call it

[10:01:01.0000] <jgraham>
takkaria: That sounds awesome. I will have a look when I have a spare moment (which may not be for some time...)

[10:02:00.0000] <takkaria>
jgraham: when you do have a spare moment, give me a yell.  I'll try and sort it out so the code is somewhere public before then. :)

[10:04:00.0000] <Lachy>
jacobolus, I found the yahoo pipe source. Filtering on the name is only reliable if markp is the only one who ever writes them

[11:40:00.0000] <jacobolus>
Lachy: it could be filtered on anything else, if you like, even on a regexp or something

[11:40:01.0000] <jacobolus>
Lachy: but you're right that a category would be easiest—I just don't have control over that :)

[11:54:00.0000] <Lachy>
jacobolus, I do. I just need to know what to call the category

[11:55:00.0000] <Lachy>
any suggestions? maybe weekly-review

[11:55:01.0000] <jacobolus>
recent-highlights?

[11:55:02.0000] <jacobolus>
weekly-review sounds fine

[11:55:03.0000] <jacobolus>
if you can actually keep it weekly ;)

[13:20:00.0000] <hsivonen>
zcorpan: Mail::Mailer brokenness fixed. thanks.

[13:39:00.0000] <hsivonen>
oops Mail::Mailer still broken. sigh.

[13:48:00.0000] <hsivonen>
where might CPAN break stuff? I remove perl stuff from under /usr/local, but Mail::Mailer still fails due to a tainted variable in the -T mode

[13:48:01.0000] <hsivonen>
and the Mail::Mailer should be provided by Ubuntu

[13:48:02.0000] <hsivonen>
hmm

[14:08:00.0000] <Hixie>
Lachy: ford.

[14:09:00.0000] <Hixie>
Lachy: also, they can't very well be immortal if they died!

[14:09:01.0000] <hsivonen>
/me regrets invoking CPAN

[14:10:00.0000] <hsivonen>
I have no idea how to debug why a variable is tainted in Bugzilla after CPAN has done something

[14:12:00.0000] <Hixie>
wow, smylers' e-mail was pretty much spot on as far as my current thinking is concerned :-)

[14:14:00.0000] <Lachy>
Hixie, what do you mean by ford?

[14:14:01.0000] <Hixie>
he died.

[14:15:00.0000] <Hixie>
well, one assumes he died, anyway.

[14:15:01.0000] <Lachy>
oh

[14:15:02.0000] <Hixie>
one minute he was on a ship, then next minute the ship was in a bazillion pieces, and we never heard from him again.

[14:36:00.0000] <annevk>
Lachy, I think the blog should have tags

[14:36:01.0000] <annevk>
Lachy, and we should get rid of the categories

[14:41:00.0000] <Lachy>
annevk, tags are already supported. We just need to expose them in the template

[14:42:00.0000] <annevk>
with "the blog should have tags" I mean that existing entries should have tags added and that they (obviously) should be exposed through the Web interface :)

[14:43:00.0000] <annevk>
then we can just make up some weird tag for markp's stuff and people can rejoice

[14:43:01.0000] <annevk>
(though given the low amount of posts I wonder why they just want to follow those, especially given only one appeared so far)

[14:44:00.0000] <Hixie>
does outerHTML have any stupid quirks anyone knows about?

[14:44:01.0000] <annevk>
in two out of three browsers supporting it stuff is in uppercase?

[14:45:00.0000] <annevk>
(not sure about WebKit)

[14:45:01.0000] <Hixie>
i meant for setting more than getting

[14:47:00.0000] <Hixie>
though thanks for the getting info

[14:54:00.0000] <hsivonen>
/me feels powerless in face of perl -T

[14:55:00.0000] <Hixie>
do you need to untaint something forcibly?

[14:55:01.0000] <Hixie>
or are you trying to work out where the taint started, so you can stop that?

[14:55:02.0000] <hsivonen>
Hixie: the latter

[14:56:00.0000] <hsivonen>
specifically, if my bugzilla installation worked before touching cpan, why does it stay broken after I delete all perl stuff under /usr/local ?

[14:57:00.0000] <hsivonen>
now every time Bugzilla tries to send email, the message is considered tainted when it's passed to exec for sendmail

[14:57:01.0000] <Hixie>
can you work out which part of the message gets considered tainted?

[14:57:02.0000] <Hixie>
then again, that sounds about right, i mean, that data is coming from the database right?

[14:57:03.0000] <Hixie>
so it should be tainted

[14:58:00.0000] <hsivonen>
Hixie: I don't know how to work it out with reasonable effort.

[14:58:01.0000] <hsivonen>
Hixie: sure, but presumably before I touched cpan, something untainted the message properly

[15:00:00.0000] <Hixie>
what version are you on?

[15:00:01.0000] <hsivonen>
2.22.1-debian2, so I can't get help from bugzilla hackers

[15:01:00.0000] <Hixie>
and what's the error message exactly?

[15:01:01.0000] <hsivonen>
undef error - Insecure dependency in exec while running with -T switch at /usr/share/perl5/Mail/Mailer/sendmail.pm line 22. 

[15:01:02.0000] <hsivonen>
line 22 is exec( $exe, '-t', @$args );

[15:02:00.0000] <hsivonen>
and I already checked that $exe is not the problem

[15:02:01.0000] <hsivonen>
so something in $args is tainted

[15:03:00.0000] <Hixie>
are you sure $ENV{PATH} is cleared?

[15:03:01.0000] <hsivonen>
Hixie: I'm not. How do I find out?

[15:03:02.0000] <Hixie>
stick $ENV{PATH} = ''; somewhere before the exec call

[15:06:00.0000] <hsivonen>
Hixie: that doesn't solve the problem

[15:06:01.0000] <Hixie>
ok just before the exec() call try this:

[15:07:00.0000] <Hixie>
use Scalar::Util; warn("exe:" . tainted($exe) . "; args: " . tainted("@$args"));

[15:07:01.0000] <Hixie>
see if either of those returns true

[15:08:00.0000] <hsivonen>
where do the warnings end up?

[15:08:01.0000] <Hixie>
(if you're on an old perl (pre 5.8) you can define sub tainted() { return ! eval { eval("#" . substr(join("", @_), 0, 0)); 1 } } )

[15:08:02.0000] <Hixie>
stderr

[15:08:03.0000] <Hixie>
apache error log probably

[15:08:04.0000] <Hixie>
failing that, print it to some known location

[15:09:00.0000] <Hixie>
it should go to the same place as you see the insecure dependency message

[15:11:00.0000] <hsivonen>
undef error - Undefined subroutine &Mail::Mailer::sendmail::tainted called at /usr/share/perl5/Mail/Mailer/sendmail.pm line 23. 

[15:11:01.0000] <hsivonen>
and I'm not on old perl

[15:11:02.0000] <Hixie>
sorry make that use Scalar::Util qw(tainted);

[15:12:00.0000] <hsivonen>
process_bug.cgi: exe:0; args: 1 at /usr/share/perl5/Mail/Mailer/sendmail.pm line 22.

[15:13:00.0000] <Hixie>
ok, so $args is being tainted somehow

[15:14:00.0000] <Hixie>
use Scalar::Util qw(tainted); foreach (@$args) { warn tainted($_) }   will tell you which item in $args

[15:15:00.0000] <hsivonen>
0 and 1

[15:15:01.0000] <hsivonen>
wouldn't that be to and from?

[15:15:02.0000] <Hixie>
no idea

[15:15:03.0000] <Hixie>
but you can probably look in the code to see

[15:16:00.0000] <hsivonen>
Hixie: thanks

[15:16:01.0000] <Hixie>
np

[15:19:00.0000] <zcorpan>
is whatever feature we come up with to distinguish images-with-proper-replacement to images without going to be used correctly enough that UAs will do something special with it?

[15:19:01.0000] <zcorpan>
why can't we just allow alt="photo" and be done with it?

[15:20:00.0000] <zcorpan>
(though when i think about it, opera does something special with lack of alt -- it shows "Image")

[15:23:00.0000] <Hixie>
i have no diea

[15:23:01.0000] <Hixie>
idea

[15:23:02.0000] <Hixie>
it's somewhat up to the accessibility "experts" at this point

[15:23:03.0000] <Hixie>
if it isn't implemented, then it'll get dropped at CR

[15:23:04.0000] <Hixie>
and we'll be back to optional alt

[15:24:00.0000] <zcorpan>
the old spec text didn't allow the kind of image to be specified as alt

[15:25:00.0000] <Hixie>
indeed, that would be quite bad for accessibility

[15:26:00.0000] <zcorpan>
why?

[15:26:01.0000] <Hixie>
you couldn't distinguish images that were photos from images that represented icons of photos

[15:26:02.0000] <Hixie>
makes for a pretty confusing ui

[15:27:00.0000] <zcorpan>
i think you could figure that out based on context

[15:27:01.0000] <Hixie>
*shrug*

[15:27:02.0000] <Hixie>
can anyone think of any way to get a hold of any of the nodes that are created in  foo.outerHTML = '...'  if 'foo' has no parent node?

[15:28:00.0000] <zcorpan>
implicit document fragment node? not sure i understand the question

[15:30:00.0000] <Hixie>
if i do:   <script> var d = document.createElement('div'); d.outerHTML = '<form id=a>'; </script> how can i get hold of the form element?

[15:30:01.0000] <Hixie>
there's no way, right?

[15:32:00.0000] <annevk>
/me can't say no for certain

[15:36:00.0000] <zcorpan>
seems so... so outerHTML is basically a no-op in that case

[15:36:01.0000] <Hixie>
yeah

[15:36:02.0000] <Hixie>
good

[15:37:00.0000] <Hixie>
that way i don't have to worry about what its parentNode is to get a context node...

[15:38:00.0000] <annevk>
in opera if you do d.outerHTML afterwards it says "<form id=a>"

[15:38:01.0000] <annevk>
/me has no idea what that means

[15:38:02.0000] <Hixie>
lol

[15:38:03.0000] <Hixie>
that's funny

[15:39:00.0000] <annevk>
(it also says d is an HTMLDivElement object)

[15:39:01.0000] <Hixie>
maybe it's setting the actual JS property and not the DOM attribute

[15:39:02.0000] <Hixie>
try setting it to something that's not HTML syntax

[15:39:03.0000] <Hixie>
and see if it fixes it up

[15:39:04.0000] <Hixie>
or if it just allowed anything

[15:39:05.0000] <Hixie>
e.g. <fOrM

[15:39:06.0000] <Hixie>
does it become <form> ?

[15:40:00.0000] <annevk>
no

[15:40:01.0000] <annevk>
"x" stays "x"

[15:40:02.0000] <Hixie>
then it's not really outerHTML

[15:40:03.0000] <annevk>
"<x" stays "<x"

[15:41:00.0000] <annevk>
setting outerHTML in other cases does work

[15:41:01.0000] <annevk>
confusing

[15:42:00.0000] <zcorpan>
in ie it's <DIV></DIV>

[15:43:00.0000] <annevk>
parentNode == null probably makes browsers invoke special codepaths

[15:43:01.0000] <annevk>
so outerHTML is going to make it in after all?

[15:43:02.0000] <Hixie>
everyone except FF implements it

[15:43:03.0000] <Hixie>
might as well

[15:44:00.0000] <Hixie>
i'm not doing it for XML though

[15:44:01.0000] <annevk>
you know you use such arguments the other way around as well, right?

[15:44:02.0000] <annevk>
:)

[15:44:03.0000] <Hixie>
depends how hard it is to spec :-P

[15:45:00.0000] <Hixie>
bigger problem is what to do when the parent is a Document object

[15:45:01.0000] <zcorpan>
webkit throws when there's no parent it seems

[15:46:00.0000] <annevk>
per some old testcase here http://tinymce.moxiecode.com/opera/documentelement_outerhtml.htm it should do something

[15:47:00.0000] <annevk>
maybe the same as document.innerHTML

[15:47:01.0000] <Hixie>
no, that would blow away comments and doctypes

[15:47:02.0000] <zcorpan>
webkit and ie throw when parent is document

[15:47:03.0000] <Hixie>
sounds good to me!

[15:49:00.0000] <annevk>
zcorpan, they don't show anything in the iframe for that test?

[15:49:01.0000] <Hixie>
btw html5 has a dependency on xhr now :-)

[15:50:00.0000] <Hixie>
(well, web workers, technically)

[15:50:01.0000] <annevk>
/me finds http://en.wikipedia.org/wiki/Ian_Hickson

[15:50:02.0000] <Hixie>
haha sweet, i actually have a wikipedia page now

[15:51:00.0000] <Hixie>
awww, no controversy in the talk page

[15:51:01.0000] <Hixie>
how sad

[15:51:02.0000] <zcorpan>
annevk: right

[15:52:00.0000] <Hixie>
i guess the right exception is INVALID_ACCESS_ERR here

[15:52:01.0000] <Hixie>
hough NO_MODIFICATION_ALLOWED_ERR, NOT_SUPPORTED_ERR and HIERARCHY_REQUEST_ERR are all tempting

[15:52:02.0000] <zcorpan>
log: Error: NO_MODIFICATION_ALLOWED_ERR: DOM Exception 7

[15:52:03.0000] <zcorpan>
webkit

[15:52:04.0000] <Hixie>
sounds good to me!

[15:53:00.0000] <Hixie>
i never get to throw that one

[15:53:01.0000] <zcorpan>
hah

[15:53:02.0000] <Hixie>
does IE throw for no prent node too?

[15:53:03.0000] <Hixie>
parent

[15:53:04.0000] <zcorpan>
no

[15:53:05.0000] <Hixie>
ok

[15:54:00.0000] <zcorpan>
neither does opera

[15:55:00.0000] <annevk>
is HIERARCHY not better? though maybe I don't care

[15:56:00.0000] <zcorpan>
fwiw, ie throws if you try to set outerHTML on head or body too

[15:56:01.0000] <zcorpan>
or innerHTML on the root

[15:56:02.0000] <zcorpan>
but we don't align with ie very much for innerHTML

[15:58:00.0000] <zcorpan>
/me tests documentfragment

[16:01:00.0000] <zcorpan>
opera does the same thing as it does for document

[16:01:01.0000] <zcorpan>
webkit throws

[16:02:00.0000] <zcorpan>
ie does the same thing as it does for a normal element parent

[16:02:01.0000] <zcorpan>
/me likes the ie approach

[16:02:02.0000] <annevk>
I hope I can make fixes and such to XHR / AC / Media Queries tomorrow

[16:03:00.0000] <Hixie>
don't break my dependency! :-P

[16:03:01.0000] <annevk>
I'll be off for the rest of the week and next week is the CSS WG meeting...

[16:03:02.0000] <annevk>
summer is slow anyways I suppose

[16:03:03.0000] <hsivonen>
zcorpan: please let me know if you see the taint mode message with bugzilla.validator.nu.

[16:03:04.0000] <Hixie>
oh, let me know how the csswg meeting goes

[16:03:05.0000] <annevk>
/me looks for the depedency

[16:03:06.0000] <Hixie>
i use NETWORK_ERR

[16:03:07.0000] <jacobolus>
what does the CSS WG plan to do at their meeting?

[16:03:08.0000] <jacobolus>
maybe figure out hypenation? ;)

[16:04:00.0000] <jacobolus>
erm, hyphenation

[16:04:01.0000] <hsivonen>
the message shows up only on certain bugs but not on others. very weird.

[16:04:02.0000] <zcorpan>
hsivonen: sure, whenever i find another bug :)

[16:04:03.0000] <annevk>
Hixie, I was sort of intrigued by your hijacking of 19, should NETWORK_ERR & ABORT_ERR use 20 and 21 instead?

[16:04:04.0000] <Hixie>
sounds good to me

[16:04:05.0000] <Hixie>
i just used 19 because it was the first free one after 18

[16:05:00.0000] <annevk>
right

[16:05:01.0000] <Hixie>
and i hijacked 18 cos it was the first free one after 17

[16:05:02.0000] <Hixie>
i updated the wiki exception codes page

[16:05:03.0000] <jacobolus>
annevk: http://www.w3.org/TR/css3-text/#hyphenate has said since the beginning of march: “The WG plans to discuss it at our next face-to-face meeting at the end of March.”

[16:05:04.0000] <Hixie>
as far as i'm concerned, that's normative http://wiki.whatwg.org/wiki/Exception_Codes :-)

[16:05:05.0000] <annevk>
I see

[16:06:00.0000] <jacobolus>
(actually, since march 2007!)

[16:06:01.0000] <annevk>
jacobolus, heh, hyphenation is difficult and the editors are working on some other stuff too, iirc

[16:06:02.0000] <Hixie>
the csswg is so dysfunctional

[16:06:03.0000] <Hixie>
they try to do everything by consensus

[16:06:04.0000] <Hixie>
and the keep reopening old issues

[16:07:00.0000] <Hixie>
etc

[16:07:01.0000] <zcorpan>
why change the NETWORK_ERR and ABORT_ERR codes now?

[16:07:02.0000] <zcorpan>
(are they implemented?)

[16:07:03.0000] <Hixie>
if they are we shouldn't change them

[16:07:04.0000] <Hixie>
it's not a big deal

[16:07:05.0000] <jacobolus>
if hyphenation worked in browsers (or even just webkit/gecko), the web would be so so so much nicer

[16:07:06.0000] <jacobolus>
text could actually be justified

[16:07:07.0000] <jacobolus>
and not look like crap

[16:07:08.0000] <jacobolus>
s/be/look/

[16:07:09.0000] <annevk>
zcorpan, if WebKit ships with them we won't

[16:07:10.0000] <Hixie>
jacobolus: there are some issues that i would consider to be a much higher priority that are languishing as much :-)

[16:08:00.0000] <annevk>
zcorpan, if WebKit doesn't it seems better to change them for consistency as 101 and 102 are sort of out of the blue

[16:08:01.0000] <Hixie>
e.g. getting a decent box model for ui

[16:08:02.0000] <zcorpan>
annevk: L&S was before, no?

[16:08:03.0000] <annevk>
yeah, UI, transitions, gradients, etc., menus maybe

[16:08:04.0000] <annevk>
zcorpan, well, that spec should really be killed

[16:08:05.0000] <jacobolus>
Hixie: sure.  i'm self-interested, in that I want to be able to use Webkit for my own experiments w/ trying to make tools for my personal use for reading long documents on a screen.  but still :)

[16:08:06.0000] <zcorpan>
/me doesn't really care

[16:09:00.0000] <annevk>
zcorpan, nobody really cares, but we should try to minimize confusion :)

[16:17:00.0000] <annevk>
why won't outerHTML work for XML btw? there's no extra problems there, right?

[16:17:01.0000] <Hixie>
trying to keep the damage to a minimum

[16:17:02.0000] <annevk>
i guess you need to do the isHTML check anyway

[16:18:00.0000] <annevk>
btw, for documentElement.outerHTML, wouldn't documentElement.parentNode.replaceChild(...) do the trick?

[16:18:01.0000] <Hixie>
the fragment parsing algorithm doesn't handle the root element

[16:19:00.0000] <Hixie>
you'd have to have a whole other separate codepath for it

[16:19:01.0000] <zcorpan>
Hixie: what are you going to do with documentFragment?

[16:19:02.0000] <annevk>
Hixie, ok, fair enough

[16:19:03.0000] <Hixie>
zcorpan: ?

[16:20:00.0000] <zcorpan>
Hixie: when the parentNode is a document fragment

[16:20:01.0000] <Hixie>
oh

[16:21:00.0000] <Hixie>
uh

[16:21:01.0000] <Hixie>
i guess we pretend the document element is a <body>

[16:22:00.0000] <zcorpan>
<zcorpan> opera does the same thing as it does for document

[16:22:01.0000] <zcorpan>
<zcorpan> webkit throws

[16:22:02.0000] <zcorpan>
<zcorpan> ie does the same thing as it does for a normal element parent

[16:22:03.0000] <zcorpan>
* zcorpan likes the ie approach

[16:22:04.0000] <Hixie>
right

[16:23:00.0000] <Hixie>
so i guess we pretend the document element is a <body>

[16:23:01.0000] <Hixie>
er, s/document element/context element/

[16:23:02.0000] <Hixie>
matches IE

[16:23:03.0000] <Hixie>
right?

[16:23:04.0000] <zcorpan>
yep

[16:23:05.0000] <Hixie>
eeeeeexcellent

[16:23:06.0000] <Hixie>
good catch, i totally flaked on that and didn't even think to ask what you mean above

[16:24:00.0000] <Hixie>
meant

[16:28:00.0000] <Hixie>
/me next looks at insertAdjacentHTML() and shudders

[16:30:00.0000] <zcorpan>
Hixie: shouldn't execCommand('insertHTML') throw in an xml context?

[16:30:01.0000] <annevk>
Hixie, how about URL_MISMATCH_ERR becoming 21 and I get 19 and 20 so XHR can define 18-20?

[16:30:02.0000] <Hixie>
annevk: 18 is going into DOM Core eventually

[16:31:00.0000] <Hixie>
annevk: i thought webkit implemented 101 102?

[16:31:01.0000] <annevk>
Hixie, weinig says this is ok for WebKit

[16:31:02.0000] <weinig>
Hixie: I don't see it as a real compatibility issue

[16:31:03.0000] <Hixie>
well if you really want to change it then sure *shrug*

[16:31:04.0000] <annevk>
Hixie, I think NETWORK and ABORT can go into DOM Core as well eventually

[16:31:05.0000] <Hixie>
feel free to pick whatever numbers you want

[16:31:06.0000] <Hixie>
update the wiki and i'll adjust as necessary

[16:31:07.0000] <annevk>
ok

[16:31:08.0000] <Hixie>
zcorpan: maybe

[16:32:00.0000] <annevk>
/me was just about to offer to edit the wiki

[16:32:01.0000] <Hixie>
zcorpan: it doesn't use XML, though

[16:32:02.0000] <weinig>
Hixie: I will change it tonight and see if stuff breaks

[16:32:03.0000] <weinig>
call it a trial run

[16:32:04.0000] <Hixie>
k

[16:32:05.0000] <weinig>
/me implemented a dumb version of insertAdjacentHTML recently

[16:32:06.0000] <zcorpan>
Hixie: otherwise you can inject "malformed" stuff in xml that is so carefully avoided in the rest of the dom

[16:33:00.0000] <Hixie>
zcorpan: *shrug*

[16:33:01.0000] <Hixie>
zcorpan: think of it as an editing api

[16:33:02.0000] <Hixie>
zcorpan: where the user types html without knowing the underlying dom is xml

[16:40:00.0000] <annevk>
/me changed a few normative numbers around :p

[16:43:00.0000] <weinig>
annevk: one big reason I am not concerned is that I don't think those numbers match IE, do they?

[16:43:01.0000] <Hixie>
IE doesn't throw DOMException objects

[16:44:00.0000] <weinig>
Hixie: good point

[16:44:01.0000] <weinig>
/me remembers that discussion fondly now :p

[16:44:02.0000] <annevk>
you mean the one in Seattle? ...

[16:44:03.0000] <weinig>
annevk: yes

[16:45:00.0000] <annevk>
yeah...

[16:45:01.0000] <weinig>
annevk: is there any reason for non same origin request to not make upload notifications?

[16:45:02.0000] <Hixie>
hsivonen: "http://www.whatwg.org/specs/web-apps/current-work/source-whatwg":7099.29-7100.28: info warning: Element name "dfnode" cannot be represented as XML 1.0.

[16:46:00.0000] <Hixie>
hsivonen: seems that error message is bogus (the element name is unknown for sure, but it's fine in xml)

[16:47:00.0000] <annevk>
weinig, I suppose at that point the preflight check already happened and it's ok

[16:48:00.0000] <weinig>
annevk: ok, good

[16:48:01.0000] <annevk>
weinig, even better, now we dropped the PI the same might go for readyState == 2

[16:48:02.0000] <weinig>
/me nods

[16:49:00.0000] <weinig>
/me eek, I think webkit already does that

[16:49:01.0000] <weinig>
/me did not read close enough

[16:50:00.0000] <zcorpan>
nn

[16:51:00.0000] <zcorpan>
Hixie: just one thing: "Note: These are the only ways to make a DOM unserializable." is false given execCommand('insertHTML') :P

[16:52:00.0000] <annevk>
weinig, given that there's no processing instruction I don't think you're in much trouble

[16:52:01.0000] <Hixie>
zcorpan: hmm

[16:52:02.0000] <Hixie>
zcorpan: send mail, something just came up here and i have to run

[16:52:03.0000] <annevk>
weinig, I'm writing an XMLHttpRequest Level 2 comment e-mail now to remind myself to fix this

[16:52:04.0000] <Hixie>
zcorpan: but thanks!

[16:52:05.0000] <Hixie>
a lot :-)

[16:53:00.0000] <zcorpan>
i'll file a bug

[16:53:01.0000] <Hixie>
cool

[16:53:02.0000] <Hixie>
/me goes

[16:58:00.0000] <annevk>
roc, "which probably tells people immediately that I'm an annoying freak, but I think it's good to be open about that." :p


2008-08-11
[17:14:00.0000] <annevk>
does anyone know what i7 stands for in the new Intel product line?

[17:14:01.0000] <annevk>
(press release doesn't say)

[17:19:00.0000] <weinig>
annevk: nice

[19:45:00.0000] <takkaria>
if anyone has an example of a page which ends up needing to be reparsed because of html5's charset algorithm when a meta tag is encountered, I'd very much appreciate a pointer to it

[20:03:00.0000] <Hixie>
takkaria: <script>alert('A');</script><!-- xx megabytes xx --><meta charset="C">

[20:03:01.0000] <Hixie>
takkaria: where A is a byte that is interpreted differently in two encodings

[20:03:02.0000] <Hixie>
and C is the one of those encodings that isn't hte one the browser picks by default

[20:03:03.0000] <takkaria>
ah, I get you.  I thought it might be mostly to do with scripting

[20:03:04.0000] <Hixie>
doesn't have to be scripting

[20:03:05.0000] <Hixie>
could be: A<!-- xx megabytes xx --><meta charset="C">

[20:04:00.0000] <Hixie>
it's just more obvious with an alert() because then you get to see that it shows up twice

[20:04:01.0000] <takkaria>
right!  got you, thanks

[20:04:02.0000] <Hixie>
np

[20:05:00.0000] <takkaria>
there is a web browser equipped with the HTML5 parsing algorithm now, though it's not got scripting so isn't the most useful browser for testing all the bits that require scripting :)

[20:05:01.0000] <Hixie>
:-)

[20:05:02.0000] <Hixie>
is scripting on the agenda?

[20:06:00.0000] <takkaria>
at some point in the future, I believe, but nothing very soon

[20:06:01.0000] <Hixie>
k

[20:06:02.0000] <takkaria>
otoh, I expect there'll be useful feedback from users of NetSurf when pages break

[22:50:00.0000] <Hixie>
sweet

[22:50:01.0000] <Hixie>
in IE you can insert multiple nodes into the document (adjacent to the root element) using insertAdjacentHTML()

[22:53:00.0000] <Hixie>
and iAH() doesn't work with table bits in IE

[23:22:00.0000] <Hixie>
anyone got an old version of IE? like IE7 or IE6?

[23:22:01.0000] <Hixie>
i'd like http://junkyard.damowmow.com/336 tested with one of those

[23:23:00.0000] <krijnh>
024578A9 (IE6)

[23:23:01.0000] <Hixie>
thanks

[23:23:02.0000] <krijnh>
021453786A9 (IE5.5)

[23:23:03.0000] <Hixie>
wow, cool, thanks

[23:24:00.0000] <krijnh>
Run with IETester..

[23:24:01.0000] <Hixie>
IETester?

[23:24:02.0000] <krijnh>
http://www.my-debugbar.com/wiki/IETester/HomePage

[23:24:03.0000] <Hixie>
neat

[23:25:00.0000] <krijnh>
No idea if it's reliable :)

[23:25:01.0000] <Hixie>
yeah i was about to say, i wonder if that's reliable with javascript: and iframes

[23:26:00.0000] <krijnh>
It's the only way for me to get some different versions of IE running on Vista

[00:21:00.0000] <aboodman2>
Hixie: I keep wishing that support for conversations via MessageChannel was more built-in

[00:22:00.0000] <Hixie>
how do you mean?

[00:22:01.0000] <aboodman2>
like, when you get a message, you always get an object you can reply to the sender with. like it is built into the api.

[00:23:00.0000] <aboodman2>
It's just a little thing, but I was wondering if you could think of some way to make it work.

[00:23:01.0000] <aboodman2>
I keep wanting the sender to be able to do:

[00:23:02.0000] <aboodman2>
var port = worker(or iframe).sendMessage("foo");

[00:23:03.0000] <aboodman2>
and then the port represents the conversation

[00:24:00.0000] <aboodman2>
and on the other side, the worker (or iframe) could do globalObject.addEventListener("message", function(e) { e.port.sendMessage("reply") }, false)

[00:24:01.0000] <Hixie>
a port _is_ a conversation

[00:24:02.0000] <aboodman2>
right, i know

[00:24:03.0000] <Hixie>
if you want to split the conversation off, just send a port with the message

[00:25:00.0000] <Hixie>
and use that port for the subconversation

[00:25:01.0000] <aboodman2>
i know, it seems slightly less elegant for some reason to send the port as an argument

[00:25:02.0000] <Hixie>
i don't understand you example fragments

[00:25:03.0000] <Hixie>
your

[00:25:04.0000] <aboodman2>
one second

[00:33:00.0000] <aboodman2>
k, here is how it works now:

[00:33:01.0000] <aboodman2>
var channel = new MessageChannel();

[00:33:02.0000] <aboodman2>
var worker = createWorker("foo.js");

[00:33:03.0000] <aboodman2>
channel.port1.addEventListener("message", function(e) { ... }, false);

[00:33:04.0000] <aboodman2>
worker.sendMessage("init", channel.port2);

[00:33:05.0000] <aboodman2>
here is how I wish it would work

[00:33:06.0000] <aboodman2>
var worker = createWorker("foo.js");

[00:33:07.0000] <aboodman2>
var port = worker.sendMessage("init");

[00:33:08.0000] <aboodman2>
port.addEventListener("message", function(e) { ... }, false);

[00:34:00.0000] <aboodman2>
the worker code would look the same as it does now. e.port would contain a port that could talk back to the return value from sendMessage()

[00:35:00.0000] <aboodman2>
I do not know how to make this work with your goal of wanting to be able to send arbitrary ports though.

[00:35:01.0000] <Hixie>
/me reads (sorry, had a cat petting emergency)

[00:37:00.0000] <Hixie>
note that |port.addEventListener("message", function(e) { ... }, false);| is better written as |port.onmessage = function (e) { ... };|

[00:37:01.0000] <aboodman2>
why?

[00:37:02.0000] <aboodman2>
because otherwise i need to call start()?

[00:37:03.0000] <Hixie>
it's shorter, and doesn't require calling start()

[00:37:04.0000] <Hixie>
but that's orthogonal

[00:38:00.0000] <Hixie>
i agree that it would be neat to have a method that creates a port and includes it

[00:38:01.0000] <Hixie>
and returns a matching port

[00:38:02.0000] <Hixie>
you would likely not want that to happen every time, creating a port is presumably not free

[00:39:00.0000] <Hixie>
btw, it's now worker.port.sendMessage(), since sicking wanted a separate Worker object returned from createWorker()

[00:39:01.0000] <aboodman2>
yeah.

[00:40:00.0000] <Hixie>
we could have port.postMessageWithNewChannel('message') which returns a port

[00:40:01.0000] <Hixie>
though that name sucks

[00:41:00.0000] <aboodman2>
it seems like it would be possible for implementations to know if the port is used and delay the hard work until then

[00:42:00.0000] <Hixie>
true

[00:42:01.0000] <Hixie>
maybe

[00:42:02.0000] <aboodman2>
another thing.

[00:42:03.0000] <aboodman2>
let's say you send port2 to a friend.

[00:43:00.0000] <aboodman2>
and then you start doing things like:

[00:43:01.0000] <aboodman2>
port2.onmessage = function(e) { alert('hi') }

[00:43:02.0000] <aboodman2>
that's totally nonsensical, right?

[00:44:00.0000] <Hixie>
yeah

[00:44:01.0000] <Hixie>
once you send port2, it'll never receive a message again

[00:47:00.0000] <Hixie>
so should i make postMessage() always create a new channel and send a port?

[00:47:01.0000] <Hixie>
seems expensive to me

[00:47:02.0000] <Hixie>
even if theoretically it could be optimised

[00:48:00.0000] <aboodman2>
no

[00:48:01.0000] <aboodman2>
I'm not sure about sicking but a lot of my feedback has been wrong based on not totally understanding what you were trying to do or what the spec proposed.

[00:49:00.0000] <aboodman2>
so i'm just registering the wish that conversations were more elegant because that is a real problem with gears (and postMessage) right now

[00:50:00.0000] <aboodman2>
but i don't have a concrete suggestion

[00:52:00.0000] <Hixie>
maybe port.startConversation('message'); which returns a port and sends a new port

[00:52:01.0000] <aboodman2>
yeah, i think something like that would be cool

[01:30:00.0000] <Hixie>
aboodman2: ok, added startConversation()

[01:30:01.0000] <aboodman2>
don't do that

[01:31:00.0000] <aboodman2>
i was speaking theoretically

[01:31:01.0000] <Hixie>
you said it would be cool

[01:31:02.0000] <Hixie>
i agree

[01:32:00.0000] <Hixie>
man, you people need to stop telling me that things would be good ideas if you don't think they're good ideas

[01:32:01.0000] <Hixie>
anyway, it's in now, but easy to remove if we decide we don't want it

[01:32:02.0000] <aboodman2>
what was the expense you were thinking of to create a port?

[01:34:00.0000] <Hixie>
creating a port and sending it requires creating a MessageChannel and three MessagePorts instead of just creating the two ports already entangled in the right thread spaces

[01:35:00.0000] <aboodman2>
can you get from a port object to a channel?

[01:35:01.0000] <Hixie>
no

[01:35:02.0000] <Hixie>
a channel is basically just an array of two ports

[01:36:00.0000] <aboodman2>
right, so you just need to create the port to return

[01:36:01.0000] <aboodman2>
the port seems like it just wraps a unique integer

[01:36:02.0000] <aboodman2>
and then on the other side you have to create the other port, which can be done lazily

[01:36:03.0000] <Hixie>
right

[01:38:00.0000] <Hixie>
but without startConversation(), the author has to create a new messagechannel, then call postmessage, which creates a third port

[01:38:01.0000] <Hixie>
hence startConversation() is cheaper

[01:39:00.0000] <Hixie>
i wish sicking would send feedback

[01:39:01.0000] <Hixie>
he was all like "we have to have this done in 11 days or it's all over!" last week

[01:39:02.0000] <Hixie>
barely heard three e-mails out him since

[01:39:03.0000] <Hixie>
out of him

[01:40:00.0000] <gavin_>
he was at blackhat last week, fwiw

[01:41:00.0000] <Hixie>
yeah

[01:41:01.0000] <Hixie>
so? :-)

[01:41:02.0000] <Hixie>
i was off sick last week, didn't stop me from responding to all his mails :-)

[01:44:00.0000] <gavin_>
it's harder to be around a computer 24/7 at blackhat than it is when you're off sick :)

[01:45:00.0000] <gavin_>
casino security doesn't appreciate laptops at the blackjack tables

[01:45:01.0000] <Hixie>
i'm just saying'

[01:46:00.0000] <Hixie>
er, sayin'

[01:46:01.0000] <Hixie>
he was volunteering to edit it because it was so urgent

[01:46:02.0000] <Hixie>
presumably because i was a bottleneck

[01:46:03.0000] <gavin_>
yeah- did you emailhima bout it?

[01:47:00.0000] <aboodman2>
Hixie: it's pretty hard to digest these.

[01:47:01.0000] <Hixie>
aboodman2: hm?

[01:47:02.0000] <gavin_>
I think you'd get better results with that than with griping on IRC when he isn't her e:)

[01:47:03.0000] <Hixie>
i'm just ranting, i'm sure he's replying as fast as he can :-)

[01:48:00.0000] <Hixie>
i didn't e-mail him specifically to ask why he hadn't replied, i assumed he had good reasons :-)

[01:48:01.0000] <Hixie>
aboodman2: these what?

[01:54:00.0000] <aboodman2>
Hixie: specs

[01:55:00.0000] <Hixie>
sure

[01:55:01.0000] <Hixie>
i made a bunch of demos which hopefully help explain the worker model a bit

[01:55:02.0000] <aboodman2>
yes, they help a lot.

[02:42:00.0000] <Lachy>
I finally found a solution to allow me to stream olympics from NBCs website, even though I'm not in the USA! :-)

[02:43:00.0000] <hendry>
Lachy: which is? proxy? ?:)

[02:44:00.0000] <Lachy>
I forward all connections using ssh through my web server which is in the USA

[02:44:01.0000] <Lachy>
using: ssh -D 8080 -fN lachy.id.au  - and then setting my setting the SOCKS proxy in my browser to localhost 8080

[02:45:00.0000] <Lachy>
http://ubuntu-tutorials.com/2008/06/18/tunnel-web-and-dns-traffic-over-ssh/

[02:47:00.0000] <gavin_>
there are no australian sites that offer internet feeds of the ollympics?

[02:47:01.0000] <gavin_>
perhaps http://www.cbc.ca/olympics/ would work better?

[02:48:00.0000] <gavin_>
oh, I forgot you're in norway

[02:50:00.0000] <Hixie>
well my 1.5TB disk just started makin horrific noises

[02:50:01.0000] <Hixie>
so i guess it's time to get a drobo

[02:50:02.0000] <hendry>
Lachy: i'm IP restricted from BBC news covereage, even though I am in the UK http://www.flickr.com/photos/hendry/2753042810/

[02:52:00.0000] <gavin_>
I guess the broadcasters are forced to implement some sort of geographical restrictions

[02:52:01.0000] <hsivonen>
Hixie: http://www.whatwg.org/specs/web-apps/current-work/source-whatwg is 404

[02:52:02.0000] <hendry>
gavin_: forced by "exclusive deals" and greed

[02:53:00.0000] <Hixie>
hsivonen: yeah, it's only not 404 while i'm running the gen script

[02:53:01.0000] <Hixie>
hsivonen: the markup just had <dfncode> or something along those lines

[02:53:02.0000] <Hixie>
no embedded nulls or anything

[02:56:00.0000] <hsivonen>
Hixie: ok

[02:58:00.0000] <Lachy>
gavin_, yahoo7.com.au/Olympics has streaming, but I need access to server in Australia that I can tunnel through. I would prefer that, cause that only requires flash. I had to install silverlight to watch NBC, and I wish I didn't

[02:58:01.0000] <gavin_>
no norwegian sites?

[02:59:00.0000] <hendry>
/me is there is a flickr or youtube of short mp3 clips, so i can embed sound into my blog with (urg) flash (fall back to <audio> would be a plus)

[02:59:01.0000] <Lachy>
not that I know of. But even if they did, they would have norwegian commentary, not english

[02:59:02.0000] <Lachy>
there is also SOPCast, which has live streaming, but that only seems to have a chinese language stream

[03:00:00.0000] <Lachy>
and its windows only, so I can only run it in virtual machine

[03:02:00.0000] <Hixie>
ok, drobo purchased

[03:02:01.0000] <Hixie>
i'm fed up of disks dying all the time, hopefully this will make it a non-issue

[03:02:02.0000] <Lachy>
Hixie, I was thinking about getting myself a drobo too

[03:03:00.0000] <Hixie>
they seem pretty sweet

[03:03:01.0000] <Lachy>
I don't have problem with disks dying, but I am running out of storage space, so I need an economical solution that is cheaper to setup than a RAID

[03:07:00.0000] <Hixie>
can't you just buy a new disk and add it to the usb bus?

[03:08:00.0000] <hsivonen>
my HFS+ filesystem holding my real-world test data and Time Machine backups just died

[03:08:01.0000] <hsivonen>
/me wishes HFS+ were more robust

[03:08:02.0000] <hsivonen>
I guess next time I should put the real-world data on a disk image instead of stressing the catalog file of the concrete fs

[03:09:00.0000] <Hixie>
i made a point of not putting anything but my time machine backups on the time machine drive

[03:09:01.0000] <Lachy>
I already have 2 external 1TB drives

[03:09:02.0000] <Hixie>
(that's the one that just started making weird high pitches screeching noises)

[03:09:03.0000] <hsivonen>
Hixie: probably a good idea

[03:09:04.0000] <Hixie>
(i just turned it off without checking if it was ok)

[03:10:00.0000] <Hixie>
(don't want to cause it to fail before i have something to replace it)

[03:10:01.0000] <hsivonen>
my Time Machine disk never fills up, because HFS+ dies first and I have to restart Time Machine history

[03:12:00.0000] <Hixie>
my 1.5TB disk was down to a few dozen GB

[03:12:01.0000] <Hixie>
but unrelated to that i've had hell getting it to work over the network

[03:13:00.0000] <hsivonen>
I have high hopes for zfs

[03:13:01.0000] <Hixie>
my disk failures have always sounded very hardware-related

[03:13:02.0000] <Hixie>
i had a dell laptop once whose motherboard had a bug that was actually destroying hard disks

[03:14:00.0000] <Hixie>
and my mac mini has had two disks die on it (both maxtor)

[03:14:01.0000] <Hixie>
(including this latest one)

[03:15:00.0000] <Hixie>
my plan once i have the drobo is to get spinrite and put it on an old pc i have lying around and see if i can get the dead maxtor 750GB disk and the potentially dead maxtor 1.5TB disk up and working again

[03:15:01.0000] <gavin_>
/me can't recall ever suffering from a hard drive failure

[03:15:02.0000] <Hixie>
and if so, use them as an external disk for my girlfriend's timemachine instead of going over the network

[03:16:00.0000] <Hixie>
anyway, bed time first

[03:17:00.0000] <Hixie>
btw anyone know why <script src=""> doesn't support javascript: URLs? (zcorpan sent mail late last year about it)

[03:20:00.0000] <gavin_>
does it not work in gecko?

[03:20:01.0000] <gavin_>
/me would expect it o

[03:21:00.0000] <Hixie>
it does nothing

[03:22:00.0000] <Hixie>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E...%3Cscript%20src%3D%22javascript%3Aalert(%27test%27)%3B%27test%27%22%3E%3C%2Fscript%3E

[03:23:00.0000] <gavin_>
well, "alert" isn't valid JS

[03:24:00.0000] <gavin_>
that reutns alert.toString()

[03:24:01.0000] <gavin_>
which would justredefine alert

[03:24:02.0000] <gavin_>
...with invalid JS code

[03:24:03.0000] <gavin_>
nevermind

[03:30:00.0000] <webben>
Hixie: Do you have a view on whether, if RDFa in XHTML becomes a Recommendation (currently it's CR), it would be conforming with HTML5 to use its elements and attributes in an XML document also using the new RDFa attributes, which are also in http://www.w3.org/1999/xhtml ( http://www.w3.org/TR/2008/CR-rdfa-syntax-20080620/ ) ?

[03:30:01.0000] <webben>
or are these drafts fundamentally incompatible?

[03:31:00.0000] <Hixie>
sorry i don't quite follow the question

[03:32:00.0000] <webben>
Hixie: I guess it boils down to: if some elements and attributes are placed into the HTML namespace by another W3C recommendation, can they be used in an XML document also using elements and attributes in the HTML namespace which are defined by HTML5?

[03:34:00.0000] <webben>
Would that break document conformance with HTML5, for instance?

[03:34:01.0000] <hsivonen>
Hixie: I'm unable to reproduce the dtcode bug

[03:35:00.0000] <annevk>
webben, it would

[03:36:00.0000] <webben>
annevk: Would it break conformance in a different way than (for example) including some elements in a completely different namespace (I dunno, MusicML or something) would?

[03:37:00.0000] <webben>
annevk: i.e. would it have the same relationship to HTML5 as the XML documents include XHTML and MathML do to XHTML 1.0? http://www.w3.org/TR/xhtml1/#well-formed

[03:37:01.0000] <webben>
*as XML documents that include

[03:37:02.0000] <hsivonen>
webben: as implemented, they both get the same type of error message

[03:38:00.0000] <hsivonen>
interestingly, my "eRDF5" trial balloon got Warnocked, even though a couple of people had tried to stir discussion about RDFao

[03:38:01.0000] <hsivonen>
s/ao/a/

[03:39:00.0000] <hsivonen>
Hixie: oops. I can reproduce the dtcode thing now

[03:39:01.0000] <hsivonen>
will investigate

[03:40:00.0000] <webben>
hsivonen: Thanks.

[03:52:00.0000] <mcarter>
does anyone know what a browser is supposed to do when (in an xhr response) it receives data ending with a partial unicode character -- should it trigger a readystate 3 with only the last complete character?

[03:53:00.0000] <annevk>
it should simply try to decode it

[03:53:01.0000] <annevk>
the partial character will turn into U+FFFD

[03:54:00.0000] <hsivonen>
mcarter: do you mean really ending or the first buffer containing a partial character with potentially more data coming up?

[03:54:01.0000] <mcarter>
hsivonen, actually, I'm curious about both cases

[03:55:00.0000] <annevk>
same for both

[03:55:01.0000] <Hixie>
webben: it has the same conformance status as web forms 2 vs html4

[03:55:02.0000] <Hixie>
webben: using it would be conforming to the spec that makes up those attributes, but not to html5 proper

[03:55:03.0000] <hsivonen>
mcarter: if the stream ends, I'd do what annevk said. If the stream doesn't end, I'd hold onto the partial character expecting more data and behave as if that character hadn't yet been received as far as scripts go

[03:56:00.0000] <webben>
Hixie: That seems a reasonable answer. :) Thanks.

[03:56:01.0000] <Hixie>
webben: it doesn't matter what the namespace is, though

[03:56:02.0000] <Hixie>
webben: from the point of view of conformance

[03:56:03.0000] <mcarter>
annevk, how could the web work at all if it turned it into a U+FFFD character? sometimes unicode characters get split into two seperate packets, and half of the character might come through before the other half

[03:56:04.0000] <webben>
Hixie: okay

[03:57:00.0000] <hsivonen>
mcarter: that is, I'd expect the partial character stay inside the decoder and the decoder not outputting more data until it has at least one complete character to output

[03:57:01.0000] <mcarter>
hsivonen, that makes sense

[03:57:02.0000] <Hixie>
webben: i mean, the spec could say it's the xhtml namespace, or that it's the foobar namespace, html5 doesn't care, html5 just cares about "metadata elements" and "phrasing elements" and so on, the categories

[04:00:00.0000] <hsivonen>
fwiw, I think RDFa could be mostly harmless if it used full URIs instead of CURIEs and, therefore, didn't need to know the namespace mapping context at each node

[04:00:01.0000] <hsivonen>
I'm strongly opposed to importing any CURIE or other qnames-in-content stuff into HTML5

[04:01:00.0000] <Hixie>
RDFa would be mostly harmless in the same way that products in sky mall magazines are mostly harmless

[04:02:00.0000] <Hixie>
they're solving a problem that people think they have when they're stuck 30,000ft up for 12 hours, but when they get down to earth, they don't find it actually fits in with their workflow

[04:04:00.0000] <Hixie>
i should totally make a sky mall catalogue of web standards

[04:04:01.0000] <annevk>
:)

[04:06:00.0000] <hsivonen>
I'd be interested to know how RDFa ended up in the HTML WG charter

[04:07:00.0000] <annevk>
probably similar to how the word XForms ended up there

[04:08:00.0000] <annevk>
or ITS

[04:09:00.0000] <hsivonen>
Hixie: dtcode issue fixed

[04:09:01.0000] <Hixie>
cool, what was it?

[04:10:00.0000] <hsivonen>
Hixie: no longer saying that <dtcode> cannot be mapped to XML

[04:10:01.0000] <Hixie>
right i mean what was the source of the bug? :-)

[04:11:00.0000] <hsivonen>
Hixie: I had forgotten the NCName check and only checked for customness when name policy was allowing XML violations

[04:32:00.0000] <annevk>
Hixie, insertAdjacentHTML is not an attribute

[04:32:01.0000] <annevk>
Hixie, the sentence outerHTML and insertAdjacentHTML attributes should probably say "members"

[08:35:00.0000] <zcorpan>
i find it amusing that people have opinions about the <script src=javascript:'...'> thing

[08:44:00.0000] <annevk>
maybe there was not enough e-mail

[10:02:00.0000] <zcorpan>
firefox adds a playbackrate=1 attribute to <video> elements

[10:02:01.0000] <takkaria>
fun times

[10:02:02.0000] <hsivonen>
zcorpan: do you mean in add the attribute to teh DOM even if absent from source?

[10:03:00.0000] <zcorpan>
hsivonen: yes

[10:03:01.0000] <zcorpan>
if you have <video playbackrate=2> in source firefox still says it's 1

[10:03:02.0000] <hsivonen>
zcorpan: that sucks. is there a bug report already?

[10:03:03.0000] <zcorpan>
if you set video.playbackRate = 3 firefox updates the content attribute

[10:04:00.0000] <zcorpan>
dunno

[10:05:00.0000] <hsivonen>
not with the obvious word in description, at least

[10:06:00.0000] <zcorpan>
same with <audio>

[10:06:01.0000] <hsivonen>
I'll file a bug.

[10:07:00.0000] <zcorpan>
cool

[10:11:00.0000] <hsivonen>
aargh. my tokenizer loop has gone over 8000 byte codes again

[10:16:00.0000] <hsivonen>
https://bugzilla.mozilla.org/show_bug.cgi?id=450115

[12:10:00.0000] <annevk>
did @whatwg.org just get spam?

[12:11:00.0000] <annevk>
http://googlemobile.blogspot.com/2008/08/google-translate-now-for-iphone.html -_- instead of actually making a normal Web page that just works they add special trickery for the iPhone

[16:40:00.0000] <takkaria>
oops, I almost replied to RB then


2008-08-12
[17:18:00.0000] <Hixie>
/me sends another private e-mail to kristof trying to convince him to use proper quoting style

[18:32:00.0000] <Hixie>
i'm down to only 107 e-mails from henri to deal with

[18:33:00.0000] <Hixie>
and there are 1500 or so e-mails in folders i've deferred for one reason or another, and only 500 others

[18:33:01.0000] <Hixie>
the end is in sight!

[18:33:02.0000] <Hixie>
only a year or two to go

[18:33:03.0000] <Hixie>
to get this feedback to sane numbers

[18:34:00.0000] <csarven>
You should just text-to-speech and listen to all the emails while you sleep! :P

[18:34:01.0000] <Hixie>
i've read most of them already

[18:49:00.0000] <Hixie>
why does firefox2 scroll the page when the notification widget comes up

[18:49:01.0000] <Hixie>
weird

[18:49:02.0000] <Hixie>
i wish i could use ff3 on my linux box

[18:58:00.0000] <BlueG>
I am curious about the relationship between the WHATWG and the W3C HTML WG. Ian Hickson is in both groups and is the editor for the spec, right?

[18:58:01.0000] <Hixie>
i am, yes

[18:59:00.0000] <Hixie>
html5 is sort of a joint venture between the two groups

[18:59:01.0000] <BlueG>
ok, I am not entire sure I understand why there are two working groups and what is the difference in their roles?

[18:59:02.0000] <Hixie>
it's sort of a historical thing

[18:59:03.0000] <BlueG>
ok

[18:59:04.0000] <Hixie>
whatwg started off in 2004 or so when the w3c told the browser vendors who founded the whatwg that the w3c didn't want to work on html5

[18:59:05.0000] <Hixie>
they then founded the whatwg to work on html5 in public

[19:00:00.0000] <BlueG>
so, people who are interested in the developing spec can join either, or both wg?

[19:00:01.0000] <Hixie>
later on (late 2006ish to early 2007) the w3c changed their minds and asked if the html5 work could happen in the w3c, so it now happens in both

[19:00:02.0000] <Hixie>
yup

[19:01:00.0000] <Hixie>
http://blog.whatwg.org/w3c-restarts-html-effort has instructions on how to join the w3c group

[19:01:01.0000] <BlueG>
ok

[19:01:02.0000] <Hixie>
http://www.whatwg.org/mailing-list#specs has instructions on joining the whatwg group

[19:02:00.0000] <BlueG>
what are the current differences between the two?

[19:02:01.0000] <Hixie>
the main difference is that in theory microsoft participate only in the w3c list

[19:02:02.0000] <Hixie>
in practice they don't really participate there either very much

[19:03:00.0000] <BlueG>
ok

[19:05:00.0000] <Hixie>
there are certain topics that just get discussed more on one list than the other

[19:06:00.0000] <Hixie>
for example video and the new web workers stuff is more discussed on the whatwg list

[19:06:01.0000] <Hixie>
and <img alt=""> and <table> semantics are more discussed on the w3c list

[19:06:02.0000] <BlueG>
so it is probably good to join both, at least if you have broad interest in the development

[19:06:03.0000] <Hixie>
but that's mostly a function of which list people who care about those topics are primarily on

[19:06:04.0000] <Hixie>
yeah, though to be honest you'll find that's a lot of traffic :-)

[19:06:05.0000] <BlueG>
ok

[19:07:00.0000] <Hixie>
it's died down now a bit, and the w3c list gets about 500 e-mails a month, but there was a time where 1200 e-mails amonth wasn't uncommon

[19:07:01.0000] <Hixie>
whatwg has a lot less traffic, and it tends to be somewhat more technically focussed

[19:08:00.0000] <Hixie>
but that's mostly because the whatwg list is an older and more mature community now, so expected behaviour is more understood

[19:09:00.0000] <Hixie>
(whatwg has about 300 a month)

[19:12:00.0000] <BlueG>
it looks easier to join whatwg

[19:12:01.0000] <Hixie>
there's that too :-)

[19:12:02.0000] <BlueG>
haha

[19:13:00.0000] <BlueG>
ok

[19:14:00.0000] <Hixie>
you can also read the w3c list online if you don't want to post much: http://lists.w3.org/Archives/Public/public-html/latest

[19:14:01.0000] <BlueG>
ok

[19:16:00.0000] <Hixie>
this irc channel is sort of a hub for a lot of the html5 activity (mostly because i rant on here while i'm editing the spec :-) ) so if you hang out here you'll likely here about anything interesting going on

[19:17:00.0000] <Hixie>
there are logs for this channel too, see the /topic for details

[19:20:00.0000] <BlueG>
ok

[19:26:00.0000] <BlueG>
there seems to be a problem with the signup page for whatwg, its throwing an error

[19:26:01.0000] <Hixie>
hm, crap

[19:26:02.0000] <Hixie>
/me looks

[19:33:00.0000] <Hixie>
try sending mail to whatwg-join⊙wo no idea what's up with mailman.

[19:33:01.0000] <Hixie>
i'll have to send dreamhost a support request if it's still broken tonight

[19:35:00.0000] <BlueG>
ok

[19:36:00.0000] <Hixie>
right, dinner time. i'll be back in a bit. hopefully dev.w3.org's cvs server and whatwg's mailman server will be in a better state upon my return.

[21:09:00.0000] <BlueG>
whatwg-join⊙wo doesn't seem to have worked either, and the mailman page is still giving the same error

[22:40:00.0000] <Hixie>
BlueG: reported it, hopefully they'll fix it soon

[22:41:00.0000] <BlueG>
ok, thanks

[01:18:00.0000] <annevk>
hmm, wtf? http://forums.whatwg.org/viewtopic.php?t=201

[01:27:00.0000] <hsivonen>
annevk: I don't see a usual spam payload. perhaps the author just has a lot to post.

[01:37:00.0000] <zcorpan>
Hixie: <script src=javascript:a=1> and <script src=javascript:"a=1"> are different

[01:37:01.0000] <zcorpan>
Hixie: i'm not sure i understand which you've specced but the first should work and the second not

[01:37:02.0000] <Hixie>
no, neither works

[01:38:00.0000] <annevk>
hsivonen, yeah...

[01:39:00.0000] <annevk>
Hixie, insertAdjacentHTML is at one point referred to as "attribute"

[01:39:01.0000] <zcorpan>
Hixie: oh. hmm. the first works in opera

[01:51:00.0000] <zcorpan>
annevk: thanks (re spam)

[02:31:00.0000] <zcorpan>
"It may have been a mere warning, but it made a lot, lot, lot of people anxious and upset. So, by popular demand – and also because the XHTML working group are preparing a revised note on XHTML and media types −　the warning is gone." -- http://www.w3.org/QA/2008/08/markup_validator_updated.html

[02:31:01.0000] <Hixie>
zcorpan: yeah, zcorpan (i think) mentioned that already

[02:31:02.0000] <Hixie>
er

[02:31:03.0000] <Hixie>
s/zcorpan/annevk/

[02:32:00.0000] <Hixie>
zcorpan: what was the waring?

[02:32:01.0000] <Hixie>
oh i see

[02:33:00.0000] <Hixie>
how silly

[02:33:01.0000] <Hixie>
oh well

[02:33:02.0000] <Hixie>
i look forward to henri's validator giving the w3 one a run for its money

[02:55:00.0000] <hsivonen>
/me notes that comments can be implemented with less buffer copying if stuff is always "appended" but emitting the comment tokens trims the right number of trailing hyphens

[02:56:00.0000] <hsivonen>
the joke is that the SGML people used the longest names for the most common things they talked about

[02:57:00.0000] <hsivonen>
but as takkaria abserved about the # of states for doctypes, it seems that the importance of a given piece of SGML-legacy syntax in inversely proportional to its usefulness

[02:58:00.0000] <hsivonen>
it's dumb how many states go for doctypes and comments compared to tags

[03:00:00.0000] <hsivonen>
I need to eliminate 129 bytes from my tokenizer loop

[03:14:00.0000] <Hixie>
hsivonen: the only reason tags are simpler is because we skipped out on 90% of the syntax for tags, whereas for doctypes we went all-out in doing the sgml-compatible thing

[03:17:00.0000] <hsivonen>
Hixie: what did you skip? <> and </> still invoke stuff even though different stuff

[03:17:01.0000] <hsivonen>
good thing we aren't doing internal subsets

[03:18:00.0000] <Hixie>
<foo/ /, <> and </> implying the right tags, etc

[03:18:01.0000] <hsivonen>
I should go read the XML archives and find out who perpetrated all the DTD stuff

[03:18:02.0000] <hsivonen>
oh right, <foo/  /

[03:18:03.0000] <Hixie>
philip taylor (w) claims that <script src=javascript:"alert(1)"></script> shows an alert in SeaMonkey 1.1.11

[03:19:00.0000] <hsivonen>
I mean, one has to respect Goldfarb for starting all this, but there would have been some serious opportunity for syntax simplification

[03:20:00.0000] <annevk>
I think SeaMonkey is Gecko 1.8, not 1.9

[03:22:00.0000] <Hixie>
yeah that was my guess too

[03:23:00.0000] <annevk>
he's probably right in that it's possible to change

[03:23:01.0000] <hsivonen>
134 bytecodes to cut...

[03:23:02.0000] <hsivonen>
/me doesn't like arbitrary VM limits

[03:25:00.0000] <hsivonen>
Sun is now all like "there are more JVM languages than Java"

[03:25:01.0000] <hsivonen>
but then they keep around this silly 8000 byte limit that makes it harder to compile other languages into bytecode

[03:25:02.0000] <hsivonen>
or to write huge state machines by hand

[03:26:00.0000] <Hixie>
what is this limit?

[03:27:00.0000] <hsivonen>
Hixie: if a method takes more than 8000 bytes, it doesn't get JITed

[03:27:01.0000] <hsivonen>
which sucks *badly*

[03:27:02.0000] <Hixie>
that's odd

[03:31:00.0000] <Hixie>
so about 1% of files in philip's sample have meta content-language

[03:31:01.0000] <hsivonen>
it seems that the Eclipse Java Compiler outputs very naïve bytecode

[03:31:02.0000] <Hixie>
i suppose i should check what fraction of those have useful content-language values that aren't redundant with other language information

[03:31:03.0000] <hsivonen>
I guess I should check if javac already compiles this to under 8000 bytes

[03:31:04.0000] <hsivonen>
Hixie: FrontPage outputs Content-Language

[03:32:00.0000] <Hixie>
usefully?

[03:32:01.0000] <hsivonen>
and it affects spell checking, which is good for getting people to change it

[03:32:02.0000] <hsivonen>
Hixie: yes, unless it gets overriden by <span lang> and the user doesn't notice

[03:32:03.0000] <Hixie>
sigh

[03:33:00.0000] <hsivonen>
Hixie: they tried to be really smart and tied the input language to the input method

[03:33:01.0000] <Hixie>
with lang="", xml:lang="", meta content-language treated as lang, and maybe even http content-language treated per mime, determining the language of a node is going to be one heck of a pain in the ass

[03:33:02.0000] <hsivonen>
which works if your universe only includes American English, Japanese, Chinese, Korean and Hebrew

[03:34:00.0000] <hsivonen>
but it sucks pretty badly when Europeans use their national keyboards for English

[03:36:00.0000] <annevk>
Hixie, Firefox at least already does that for :lang()

[03:37:00.0000] <annevk>
Hixie, we have some bugs on Opera to get it to do the same (or it might already have been fixed, forgot)

[03:37:01.0000] <annevk>
Content-Language, <meta http-equiv=content-language>, lang=""

[03:38:00.0000] <Hixie>
so what happens when Content-Language has multiple languages set?

[03:39:00.0000] <annevk>
pick the first

[03:39:01.0000] <zcorpan>
i don't particularly like that content-language is involved at all

[03:39:02.0000] <zcorpan>
but oh well

[03:41:00.0000] <Hixie>
m either

[03:43:00.0000] <Hixie>
should i make the Content-Language pragma non-conforming?

[03:43:01.0000] <Hixie>
i guess there's no point

[03:44:00.0000] <annevk>
we could try to remove support for it from browsers I suppose

[03:46:00.0000] <hsivonen>
I've already written code that supports <meta http-equiv=content-language> between <html lang> and HTTP

[03:46:01.0000] <annevk>
and you'd be sad to remove it? :)

[03:47:00.0000] <hsivonen>
yeah :-)

[03:57:00.0000] <zcorpan>
our developers don't mind removing code :)

[03:58:00.0000] <zcorpan>
but we'd have to change :lang() in that case too, no point being inconsistent

[03:58:01.0000] <Hixie>
content-language pragma defined

[04:11:00.0000] <annevk>
zcorpan, theoretically :lang() would match the language of the element, in which case removing support for content-language would "just work"

[04:18:00.0000] <annevk>
Hixie, "This pragma not exactly equivalent to the HTTP"

[04:21:00.0000] <Hixie>
thx

[04:23:00.0000] <annevk>
Hixie, the first note in "Dynamic markup insertion" has that other error I was trying to get you to fix

[04:23:01.0000] <annevk>
Hixie, the relevant bit reads "The outerHTML and insertAdjacentHTML attributes, on the other hand, only apply to Element nodes."

[04:25:00.0000] <Hixie>
already fixed

[04:26:00.0000] <zcorpan>
Hixie: the spec doesn't ban multiple content-language declarations

[04:26:01.0000] <zcorpan>
Hixie: also perhaps it should require <html lang> to match

[04:27:00.0000] <annevk>
Hixie, also, that's the only place the method is mentioned without parenthesis

[04:27:01.0000] <zcorpan>
(if present)

[04:27:02.0000] <Hixie>
yes it does, "There must not be more than one meta element with any particular state in the document at a time."

[04:27:03.0000] <zcorpan>
oh

[04:27:04.0000] <Hixie>
annevk: hm, good point

[04:27:05.0000] <Hixie>
zcorpan: i think matching the lang="" is unnecessary, though i agree a warning might be useful there if henri wants to give one

[04:28:00.0000] <annevk>
(another inconsistency I noticed was document.write() versus just write(); most methods are defined without accessor in front of them in the rest of the spec)

[04:28:01.0000] <Hixie>
yeah, i use the full name for a handful of them because that's how most people refer to them

[04:28:02.0000] <Hixie>
window.open() is the same iirc

[04:28:03.0000] <annevk>
window.open() has both styles

[04:28:04.0000] <Hixie>
yeah

[04:29:00.0000] <annevk>
it's <dfn>open()</dfn> but <dfn>document.write()</dfn>, etc.

[04:29:01.0000] <annevk>
oh well

[04:29:02.0000] <annevk>
maybe I should care about that after CR :)

[04:29:03.0000] <Hixie>
indeed :-D

[04:30:00.0000] <Hixie>
ok bed time

[04:30:01.0000] <Hixie>
nn

[04:32:00.0000] <hsivonen>
/me wishes Java allowed goto

[04:33:00.0000] <hsivonen>
all this high-levelness interferes with writing state machines

[04:42:00.0000] <zcorpan>
Hixie: are you saying that IE6 has less than 1% of the browser install base? :)

[05:19:00.0000] <zcorpan>
/me notes that <html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="sv"> validates in v.nu under xhtml 1.0

[05:21:00.0000] <hsivonen>
zcorpan: which parser?

[05:22:00.0000] <zcorpan>
hsivonen: xml

[05:22:01.0000] <hsivonen>
makes sense

[05:22:02.0000] <zcorpan>
i had expected it to complain about the language mismatch

[05:22:03.0000] <hsivonen>
so many legacy holes to plug

[05:25:00.0000] <hsivonen>
zcorpan: recorded: http://bugzilla.validator.nu/show_bug.cgi?id=286

[05:25:01.0000] <hsivonen>
thanks

[05:59:00.0000] <hsivonen>
performance tuning is hard

[06:08:00.0000] <annevk>
argh, I thought my camera could handle CompactFlash but it can't...

[06:08:01.0000] <annevk>
anyone in need of a CF card?

[06:12:00.0000] <hsivonen>
lesson learned today: never try doing two performance "optimizations" without benchmarking in between

[06:15:00.0000] <hsivonen>
another lesson learned today: it's more efficient to copy data than to check if data needs to be copied

[06:17:00.0000] <jgraham>
annevk: What CF card?

[06:17:01.0000] <jgraham>
/me was thinking of getting one

[06:17:02.0000] <jgraham>
(more)

[06:18:00.0000] <jgraham>
hsivonen: That latter lesson seems rather situation-specific

[06:19:00.0000] <hsivonen>
jgraham: very likely, yes.

[06:19:01.0000] <hsivonen>
anyway, I implemented copy avoidance for names and comments

[06:19:02.0000] <hsivonen>
and attribute values without entity refs

[06:20:00.0000] <hsivonen>
so that copy happened late only if buffer boundary fell inside the value of of an attribute value had a & in it

[06:21:00.0000] <hsivonen>
turns out that checking for each character if mode is copying or copy avoidance sucked big time

[06:22:00.0000] <hsivonen>
so at least doing an 'if' per charecter sucked

[06:22:01.0000] <hsivonen>
in non-Java languages, function pointers could be used to swap the appending function or something

[06:23:00.0000] <annevk>
jgraham, SanDisk 16GB Extreme III

[06:25:00.0000] <annevk>
/me is off to buy a SDHC card

[06:25:01.0000] <jgraham>
annevk: Wow that's big. 

[06:25:02.0000] <annevk>
it was the biggest I could find, but apparently it didn't work for my camera :/

[06:25:03.0000] <jgraham>
I'm happy to buy it off you assuming it's not too expensive 

[06:26:00.0000] <annevk>
I paid 145 EUR

[06:26:01.0000] <jgraham>
Wow, that is really quite expensive

[06:26:02.0000] <annevk>
is it?

[06:26:03.0000] <jgraham>
Isn't it cheaper to get more smaller cards?

[06:26:04.0000] <hsivonen>
2 GB is like 12 euros

[06:27:00.0000] <jgraham>
annevk: I just mean in absolute terms

[06:27:01.0000] <annevk>
/me compared it with 2 8GB cards and that didn't matter

[06:28:00.0000] <annevk>
I wanted a big one so I didn't have to swap and could do HD movies and such

[06:28:01.0000] <annevk>
seems that SDHC only maxes at 8 GB though for a lot less

[06:28:02.0000] <jgraham>
I bought a not-quite-as-good 4Gb card for £15ish which is like 80 euros for 16Gb

[06:29:00.0000] <jgraham>
The advantage of more smaller cards is that losing one in't such a big deal

[06:31:00.0000] <annevk>
well, mine would basically never leave my camera :)

[06:31:01.0000] <annevk>
anyways, really going now

[07:13:00.0000] <annevk>
ok, that was expensive, I should see into returning it

[07:14:00.0000] <annevk>
just bought SDHC 8GB Ultra II for 32 EUR

[08:00:00.0000] <zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cstyle%3Eembed%2C%20object%2C%20iframe%20%7B%20width%3A123px%3B%20height%3A45px%3B%20%7D%3C%2Fstyle%3E%0D%0A%3Cscript%3E%0D%0Aonload%3Dfunction()%7B%0D%0Avar%20e%20%3D%20document.body.firstChild%3B%0D%0Ado%20%7B%0D%0Aw(e.nodeName%20%2B%20'%3A%20'%20%2B%20e.width%20%2B%20'%2C%20'%20%2B%20e.height)%0D%0A%7D%20while%20(e%20%3D%20e.nextSibling)%3B%0D%0A%7D%0D%0A%3C%2F

[08:00:01.0000] <zcorpan>
script%3E%0D%0A%3Cembed%3E%3Cobject%3E%3C%2Fobject%3E%3Ciframe%3E%3C%2Fiframe%3E%3Cimg%3E%3Cvideo%3E%3C%2Fvideo%3E

[08:00:02.0000] <zcorpan>
(uploaded)

[09:45:00.0000] <jgraham>
It is the intent of the spec that <video><source type=type1><source type=type2><object></object></video> won't display the <object> if the browser doesn't support either type1 or type2 videos, right? (where is this specified normatively. I could only obviously see a should-level requirement when I scanned the spec). Would displaying the fallback in this case be helpful for migration from non <video> browsers without requiring a js-api for detecting 

[09:50:00.0000] <Lachy>
jgraham, IIRC, the fallback for <video> is only intended for browsers that don't support the element at all. The spec says "Content may be provided inside the video element. User agents should not show this content to the user"

[09:50:01.0000] <Lachy>
I'm not sure why it only says should instead of must, though

[09:52:00.0000] <jgraham>
Lachy: Yeah that was the requirement I found. The use case it doesn't address is the wikimedia one where you want to use <video>+ogg for firefox, want to use an applet+ogg for <video>-supporting Safari and can't transcode for some reason, possibly legal

[09:53:00.0000] <jgraham>
(so <video>+mp4 for Safari is not possible)

[09:53:01.0000] <jgraham>
(where mp4 is H.264 or whatever it's called)

[11:40:00.0000] <zcorpan>
/me reads http://www.w3.org/MarkUp/2008/ED-xhtmlmime-20080618/

[11:41:00.0000] <zcorpan>
/me is amused that the abstract contains rfc2119 words

[11:45:00.0000] <zcorpan>
"Elements and attributes in those document types belong to the XHTML namespace"

[11:57:00.0000] <zcorpan>
"However, authors should be aware that such a document [served as 'application/xml'] may not always be processed as XHTML (e.g. hyperlinks may not be recognized), depending on user agents."

[11:59:00.0000] <zcorpan>
"Authors SHOULD explicitly identify the XHTML namespace through the namespace declaration when they serve an XHTML Family document as 'application/xml' to facilitate the chance for reliable processing."

[11:59:01.0000] <zcorpan>
well i guess i could quote the whole thing here, it's all amusing

[12:11:00.0000] <zcorpan>
appendix c becomes appendix a

[12:14:00.0000] <zcorpan>
"The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119]."

[12:14:01.0000] <zcorpan>
later on:

[12:14:02.0000] <zcorpan>
"Like all of this document, this Appendix is informative. It contains no absolute requirements, and should NEVER be used as the basis for creating conformance nor validation rules of any sort. Period."

[12:17:00.0000] <zcorpan>
more consice A1: Rationale: They are rendered in some crappy mobile browsers and they'll trigger quirks mode in IE

[12:20:00.0000] <zcorpan>
A2: Rationale: Netscape Navigator 4 would think the slash is part of the tag name when there are no attributes. Also HTML UAs interpret </br> as <br>.

[12:22:00.0000] <zcorpan>
A3: Rationale: since HTML UAs just ignore the slash it's interpreted as a start tag.

[12:24:00.0000] <zcorpan>
A4 is just bogus

[12:25:00.0000] <zcorpan>
A5 seems partially bogus

[12:26:00.0000] <zcorpan>
A7 doesn't give rationale for the DO NOT

[12:27:00.0000] <zcorpan>
nor A8

[12:30:00.0000] <zcorpan>
A11 is wrong

[12:31:00.0000] <zcorpan>
A12 has bogus rationale

[12:32:00.0000] <zcorpan>
A13 seems backwards

[12:33:00.0000] <zcorpan>
A14 has bogus rationale

[12:57:00.0000] <Hixie>
BlueG: the mailman bug was fixed, you should be able to subscribe now

[15:20:00.0000] <zcorpan>
hsivonen: bookmarklet for v.nu: http://tinyurl.com/6jvpmv

[15:21:00.0000] <zcorpan>
works for both html and xhtml

[15:27:00.0000] <zcorpan>
http://tinyurl.com/65o6bz works with svg too

[15:30:00.0000] <zcorpan>
sorry, http://tinyurl.com/5mmt3l

[15:57:00.0000] <zcorpan>
should we expect episode 2 today?

[16:26:00.0000] <Hixie>
oh jesus

[16:26:01.0000] <Hixie>
kristof and garett talking to each other

[16:37:00.0000] <roc>
I thought that was part of your clever plan

[16:39:00.0000] <Hixie>
it does keep them occupied and away from discussing other things

[16:39:01.0000] <Hixie>
but i'm not sure it's wise


2008-08-13
[18:36:00.0000] <Hixie>
i accidentally wrote </li> instead of </ol> and the validator said "error fatal: Too many messages" after spewing about one error per paragraph in html5

[18:36:01.0000] <Hixie>
i may have to admit that the html5 parser's recovery is not necessarily what a validator should try to do

[23:55:00.0000] <heycam>
"A person's name is not the title of a work — even if people call that person a piece of work" -- heh

[01:09:00.0000] <hsivonen>
Hixie: I found obsolete attributes useful today: http://hsivonen.iki.fi/performance-mistake/

[01:10:00.0000] <hsivonen>
Hixie: what would have been the right way to do the table cell alignment?

[01:11:00.0000] <hsivonen>
huh? is the W3C establishing a new *Activity* for EOT?

[01:22:00.0000] <virtuelv>
hsivonen: url? (that eot thing)

[01:22:01.0000] <hsivonen>
virtuelv: http://lists.w3.org/Archives/Public/www-archive/2008Aug/att-0010/EOT-charter-draft-1.html

[01:24:00.0000] <virtuelv>
/me shall refrain from commenting further

[02:08:00.0000] <zcorpan>
Hixie: "and required that if two of these attributes are specified (or if all three are specified, in text/html), they have the same value." -- afaict it's not allowed to specify all 3 per the spec

[02:09:00.0000] <zcorpan>
"Authors must not use the xml:lang attribute (that is, the lang attribute with the xml prefix in the http://www.w3.org/XML/1998/namespace namespace) in HTML documents."

[02:10:00.0000] <hsivonen>
xml:lang causes so much trouble

[02:11:00.0000] <hsivonen>
in retrospect, XML 1.0 should have reserved id, class and lang

[02:13:00.0000] <zcorpan>
hsivonen: i've came to that conclusion too

[02:13:01.0000] <zcorpan>
s/came/come/

[02:14:00.0000] <hsivonen>
/me decides to start measuring the performance cost of the HTML5 tree construction layer

[03:44:00.0000] <Hixie>
hsivonen: i think we may want to expose charoff on col or td

[03:44:01.0000] <Hixie>
hsivonen: but there's not much point before UAs have any intention of implementing it

[03:45:00.0000] <Hixie>
zcorpan: yeah i guess you can't include all three on one element without violating another rule before worrying about the values anyway

[03:45:01.0000] <hsivonen>
Hixie: how about getting to use align conformingly in the mean time?

[03:45:02.0000] <Hixie>
what's wrong with the css equivalent?

[03:46:00.0000] <Hixie>
td + td { text-align: right; }

[03:47:00.0000] <Hixie>
anyway bed time now

[03:55:00.0000] <hsivonen>
Hixie: that violates the separation of content and style. I the rules in the style sheet should not have to depend on what number of columns my tables have and which columns have which alignment

[05:47:00.0000] <takkaria>
hsivonen: interesting the results you got with your performance changes

[05:48:00.0000] <takkaria>
hsivonen: because in hubbub, I was briefly toying with having a buffer only when required and switching to it if NULs/CRs/entities were encountered

[05:49:00.0000] <takkaria>
hsivonen: turns out it added a lot of extra macros, removed clarity, and it ended up easier and about as fast to do unconditional buffer writes

[05:50:00.0000] <hsivonen>
takkaria: I guess it's comforting that C is like that, too, and this isn't just a Java thing ;-)

[05:51:00.0000] <takkaria>
I wish I knew x86 well enough to speculate on why it's the case

[05:52:00.0000] <hsivonen>
today, I measured the perf cost of the HTML5 tree builder compared to an XML-ish tree builder

[05:52:01.0000] <hsivonen>
I'll blog about that later

[05:53:00.0000] <hsivonen>
executive summary: the Validator.nu HTML Parser's tokenizer is almost as fast as Xerces without the HTML tree builder complexity

[05:55:00.0000] <takkaria>
that's pretty speedy

[05:56:00.0000] <takkaria>
Hubbub's tokeniser tokenises HTML5 in about 0.5s

[05:56:01.0000] <takkaria>
which is something like 5MB/s

[05:58:00.0000] <takkaria>
however, libxml tokenises and produces a tree in 0.8s.  so there's quite a lot of scope for improvement, or so I hope

[06:03:00.0000] <hsivonen>
takkaria: do you count IO when benchmarking the tokenizer?

[06:04:00.0000] <takkaria>
I have some perf tests around which basically mmap() a file into memory and then pass the memory thus mapped to libxml2/hubbub

[06:05:00.0000] <takkaria>
I guess I could read an entire file properly rather than letting the OS do it, but I don't think it would make that much difference

[06:06:00.0000] <takkaria>
at some point it would be nice to do proper somewhat-scientific benchmarks, though right now getting hubbub fully-funtional takes more priority

[06:11:00.0000] <hsivonen>
with Java, IO details make a huge difference

[06:12:00.0000] <hsivonen>
which is why I run benchmarks from RAM when I don't want to benchmark IO

[06:14:00.0000] <hsivonen>
(I already tweaked IO earlier)

[06:14:01.0000] <hsivonen>
(It might not be a bad idea to rerun tests with IO included)

[06:27:00.0000] <takkaria>
my benchmarking is really unscientific, and consists of writing test apps and running them through `time` on the command-line

[06:28:00.0000] <takkaria>
otoh, mmap() after a couple of runs results in fairly unchanging data, so it's good enough for me

[06:32:00.0000] <hsivonen>
I do timing inside Java and warm the VM up first

[06:32:01.0000] <takkaria>
yeah, you have the advantage there :)

[06:36:00.0000] <hsivonen>
performance tuning on HotSpot is pretty annoying

[06:37:00.0000] <hsivonen>
because doing the wrong thing can lead to HotSpot not JITting stuff

[06:37:01.0000] <hsivonen>
which means about tenth the performance

[06:54:00.0000] <hsivonen>
I wonder how much I could speed up the tokenizer if I removed all the run-time configurability

[07:37:00.0000] <Lachy>
JohnResig, yt?

[07:39:00.0000] <Lachy>
JohnResig, your selectors api testsuite is trying to load "data/iframe.html" in an iframe, but it's returning 404. http://ejohn.org/apps/selectortest/data/iframe.html

[07:42:00.0000] <JohnResig>
Lachy: correct - the URL wasn't important to the test itself

[07:47:00.0000] <Lachy>
JohnResig, ok. So can the iframe be removed, or the src attribute removed?

[07:48:00.0000] <JohnResig>
Lachy: the src is probably safe to remove

[07:48:01.0000] <Lachy>
ok. It still works when I remove it.

[07:49:00.0000] <Lachy>
the test suite is actually working in gogi now, so it appears we fixed whatever bug was causing it to abort the test earlier

[07:49:01.0000] <JohnResig>
Lachy: cool

[07:49:02.0000] <Lachy>
except, there's some weird bug with the stylesheet

[07:50:00.0000] <Lachy>
If I leave this style in the page: .unitTest, .test { height: 10px; } then I don't get any scrollbar

[07:50:01.0000] <JohnResig>
weird

[07:50:02.0000] <Lachy>
it doesn't appear to be important, so I just removed it so I could see all the results

[07:51:00.0000] <Lachy>
oh, it's because body has a class of unitTest, so it's getting set to a height of 10px.

[07:52:00.0000] <Lachy>
oh, no, that's not it.

[07:59:00.0000] <Lachy>
JohnResig, was the test suite updated after support for namespaces were removed from the spec?

[08:00:00.0000] <JohnResig>
Lachy: I'm updating it now

[08:00:01.0000] <Lachy>
ok, thanks.

[08:00:02.0000] <JohnResig>
Lachy: we had a copy that we were using, I'm pushing itlive

[08:00:03.0000] <JohnResig>
*it live

[08:00:04.0000] <zcorpan>
interesting, http://simon.html5.org/test/html/semantics/video/events.htm doesn't finish loading in firefox (the parser stops after the first script block)

[08:01:00.0000] <Lachy>
alright. Let me know when it's up

[08:07:00.0000] <jgraham>
zcorpan: It seems to load OK for me, but then it crashes the browser 

[08:09:00.0000] <jgraham>
when you click play

[08:19:00.0000] <Lachy>
JohnResig, can you add a feature that allows me to show only failed test results? If you add class="pass" and class="fail" to each <li>, then add a checkbox or something that applies the style: .pass {display: none;}, or similar, that woud work

[08:26:00.0000] <JohnResig>
Lachy: done: http://ejohn.org/apps/selectortest/#target

[08:27:00.0000] <Lachy>
thanks

[08:28:00.0000] <zcorpan>
jgraham: i updated my firefox and now it loads and plays but it seems to play the video too fast

[08:30:00.0000] <jgraham>
zcorpan: Which platform? 

[08:30:01.0000] <jgraham>
/me is using today's Linux build

[08:31:00.0000] <zcorpan>
jgraham: windows

[08:31:01.0000] <Lachy>
JohnResig, that scrolling bug in gogi no longer occurs, and it looks like you changed the offending css. Did that turn out to be a bug in your CSS, or is it still something wrong with gogi that I should get fixed?

[08:31:02.0000] <zcorpan>
jgraham: but i got the same results on ubuntu yesterday

[08:31:03.0000] <zcorpan>
(played too fast)

[08:32:00.0000] <jgraham>
/me goes to file a bug

[08:32:01.0000] <JohnResig>
Lachy: probably something wrong in Gogi, but it wasn't necessary in my tests

[08:34:00.0000] <Lachy>
ok. I still have a copy of the old file, so I'll investigate it later

[08:38:00.0000] <Lachy>
cool, looks like we're failing the following: stringifying null/undefined, various attribute selectors, :enabled, :disabled, :checked, and all DocumentFragment tests

[08:41:00.0000] <Lachy>
JohnResig, can you add <label> around the Show Failing Tests checkbox

[08:43:00.0000] <JohnResig>
Lachy: done

[10:03:00.0000] <virtuelv>
question (barely related to anything, really)

[10:03:01.0000] <virtuelv>
I have an element with top:0 and height:100%

[10:04:00.0000] <virtuelv>
fixed position

[10:04:01.0000] <virtuelv>
now, assume that I use full-screen zoom

[10:05:00.0000] <virtuelv>
If I now zoom, should offsetHeight of the element change?

[10:10:00.0000] <virtuelv>
nevermind, what should happen to window.innerHeight?

[10:39:00.0000] <hsivonen>
/me finds http://apache.org/xml/features/standard-uri-conformant Default: false

[10:54:00.0000] <hsivonen>
whoa! Namespace processing in Xerces is relatively more expensive than HTML5 tree building in the Validator.nu HTML Parser

[13:19:00.0000] <jgraham>
Wow I like the comment "anything [complex enough to require workers] is [...] valuable enough to be commercial software - and therefore requiring protection against illicit copying"

[13:20:00.0000] <jgraham>
Because no one ever wrote any open source software that needs threads...

[13:23:00.0000] <Lachy>
jgraham, who said that?

[13:23:01.0000] <jgraham>
Lachy: Shannon on the whatwg list. You may have stopped reading the thread...

[13:23:02.0000] <Lachy>
yeah, I haven't read much on whatwg lately

[13:25:00.0000] <csarven>
What do you guys think: 10 GETs vs. 1 GET + 50k

[13:25:01.0000] <Lachy>
?

[13:25:02.0000] <jgraham>
I would reply and mention the existence of Erlang and use cases for long running background tasks for things like web-based image editors but I don't think it would help

[13:25:03.0000] <jgraham>
50kb?

[13:26:00.0000] <jgraham>
/me suspects that 1 get + 50kb would be faster for most people

[13:26:01.0000] <jgraham>
If that was the question

[13:26:02.0000] <jgraham>
But I have no reason for thinking that

[13:33:00.0000] <virtuelv>
csarven: ? use case?

[13:34:00.0000] <virtuelv>
but in general, unless the 1 request locks up the browser/page UI, I'd propose going for it

[13:34:01.0000] <virtuelv>
10 requests add at the very least 3kb extra of data to download

[13:35:00.0000] <csarven>
jgraham virtuelv I'm applying data: URI scheme to images in CSS.

[13:35:01.0000] <roc>
every time someone mentions Erlang I reach for my gun

[13:36:00.0000] <jgraham>
roc: ?

[13:36:01.0000] <jgraham>
/me is quite scared

[13:37:00.0000] <roc>
too many people seem to believe that Erlang invented message passing

[13:37:01.0000] <virtuelv>
csarven: I'm probably slow, given that my body is in a different timezone than the one I'm actually in, without having travelled across timezones

[13:37:02.0000] <virtuelv>
csarven: but you're including images using data:

[13:38:00.0000] <virtuelv>
if you have 50 K of image data in data: URI's, I'd say you're doing it wrong

[13:38:01.0000] <jgraham>
roc: I don't know who invented message passing but Erlang is probably the most famous example of a system that uses it heavilly, right?

[13:38:02.0000] <roc>
wrong

[13:38:03.0000] <jgraham>
So what is right?

[13:39:00.0000] <roc>
well, thanks to the Erlang fanboys you may in fact be right at this point, but that's self-fulfilling ignorance

[13:39:01.0000] <jgraham>
roc: A lot of fame seems to be based around that concept...

[13:40:00.0000] <jgraham>
Anyway, I am curious about the other, better, examples

[13:40:01.0000] <virtuelv>
csarven: if your stylesheet is big, as 50 kB is, you are at some point going to end up with FOUC on a slow connection

[13:41:00.0000] <virtuelv>
csarven: I'd rather see if the images could be combined, using some variant of CSS sprites, to keep the number of requests down

[13:41:01.0000] <roc>
Mach was extremely famous and uses message passing

[13:42:00.0000] <csarven>
virtuelv I'm already using CSS sprites

[13:43:00.0000] <csarven>
It appears to be that the resulting total weight of the page is 50k more when I use data:

[13:43:01.0000] <jgraham>
roc: That's a microkernel, right? I guess it wouldn't be the right example to use if you were trying to convince a skeptic that message passing is a good idea on the web (but I don't disagree that it's pretty famous)

[13:43:02.0000] <roc>
but well before that, there were well-known programming models like actors and CSP that are focused on shared-nothing message passing

[13:44:00.0000] <virtuelv>
csarven: then you have a lot of image data, ~250kb, or so?

[13:45:00.0000] <virtuelv>
a possible optimization at that point is to store the images at a different domain/subdomain

[13:45:01.0000] <csarven>
Actually not that much no.

[13:46:00.0000] <csarven>
I must be incorrectly calculating this.

[13:51:00.0000] <virtuelv>
hm, or I might be off here, the overhead is 3/2

[13:51:01.0000] <virtuelv>
or so

[13:57:00.0000] <csarven>
Images weigh-in about: 45k

[13:58:00.0000] <csarven>
The ones that get to be used for data:

[13:59:00.0000] <csarven>
With data:uri ~200k total page weight, without data:uri ~150k

[13:59:01.0000] <csarven>
Firebug results.

[14:09:00.0000] <zcorpan>
hmm.. it would be nice if it wasn't possible to set website and signature until you have at least 1 post on the forums

[14:12:00.0000] <zcorpan>
Hixie: you think you could comment out the "Website" and "Signature" rows in http://forums.whatwg.org/profile.php?mode=register&agreed=true ?

[14:46:00.0000] <takkaria>
hsivonen: you may find http://takkaria.org/dmoz/ useful

[14:50:00.0000] <Hixie>
zcorpan: done

[14:51:00.0000] <Hixie>
hsivonen: separation of content and style doesn't imply that the style is in a vacuum, the tyle always depends on the structure of the content.


2008-08-14
[18:08:00.0000] <Hixie>
http://lists.w3.org/Archives/Member/member-i18n-core/2008Aug/0008.html is fantastic

[18:10:00.0000] <Hixie>
woot, html5 being used on actual google pages

[18:10:01.0000] <Hixie>
e.g. http://www.google.com/privacy.html

[18:36:00.0000] <jruderman>
what does that page use from html5? the short doctype and <meta charset>?

[18:37:00.0000] <jruderman>
innerHTML? ;)

[18:38:00.0000] <Hixie>
the short doctype and the intent that it comply to html5

[18:38:01.0000] <Hixie>
as opposed to the usual google policy of "hey it works in ie"

[18:38:02.0000] <jruderman>
oh, wow, the use of <h1> and alt text at the top is beautiful

[18:45:00.0000] <jacobolus>
also, it cuts out the <head>, <body> tags, all unnecessary closing tags, etc.

[18:45:01.0000] <Hixie>
indeed

[18:45:02.0000] <jacobolus>
what's the <div></div> about?

[18:46:00.0000] <Hixie>
probably to hang style off

[18:46:01.0000] <Hixie>
but i haven't checked

[18:47:00.0000] <billmason>
Looks like it just clears the floats from the lists of specific privacy policies above it.

[18:49:00.0000] <jacobolus>
ah, it does a { clear: left; }

[18:57:00.0000] <Hixie>
jesus, ITS is over-engineered for this.

[19:25:00.0000] <Hixie>
i really should figure something out as far as an event queue is concerned

[19:25:01.0000] <Hixie>
maybe i need to define the main event loop

[22:26:00.0000] <roc>
anyone know if the @font-face font linking in Safari 3.1 imposes a same-origin restriction on linked fonts, or can you link to any font on the Web from any page?

[22:37:00.0000] <bdash>
roc: it doesn't

[22:38:00.0000] <roc>
yeah, I just tested it myself, thanks

[22:38:01.0000] <bdash>
roc: there's a FIXME in the code that suggests that the policy is not set in stone though

[22:38:02.0000] <roc>
interesting

[22:39:00.0000] <roc>
I think we might want to impose a same-origin policy

[22:59:00.0000] <hsivonen>
roc: wouldn't same-origin policy prevent new business models?

[22:59:01.0000] <roc>
huh?

[22:59:02.0000] <hsivonen>
roc: like a foundry hosting a font and metering access

[23:00:00.0000] <roc>
we could support Access Controls for font file loads if the server with the font file wants to be more permissive

[23:09:00.0000] <hsivonen>
takkaria: thanks

[23:10:00.0000] <hsivonen>
Hixie: I disagree. I think if your content and style are properly separated, your style can be written ahead of time enumerating the constructs you want to use.

[23:10:01.0000] <hsivonen>
Hixie: where a 'table with headings' is a construct

[23:11:00.0000] <hsivonen>
Hixie: not table with two columns and the second column takes percentages

[01:12:00.0000] <hsivonen>
hmm. GWT 1.5 RC2 is out

[01:12:01.0000] <hsivonen>
still using *old* versions of Gecko and WebKit

[05:18:00.0000] <hsivonen>
/me needs apache gzip negotiation for dummies

[05:26:00.0000] <zcorpan>
http://simon.html5.org/specs/web-dom-core

[05:26:01.0000] <jcranmer>
bah, how do I send an 

[05:27:00.0000] <jcranmer>
= in post data properly?

[05:27:01.0000] <zcorpan>
=3D ?

[05:36:00.0000] <gDashiva>
zcorpan: the entire spec is one huge issues block? :)

[05:40:00.0000] <zcorpan>
gDashiva: yeah

[05:40:01.0000] <zcorpan>
gDashiva: before i had notes all over the place, at least now they're all in 1 place :)

[05:41:00.0000] <gDashiva>
It does serve as a kind of pseudo-toc as well

[05:54:00.0000] <Lachy>
I have to buy myself a Drobo from the USA. It's 3 times more expensive to get the first gen (USB only, no firewire and no drives) in Norway

[06:22:00.0000] <zcorpan>
hsivonen: got the "undef error - Insecure dependency in exec while running with -T switch at /usr/share/perl5/Mail/Mailer/sendmail.pm line 22." again

[06:26:00.0000] <hsivonen>
zcorpan: :-( I don't know what to do about it.

[06:26:01.0000] <hsivonen>
zcorpan: thanks

[06:28:00.0000] <hsivonen>
JohnResig: you blogged about parsing HTML in JS earlier. I got GWT to compile the Validator.nu HTML Parser into JS: http://livedom.validator.nu/

[06:28:01.0000] <JohnResig>
hsivonen: ah, neat

[06:32:00.0000] <zcorpan>
hsivonen: cool

[06:45:00.0000] <Lachy>
hsivonen, could that be merged with Hixie's live dom viewer and made to show both the HTML5 parser and the browser DOMs side by side?

[06:46:00.0000] <hsivonen>
Lachy: that should be technically possible

[06:46:01.0000] <hsivonen>
Lachy: I don't promise to do it right now

[06:54:00.0000] <zcorpan>
/me notes that <p><b><b><b><p>x<p>x<p>x is parsed differently in webkit and html5

[06:56:00.0000] <zcorpan>
webkit ends up with 10 elements, html5 with 16

[06:58:00.0000] <zcorpan>
with <p><b><b><b><b><b><b><b><b><b><p>x<p>x<p>x<p>x<p>x<p>x<p>x<p>x<p>x it's 28 vs 100

[07:14:00.0000] <zcorpan>
/me sees window.parseHtmlDocument(textarea.value, iframe.contentWindow.document, afterParse, null);

[07:14:01.0000] <zcorpan>
hsivonen: is there any documentation on how to use the js parser?

[07:14:02.0000] <hsivonen>
zcorpan: no

[07:14:03.0000] <hsivonen>
but I can document it here

[07:14:04.0000] <zcorpan>
cool

[07:14:05.0000] <hsivonen>
there is one tested entry point

[07:14:06.0000] <hsivonen>
(the above)

[07:14:07.0000] <hsivonen>
and one untested entry point

[07:15:00.0000] <hsivonen>
oops

[07:15:01.0000] <hsivonen>
correction

[07:15:02.0000] <hsivonen>
there's only one entry point

[07:15:03.0000] <hsivonen>
the first argument is the HTML5 source as a string

[07:16:00.0000] <hsivonen>
the second argument is a DOM document object, whose children will be removed and replaced by the result of the parse

[07:16:01.0000] <hsivonen>
the fourth argument is a function object that will be called when the parse finishes

[07:16:02.0000] <hsivonen>
s/fourth/third/

[07:16:03.0000] <hsivonen>
/me can't count

[07:17:00.0000] <hsivonen>
the fourth argument is supposed to be an error reporting callback, but it's not implemented

[07:17:01.0000] <hsivonen>
so the fourth argument does nothing

[07:18:00.0000] <hsivonen>
so currently, the JS parser doesn't support fragments

[07:18:01.0000] <hsivonen>
and it can't attach to an existing tree without destroying it

[07:18:02.0000] <hsivonen>
I want to add those features in the future

[07:19:00.0000] <zcorpan>
was that what you had in mind as the second entry point?

[07:19:01.0000] <hsivonen>
the fragment entry point

[07:19:02.0000] <zcorpan>
ok

[07:19:03.0000] <zcorpan>
thanks

[07:19:04.0000] <hsivonen>
but its design sucked, so I removed it

[07:19:05.0000] <hsivonen>
note that parseHtmlDocument returns before the parse is complete

[07:20:00.0000] <zcorpan>
hence the callback?

[07:20:01.0000] <hsivonen>
to avoid hanging the browser, the parser pumps data through itself using timeouts

[07:20:02.0000] <hsivonen>
yes, hence the callback

[07:22:00.0000] <hsivonen>
oh, and calling parseHtmlDocument overwrites the write and writeln methods on the document object that is passed to it

[07:22:01.0000] <hsivonen>
and replaces them with its own implementations

[07:23:00.0000] <zcorpan>
hsivonen: this is all pretty neat work

[07:23:01.0000] <hsivonen>
thanks

[07:26:00.0000] <zcorpan>
hsivonen: does the parser support innerHTML?

[07:28:00.0000] <hsivonen>
zcorpan: the Java code support the fragment cases, but the JS version doesn't provide entry to that code and doesn't overwrite innerHTML setters

[07:29:00.0000] <zcorpan>
hsivonen: ok

[07:46:00.0000] <mpilgrim>
lachy or hsivonen or someone with blog.whatwg.org admin rights: can you create a category for "This Week in HTML 5"? (or give me admin rights so i can do it myself)

[07:47:00.0000] <Lachy>
mpilgrim, sure

[07:47:01.0000] <Lachy>
what do you want me to call it?

[07:47:02.0000] <mpilgrim>
no preference

[07:48:00.0000] <zcorpan>
html5-weekly

[07:48:01.0000] <zcorpan>
?

[07:48:02.0000] <gDashiva>
as the html5 turns

[07:48:03.0000] <mpilgrim>
as long as it gets its own feed

[07:49:00.0000] <mpilgrim>
my boss's boss's boss wants to follow it

[07:49:01.0000] <mpilgrim>
seriously

[07:49:02.0000] <Lachy>
I was thinking weekly-review

[07:49:03.0000] <mpilgrim>
weekly-review works for me

[07:51:00.0000] <Lachy>
done and I added the first post too it already

[07:51:01.0000] <mpilgrim>
thanks

[07:51:02.0000] <mpilgrim>
so, nominations for this week's episode?

[07:51:03.0000] <Lachy>
http://blog.whatwg.org/category/weekly-review/feed works

[07:51:04.0000] <mpilgrim>
hooray

[07:52:00.0000] <mpilgrim>
i've already written up the bit about the navigator object

[07:53:00.0000] <Lachy>
outerHTML was apparently added.

[07:53:01.0000] <gDashiva>
huzzah

[07:54:00.0000] <Lachy>
xml:lang and lang are now allowed in both HTML and XHTML

[07:55:00.0000] <hsivonen>
Lachy: except xml:lang in HTML and xml:lang in XHTML are totally different

[07:55:01.0000] <Lachy>
hsivonen, yeah, I know

[07:56:00.0000] <hsivonen>
Almost every XMLism adopted in XHTML 1.0 causes trouble when mapping HTML5 to XML infoset

[07:57:00.0000] <Lachy>
were there backwards compatibility issues with doing magic namespace processing for xml:lang in HTML documents?

[07:58:00.0000] <jacobolus>
hsivonen: the livedom thing is neat

[07:59:00.0000] <hsivonen>
/me goes file browser bugs exposed by it

[08:00:00.0000] <jacobolus>
hsivonen: that could presumably be used to hack really old browsers to make proper doms, no?

[08:01:00.0000] <jacobolus>
(not that there's necessarily any particular use for that)

[08:01:01.0000] <hsivonen>
jacobolus: yeah, I want to make a library that sniffs for HTML5 parsing and if not present, document.writes <plaintext>, attaches the JS parser to the DOM and parses the plaintext content

[08:02:00.0000] <hsivonen>
jacobolus: I think such a JS library would be useful for bootstrapping SVG and MathML in text/html once new browsers support HTML5 parsing so that only old browsers would take a perf hit

[08:03:00.0000] <jacobolus>
ah, good point

[08:03:01.0000] <jacobolus>
also might help for support of the various structure tags

[08:07:00.0000] <jacobolus>
hsivonen: might also be nice to show a box with conformance problems in that page, if it's possible

[08:08:00.0000] <hsivonen>
jacobolus: I could add parser-level error reporting, but porting the entire conformance checker to JS is a bit too much :-)

[08:09:00.0000] <jacobolus>
hsivonen: or maybe just a button that says "check conformance" and sends the text out to your conformance checker, putting the problems found back in a box

[08:09:01.0000] <hsivonen>
GWT is pretty smart about removing dead code, though, so I'm pretty sure the code size would go up notably if the error reporting code didn't count as dead

[08:10:00.0000] <hsivonen>
jacobolus: perhaps when a browser ships with Access-Control support

[08:33:00.0000] <hsivonen>
has anyone figured out yet if the new <a> content models have uglier consequences than the <ins> content model?

[08:43:00.0000] <hsivonen>
/me finds it amusing that Hixie changed the ins/del content models in the same check-in

[08:43:01.0000] <gDashiva>
Isn't the new <a> content model just what the de-facto content model was all along?

[08:44:00.0000] <hsivonen>
gDashiva: I mean conformance consequences

[09:23:00.0000] <kangax>
is it possible to work with canvas in rhino?

[09:38:00.0000] <mpilgrim>
http://blog.whatwg.org/this-week-in-html5-episode-2

[09:50:00.0000] <mpilgrim>
/me still can't believe he's getting paid for this

[09:52:00.0000] <MikeSmith>
mpilgrim: I think those writeups a massively good. It would so great if you can manage to keep doing them

[09:52:01.0000] <MikeSmith>
Hixie's productively might make pretty hard to keep up

[09:52:02.0000] <MikeSmith>
productivity

[09:54:00.0000] <MikeSmith>
/me wishes he knew what rhino is

[09:54:01.0000] <Dashiva>
MikeSmith: The javascript engine?

[09:55:00.0000] <MikeSmith>
Dashiva: for which browser?

[09:56:00.0000] <Dashiva>
None that I know of. It's a standalone thing. In java, even.

[09:56:01.0000] <MikeSmith>
ah

[09:56:02.0000] <zcorpan>
hsivonen: http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0D%0A%3Csvg%3E%3Cscript%3Ea%3Cfoo%3Eb%3C%2Fscript%3E

[09:57:00.0000] <zcorpan>
gives different result than http://parsetree.validator.nu/?parser=html5&content=%3C%21DOCTYPE+html%3E%0D%0A%3Csvg%3E%3Cscript%3Ea%3Cfoo%3Eb%3C%2Fscript%3E&submit=Print+Tree

[09:57:01.0000] <zcorpan>
intended?

[14:38:00.0000] <zcorpan>
http://www.brandonfrohs.com/2008/08/whatwg-they-know-html-dont-they/

[14:46:00.0000] <Hixie>
if mark's boss' boss' boss wants to follow the "This week in html5" stuff, that means my boss' boss does too

[14:46:01.0000] <Hixie>
good to know!

[14:46:02.0000] <Hixie>
guess i'd better stop slacking off

[14:47:00.0000] <Hixie>
btw next time mark asks for someone to configure the whatwg blog for him or whatever

[14:47:01.0000] <Hixie>
just make him an admin

[14:47:02.0000] <Hixie>
:-P

[15:03:00.0000] <Dashiva>
We're the html-wg, not the pretty-sites-wg!

[15:03:01.0000] <Lachy>
/me commented too

[15:03:02.0000] <Lachy>
on the blog about the desing

[15:03:03.0000] <Lachy>
*design

[15:04:00.0000] <Hixie>
originally when i read that blog post i thought he was going to compliment us on not wasting time with frivalous things

[15:04:01.0000] <Dashiva>
Although I must agree default phpbb is ugly like little else

[15:04:02.0000] <Hixie>
:-)

[15:05:00.0000] <Lachy>
the default wordpress is quite ugly too. But I don't have design skills to redesign it.

[15:05:01.0000] <jacobolus1>
Dashiva: and not just ugly on the outside

[15:05:02.0000] <Dashiva>
jacobolus: Don't remind me, please :)

[15:05:03.0000] <Lachy>
photoshop and other image editors are way too complicated for me to produce anything even remotely good looking

[15:06:00.0000] <jacobolus>
Lachy: what good looking are you trying to produce?

[15:06:01.0000] <jacobolus>
there are various simple image editors …

[15:07:00.0000] <Lachy>
jacobolus, put it this way: the design of http://lachy.id.au/ is about the limit of my design skills. I know how ugly it is, but regardless of the image editor, I don't have design skills.

[15:08:00.0000] <jacobolus>
I like the color scheme, but the design is very busy :)

[15:08:01.0000] <Lachy>
but if someone were to give me a photoshop mockup, I could cut out the images and code it in a matter of days

[15:09:00.0000] <Lachy>
initally, I liked the colour scheme too. I used a colour scheme tool to pick it. But I'm quite tired of it, since I've had the same design for 4 years now

[15:13:00.0000] <Hixie>
ok i really think i'm going to have to define some sort of event loop

[15:14:00.0000] <Hixie>
that will make things much simpler to specify in all the places where we have "and then at the next available opportunity..." stuff


2008-08-15
[19:18:00.0000] <Lachy>
/me is experimenting with IRC BNC servers...

[19:19:00.0000] <Lachy>
I'm using psybnc right now, but it's overly complicated and not working well. Does anyone know of a better alternative?

[20:26:00.0000] <Hixie>
/me ran into an rdfa person earlier

[20:27:00.0000] <Hixie>
i'm now more convinced than before that html5 is following the right course in terms of rdfa, distributed extensibility, etc

[20:29:00.0000] <Lachy>
Hixie, what did the rdfa person tell you?

[20:30:00.0000] <Hixie>
that html5 was a horrible mistake, basically

[20:30:01.0000] <Hixie>
we had a long 40 minute discussion

[20:30:02.0000] <Lachy>
can you say who it was?

[20:31:00.0000] <jcranmer>
40 minutes to say "your project SUCKS" ?

[20:31:01.0000] <Hixie>
Hal Abelson

[20:32:00.0000] <Lachy>
ok, I don't know him.

[20:32:01.0000] <Hixie>
jcranmer: no, we were discussing things like whether there was any reason to believe that authors were interested in micro-level detailing of data types (as in microformats or rdfa), etc

[21:20:00.0000] <BenMillard>
Hixie, are you here still?

[21:22:00.0000] <BenMillard>
oh, you're AFK

[21:48:00.0000] <jacobolus>
Hixie: the Hal Abelson who taught SICP with Sussman for all those years?

[21:55:00.0000] <jacobolus>
funny that hearing objections would make you more convinced. :)  I guess the objections were not compelling? :)

[22:44:00.0000] <Hixie>
jacobolus: yeah, that Hal. It wasn't so much that the objections made me more convinced, so much as articulating the counterarguments made me realise how strong the arguments in favour of our current course were compared to the arguments against

[22:54:00.0000] <BenMillard>
Hixie, at a very rough count it looks like I've bookmarked (very approximately) 900 web pages which do something interesting with their markup

[22:54:01.0000] <BenMillard>
each bookmark has maybe a sentence describing why I bookmarked it

[22:55:00.0000] <BenMillard>
I indent to scrape those entries out Firefox's bookmarks.html file and then filter the markup into something halfway sane for the web

[22:55:01.0000] <BenMillard>
this should hopefully meet the midpoint criteria for my Mozilla sponsorship on 1st September

[22:56:00.0000] <BenMillard>
after that, I can tailor the document to make it more useful for HTML5 work

[22:59:00.0000] <Hixie>
cool

[23:23:00.0000] <hdh>
"The insertRow(index) method must, when invoked on an element table section, act as follows:" should read "table section element"?

[23:24:00.0000] <Hixie>
no, "table section" is a variable in that sentence

[23:25:00.0000] <Hixie>
"The insertRow(index) method must, when invoked on an element /table section/, act as follows"

[23:25:01.0000] <hdh>
ah ok

[00:05:00.0000] <hsivonen>
zcorpan: nope. not intended. I wrote the script kick-off code in the browser DOM tree builder for HTML script and forgot that SVG script can have non-text children

[00:05:01.0000] <hsivonen>
zcorpan: thanks

[01:02:00.0000] <hsivonen>
zcorpan: http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0D%0A%3Csvg%3E%3Cscript%3Ea%3Cfoo%3Eb%3C%2Fscript%3E fixed. thanks.

[01:54:00.0000] <zcorpan>
hsivonen: oh sorry. the <script>a<foo>b thing seems to be a bug in opera

[01:56:00.0000] <hsivonen>
zcorpan: no, it was a bug in the parser

[01:56:01.0000] <hsivonen>
10:05 < hsivonen> zcorpan: 

[01:56:02.0000] <hsivonen>
http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0D%0A%3Csvg%3E%3Cscript%3Ea%3Cfoo%3Eb%3C%2Fscript%3E fixed. thanks.

[01:56:03.0000] <hsivonen>
or rather, it was a bug in the script kick-off code in the GWT tree builder

[01:57:00.0000] <Hixie>
so i was talking to the translation guys at google earlier, regarding the no-translate thing

[01:57:01.0000] <Hixie>
and i was amused to find out that we just ignor lang="" altogether for the purposes of translation

[01:57:02.0000] <hsivonen>
hmm. now that I think about it, there's a pretty serious bug in the case of <script><script> in SVG

[01:57:03.0000] <Hixie>
because it's more accurate to just autodetect the language than to use the metadata

[01:58:00.0000] <hsivonen>
yay for metadata

[01:58:01.0000] <hsivonen>
Hixie: how should scripts be run when SVG scripts elements nest?

[01:58:02.0000] <Hixie>
ask the svgwg

[02:00:00.0000] <hsivonen>
their recent tendency seems to be towards doing what HTML browsers do with scripts

[02:04:00.0000] <Hixie>
sounds good to me

[02:04:01.0000] <Hixie>
so do the same as what XHTML <script>s do when nested

[02:05:00.0000] <hsivonen>
what happens with those?

[02:06:00.0000] <hsivonen>
(I should probably know, but I don't.)

[02:06:01.0000] <Hixie>
search for "When a script block is inserted into a document, the user agent must act as follows"

[02:06:02.0000] <Hixie>
though i guess the spec as written makes xml script elements not work

[02:06:03.0000] <heycam>
hsivonen, let me know what happens for that so i can append it to my SVG <script> fixes

[02:06:04.0000] <heycam>
(if you could)

[02:07:00.0000] <hsivonen>
heycam: ok.

[02:07:01.0000] <Hixie>
since the xml parser (where is it defined? oh well) presumably inserts the script element before the contents are inserted

[02:07:02.0000] <Hixie>
and thus the scripts never actually execute

[02:07:03.0000] <Hixie>
i guess that's a bug that should be fixed

[02:07:04.0000] <Hixie>
not sure how to fix it though

[02:07:05.0000] <Hixie>
maybe script elements shouldn't be inserted until they're all parsed

[02:08:00.0000] <Hixie>
or there should be something that happens when the end tag is parsed

[02:08:01.0000] <Hixie>
the latter seems best

[02:08:02.0000] <Hixie>
file a bug

[02:08:03.0000] <hsivonen>
ok

[02:14:00.0000] <zcorpan>
hsivonen: ah. i was confused by opera's agressive caching again

[03:24:00.0000] <Lachy>
I upgraded several whatwg blog accounts to Editor, giving them the ability to moderate posts, manage categories, etc.

[03:52:00.0000] <Lachy>
Hixie, yt?

[03:53:00.0000] <Lachy>
Hixie, can you send me a copy of the latest acid3 test. It appears my copy is out of date now.

[03:56:00.0000] <Lachy>
actually, could you set up a script that creates a tarball of all the latest files when I go to acid3.acidtests.org/latest.tar.gz (or whatever)

[05:27:00.0000] <zcorpan>
hsivonen: <svg></body> makes livedom hang

[05:29:00.0000] <hsivonen>
zcorpan: that's weird. thanks.

[05:29:01.0000] <hsivonen>
http://parsetree.validator.nu/?parser=html5&content=<svg><%2Fbody> works

[05:30:00.0000] <zcorpan>
indeed

[05:33:00.0000] <zcorpan>
hsivonen: is it intentional to drop comments in livedom?

[06:12:00.0000] <hsivonen>
hmm. <svg></body> doesn't hang in Hosted Mode

[06:26:00.0000] <hsivonen>
zcorpan: I'm suspecting an unfixed variant of the same compiler bug that was more obvious earlier

[06:26:01.0000] <hsivonen>
but this is hard to debug, because now there's a new compiler bug that makes debugging hard

[07:05:00.0000] <Lachy>
JohnResig, does your selectors api test suite test elements outside of the document?

[07:06:00.0000] <JohnResig>
Lachy: yes

[13:26:00.0000] <Philip`>
http://wiki.whatwg.org/wiki/User_talk:Highhi - spam

[13:52:00.0000] <Philip`>
takkaria: The image on http://takkaria.org/dmoz/ seems purely decorative and therefore must have alt=""

[14:05:00.0000] <Philip`>
"it's more accurate to just autodetect the language than to use the metadata" - wouldn't it be even more accurate to use the metadata as an input to the autodetection heuristics, since there's bound to be a non-zero correlation between the metadata and the data and therefore it's better to use that input than to ignore it?

[14:22:00.0000] <Dashiva>
How certain is that?

[14:25:00.0000] <Philip`>
How certain is what?

[14:25:01.0000] <Dashiva>
That the correlation is positive

[14:28:00.0000] <Philip`>
Pretty certain - if lang is currently entirely random and unrelated to the actual language, then I just need to create a single web page which uses lang correctly, and then there'll be a positive correlation, and the only way it'd get back to zero correlation is if someone maliciously uploads a new page with an intentionally incorrect lang, which seems pretty unlikely

[14:29:00.0000] <Philip`>
(and if they upload two incorrect pages, then there'll be a negative correlation, which you could usefully feed into your language autodetection heuristics so that your choice is biased away from the metadata because it's most likely to be incorrect)

[14:29:01.0000] <Dashiva>
But we'd have to know which

[14:30:00.0000] <Dashiva>
There could be page creators with English hardcoded, e.g.

[14:31:00.0000] <Philip`>
If you have a million pages which say lang="en" because it's hardcoded into the editor and is unrelated to the actual language, and one page that says lang="en" because it's actually English, then the probability of a page being English given that it has lang="en" is greater than the probability of it being English given no prior knowledge

[14:32:00.0000] <Philip`>
kangax: If I'm remembering correctly that you asked about Rhino canvases, you might want to see http://rhino-canvas.sourceforge.net/

[14:33:00.0000] <kangax>
Philip`: yes, I've seen this, but when running couldn't make it produce any result

[14:33:01.0000] <Dashiva>
It'd have to be significant enough to be useful

[14:33:02.0000] <Philip`>
Dashiva: (...assuming the distribution of languages on the pages created with the lang="en" editor is identical to the distribution of languages on pages without lang, or something like that)

[14:34:00.0000] <Philip`>
Dashiva: That depends solely on your definition of "useful"

[14:34:01.0000] <Dashiva>
Useful in that there exist pages that are ambiguous without it and not with it

[14:34:02.0000] <kangax>
Philip`: eventually I found out that Jaxer (running on Rhino) works with canvas nicely

[14:34:03.0000] <takkaria>
Philip`: fixed, ta

[14:34:04.0000] <kangax>
Philip`: and implements toDataURL which I needed

[14:36:00.0000] <Philip`>
Dashiva: If providing extra information results in your autodetection algorithm having a one-in-a-million chance of giving a better result, then that seems useful if you're looking at more than a million pages, and on the web you probably are :-)

[14:36:01.0000] <Dashiva>
But it probably isn't that much

[14:36:02.0000] <Dashiva>
It has to be a million pages with ambiguous content

[14:37:00.0000] <Philip`>
kangax: Ooh, Jaxer has a canvas implementation?

[14:37:01.0000] <Dashiva>
And even then there's the cost of figuring out how much weight to give it, and implement the weighting, and suffer from false positives when you overweight :)

[14:37:02.0000] <kangax>
Philip`: yep

[14:37:03.0000] <kangax>
Philip`: it managed to run all of my code (which is getting quite complex)

[14:38:00.0000] <Philip`>
kangax: If I wasn't too lazy (which I am), I'd update http://philip.html5.org/tests/canvas/suite/tests/results.html to include that implementation, which could be interesting :-)

[14:39:00.0000] <kangax>
they are using gecko if i'm not mistaken

[14:39:01.0000] <kangax>
not sure which, though

[14:39:02.0000] <kangax>
(not sure which version)

[14:40:00.0000] <Philip`>
Oh... If it's "running on Rhino", I presume that means it's doing stuff in Java, so how does it make use of Gecko?

[14:42:00.0000] <kangax>
I think via XUL

[14:43:00.0000] <kangax>
Philip`: http://www.aptana.com/node/266

[14:43:01.0000] <kangax>
"Jaxer is built on the same technology as Firefox which means we can take advantage of XUL-like functionality on the server." <-- I assume they are talking about Gecko

[14:47:00.0000] <Philip`>
/me fails to find an obvious description of technical details on what it does

[15:03:00.0000] <Philip`>
Hixie: Are you aware that the commit-watchers emails are saying "/home/ianh/svn/webapps/hooks/commit-email.pl: `/usr/bin/svnlook diff /home/ianh/svn/webapps -r 2073' failed with this output:"?

[15:12:00.0000] <Hixie>
Philip`: i was not, investigating, thanks

[15:12:01.0000] <Hixie>
Philip`: re the language thing, my understanding is that the heuristics are so significantly more accurate than lang="" that looking at lang="" wouldn't help any measurable amount

[15:15:00.0000] <Hixie>
so er

[15:15:01.0000] <Hixie>
the problem is svn is segfaulting

[15:15:02.0000] <Hixie>
wtf

[15:20:00.0000] <Hixie>
well i have no idea how to fix that

[15:20:01.0000] <Hixie>
so i guess we'll have to live with it for now

[15:32:00.0000] <gsnedders>
Hixie: can you reply to my email about datagrid ASAP?

[15:32:01.0000] <Hixie>
sent today?

[15:32:02.0000] <gsnedders>
Hixie: Just now.

[15:32:03.0000] <gsnedders>
I only just got internet back from being on holiday :)

[15:33:00.0000] <Hixie>
The interactive content vs element thing?

[15:33:01.0000] <gsnedders>
yeah

[15:33:02.0000] <gsnedders>
(I need it so I can release spec-gen 1.0RC1 tomorrow, or maybe Sun)

[15:33:03.0000] <Hixie>
the answer to the question is yes

[15:33:04.0000] <Hixie>
but editing the spec will be a bitch cos i'm in the middle of a bazillion edits

[15:34:00.0000] <gsnedders>
Hixie: Thanks. That'll do for me. :P

[15:34:01.0000] <Hixie>
do you need the spec edited?

[15:34:02.0000] <gsnedders>
Nope.

[15:34:03.0000] <Hixie>
ok cool

[15:34:04.0000] <Hixie>
yeah that was just a typo

[15:34:05.0000] <gsnedders>
(It obviously ought to be fixed, but I don't have any need for that to be done quickly)

[15:34:06.0000] <Hixie>
i always think in terms of "bla element" like flow element, phrasing element, etc

[15:34:07.0000] <gsnedders>
Yeah, I was guessing that.

[15:34:08.0000] <Hixie>
but i have to call it content to allow text nodes to fall into it, etc

[15:34:09.0000] <Hixie>
so i keep making that mistake

[15:35:00.0000] <Hixie>
anything else i can help you with?

[15:35:01.0000] <gsnedders>
I think not with

[15:35:02.0000] <gsnedders>
Have we published the Aug draft MikeSmith was talking of?

[15:35:03.0000] <takkaria>
not yet

[15:36:00.0000] <gsnedders>
All right.

[15:36:01.0000] <Hixie>
i'll be switching to spec gen asap after we do, if it's available :-)

[15:37:00.0000] <gsnedders>
spec-gen 1.0b2 is perfectly bug-free, spec-gen 1.0RC1 will cause ID gen breakages, though

[15:37:01.0000] <gsnedders>
and RC1 really is an RC :)

[15:37:02.0000] <Hixie>
hehe

[15:37:03.0000] <Hixie>
you sure are taking this seriously :-P

[15:38:00.0000] <gsnedders>
:P

[15:38:01.0000] <gsnedders>
Hixie: Or maybe I'm just a perfectionist :P

[15:39:00.0000] <gsnedders>
/me ponders how to get around the issue that he uses XPath's normalize-space() which follows XML's S, and not HTML 5 whitespace

[15:40:00.0000] <Philip`>
I suggest getting around the issue by not caring about it

[15:40:01.0000] <Hixie>
"a documented limitation of this software..."

[15:41:00.0000] <gsnedders>
All that that affects is the whitespace in the middle of "latest version" for parsing the W3C status :P

[15:41:01.0000] <gsnedders>
s/af/ef/

[15:46:00.0000] <gsnedders>
I know I could fix it, it's just a question of with a how big of a perf. hit

[15:54:00.0000] <gsnedders>
(The ID gen changes only really effect things that play around with edge cases, like HTML5 :D)

[15:54:01.0000] <Hixie>
heh

[16:15:00.0000] <Hixie>
i guess we'll have to study how people use Content-Language

[16:31:00.0000] <gsnedders>
Hixie: As an HTTP header?

[16:31:01.0000] <Hixie>
meta

[16:32:00.0000] <gsnedders>
I can give you data for HTTP header, but not meta :P


2008-08-16
[18:41:00.0000] <Hixie>
/me mumbles something about not believing that he's actually defining a fricking event loop algorithm but not seeing any sane way around it

[04:12:00.0000] <gsnedders>
Why do I read public-html and whatwg?

[04:36:00.0000] <Dashiva>
gsnedders: Because it's relevant to your interests (and future employment, maybe even)

[04:55:00.0000] <gsnedders>
We are the Priests/Of the Temples of Syrinx

[06:58:00.0000] <jgraham>
/me was hoping that gsnedders was setting up a "why does the chicken cross the road" type joke

[06:59:00.0000] <gsnedders>
jgraham: No, I wasn't.

[06:59:01.0000] <gsnedders>
jgraham: I guess I'm relying on the fact my T-shirt says, "Maybe if this shirt is witty enough someone will finally love me" for wit.

[07:52:00.0000] <hsivonen>
gsnedders: I suggest using an HTML5 parser that can be configured to follow the infoset coercion rules and, thus, maps space characters to XML S.

[08:09:00.0000] <gsnedders>
hsivonen: I have no need for the output to match XML, though

[08:09:01.0000] <hsivonen>
gsnedders: it makes the Toolchain work

[08:10:00.0000] <gsnedders>
I realised half an hour ago that normalize-space(translate(., "\x0C", " ")) would work fine

[08:57:00.0000] <hsivonen>
it takes almost 9 minutes to compile the parser into JS on 2.4 GHz Core Duo

[09:02:00.0000] <Philip`>
hsivonen: Why does http://validator.nu/?doc=http://www.teamfortress.com/ say "Error: Required attributes missing on element map." on the <map name="home">? I don't see any attributes required other than 'name'

[09:06:00.0000] <hsivonen>
Philip`: interesting.

[09:06:01.0000] <hsivonen>
Philip`: are you seeing similar bogus messages on other elements?

[09:10:00.0000] <Philip`>
hsivonen: Not that I can see

[09:10:01.0000] <Philip`>
Maybe something (maybe me) is getting confused because XHTML requires id instead of name?

[09:14:00.0000] <hsivonen>
Philip`: oh. right. it's not the HTML5 schema

[09:15:00.0000] <Philip`>
HTML 4.01 Transitional apparently

[09:16:00.0000] <Philip`>
Sounds like that's a known issue, given http://about.validator.nu/#ideas

[09:17:00.0000] <hsivonen>
/me tries to figure out why </body> hangs in the Web mode of GWT, although the code works right in Hosted mode

[09:17:01.0000] <Philip`>
Is there any debugging support for the Web mode?

[09:19:00.0000] <hsivonen>
Philip`: Firebug :-)

[09:20:00.0000] <hsivonen>
Philip`: but it's not useful due to the obfuscation

[09:21:00.0000] <hsivonen>
(GWT RC2 introduced a bug in the non-obfuscated compiler mode, so that's no good)

[09:21:01.0000] <Philip`>
Sounds like fun

[10:20:00.0000] <gsnedders>
anyone from the US around?

[10:37:00.0000] <hsivonen>
yay. the GWT compiler bug causing the hang is one of those bugs that stops being a bug when observed

[10:37:01.0000] <hsivonen>
(inserting alerts fixes the problem--presumably by foiling some compiler "optimizations")

[10:39:00.0000] <gsnedders>
nvm

[12:30:00.0000] <Philip`>
/me encounters https://mail.mozilla.org/pipermail/es-discuss/2008-August/003400.html

[12:33:00.0000] <Philip`>
(which indicates still requiring content-type opt-in to the new version, but promoting putting it in <script type> instead of HTTP content-type)

[15:27:00.0000] <Xenos>
Don't everybody talk at once, now

[15:42:00.0000] <Lachy>
Xenos, it's usually quiet in here on weekends

[15:43:00.0000] <Xenos>
Ahh

[15:43:01.0000] <gsnedders>
Until someone explodes, at least

[15:44:00.0000] <Lachy>
fortunately, exploding people are an infrequent occurrence. Though, it's messy when it does happen.

[15:44:01.0000] <gsnedders>
(Don't worry about me, I'm just talking about a mixture of Nabokov on IM, primarily Lolita and Ada or Ardor)

[15:45:00.0000] <Xenos>
How do you represent exploding people in text format?

[15:45:01.0000] <gsnedders>
/me explodes

[15:45:02.0000] <Lachy>
*BOOM*

[15:45:03.0000] <gsnedders>
Xenos: ^^

[15:45:04.0000] <Xenos>
Ahh

[15:45:05.0000] <gsnedders>
Either work.

[15:45:06.0000] <Xenos>
Doesn't seem so messy to me :p

[15:45:07.0000] <gsnedders>
You can't see myself or Lachy now.

[15:46:00.0000] <Lachy>
Xenos, you'd think differently if you saw my screen right now

[15:46:01.0000] <Lachy>
:-)

[15:46:02.0000] <Xenos>
This is true. I am impressed by your disembodied selves still handling your keyboards, though :p

[15:47:00.0000] <gsnedders>
Xenos: I don't have a keyboard, I just have diodes attached to various bits of brain splattered across this room

[15:47:01.0000] <Xenos>
Ah :P

[15:47:02.0000] <gsnedders>
Now, let me try and reassemble myself so I can go to bed.

[15:48:00.0000] <Xenos>
hehe

[15:52:00.0000] <gsnedders>
/me is wondering if he can be bothered to not be lazy for long enough to apply to uni in the US

[15:53:00.0000] <Xenos>
Why would you apply for uni in the US?

[15:53:01.0000] <Xenos>
I mean, rather than somewhere civilized :p

[15:53:02.0000] <gsnedders>
IMO (which by definition is not humble), there are better places for CS in the US than anywhere else

[15:54:00.0000] <gsnedders>
And it really would be a successful means of running away from everyone I know

[15:54:01.0000] <Xenos>
Hmm, I thought there were Indian places that were supposed to be better

[15:54:02.0000] <Xenos>
don't remember any names tho

[15:54:03.0000] <Xenos>
That it would, I guess :P

[15:55:00.0000] <gsnedders>
But I don't speak Indian :)

[15:55:01.0000] <gsnedders>
(There again, I doubt many people from my school would be going to Cambridge, for example)

[15:55:02.0000] <Xenos>
English is an official language there, I'd guess most CS education happens in that :P

[15:56:00.0000] <gsnedders>
And Latin is the official language of the Vatican, but not everything is in Latin there :P

[15:56:01.0000] <Xenos>
hehe

[15:56:02.0000] <Xenos>
There are some regions that don't even speak Hindi

[15:56:03.0000] <gsnedders>
Like Fife, Scotland?

[15:57:00.0000] <Xenos>
lol

[15:58:00.0000] <gsnedders>
Or is that not the region you meant? :P

[15:59:00.0000] <Xenos>
Wow, there are 415 different mother tongues within India, 29 of which are spoken by more than 1 million people, 122 by 100k+

[15:59:01.0000] <gsnedders>
Well, I can speak English. I can sorta speak French, and I can understand some German.

[16:00:00.0000] <gsnedders>
Oh, and I can read a fair amount of Latin.

[16:01:00.0000] <Xenos>
Then I guess Hindi is next on the list? :P

[16:02:00.0000] <gsnedders>
Nah.

[16:02:01.0000] <gsnedders>
I hate langauges :)

[16:02:02.0000] <Xenos>
hehe

[16:02:03.0000] <gsnedders>
Besides, "We're only immortal for a limited time", so why spend that time learning languages? :)

[16:03:00.0000] <Xenos>
:p

[16:03:01.0000] <Xenos>
It keeps the alzheimers away

[16:03:02.0000] <Xenos>
:p

[16:03:03.0000] <gsnedders>
(That's an awesome lyric, though)

[16:12:00.0000] <Xenos>
http://current.com/items/89204971_death_star_over_san_francisco nice

[16:13:00.0000] <gsnedders>
Photoshop'd!

[16:14:00.0000] <Xenos>
hehe

[16:14:01.0000] <Xenos>
I didn't know you could edit vids there :p

[16:17:00.0000] <gsnedders>
No, you can't, but meh!

[16:17:01.0000] <Xenos>
Thought it was kinda cool anyway :p

[16:25:00.0000] <gsnedders>
The longer anything goes on for, the less it makes sense.

[16:25:01.0000] <Xenos>
How so?

[16:26:00.0000] <gsnedders>
/me is just quoting myself out of context

[16:28:00.0000] <gsnedders>
(to give it a slight bit of context, there's a girl who constantly flirts with me, and just generally acts as if she wants me, denies it and will do nothing with me)

[16:29:00.0000] <Xenos>
That makes sense, every time you do anything in return that makes it seem that you want her validates her

[16:29:01.0000] <Xenos>
Boosts self-esteem, makes her feel more attractive, etc. She might not even be consciously aware of continually flirting

[16:30:00.0000] <gsnedders>
/me has already got into far too long discussions about this

[16:30:01.0000] <Xenos>
Anyhow, it doesn't not make sense :p

[16:31:00.0000] <gsnedders>
"talking to you is like taking a ride in a mobius loop" — molly

[16:31:01.0000] <gsnedders>
(as it always ends up back on the subject of that girl :P)

[16:32:00.0000] <gsnedders>
/me is hopeless

[16:32:01.0000] <gsnedders>
(Actually, I'm not, I'm far too hopeful, but I digress, and I ought to shut up now)

[16:33:00.0000] <Xenos>
You really like this girl, eh?

[16:33:01.0000] <gsnedders>
Xenos: I have shut up.

[16:33:02.0000] <Xenos>
The general rule in such situations is GFTOW.

[16:33:03.0000] <Xenos>
Go F*ck Ten Other Women. If you're still in love with her after you might have real problems :P

[16:35:00.0000] <gsnedders>
… Xenos says completely naïve of the fact I've just got out of months of seeing a psychologist :P

[16:35:01.0000] <Xenos>
Good. Most people should see a good psychologist for some time :p

[16:36:00.0000] <Xenos>
And if you've quit that means you're better, right?

[16:36:01.0000] <gsnedders>
In terms of what was causing me to be suicidal, yes

[16:40:00.0000] <Lachy>
That death star video is awesome :-)

[16:41:00.0000] <Xenos>
Ya :p

[16:41:01.0000] <Lachy>
it's quite well made. Some of the shots look quite realistic

[16:41:02.0000] <gsnedders>
But only some :(

[16:42:00.0000] <Xenos>
Yup

[16:42:01.0000] <Xenos>
Some

[16:42:02.0000] <Lachy>
yeah. The one with the landing didn't quite work well due to some minor camera movement

[16:42:03.0000] <Lachy>
you can see the ship shifting slightly as the camera moved

[16:47:00.0000] <Xenos>
g'nite guys

[16:50:00.0000] <jgraham>
The lighting's wrong :)

[16:50:01.0000] <jgraham>
(on the death star)

[16:55:00.0000] <gsnedders>
http://xkcd.com/331/


2008-08-17
[05:51:00.0000] <annevk>
https://mail.mozilla.org/pipermail/es-discuss/2008-August/003400.html sounds promising

[05:53:00.0000] <annevk>
already mentioned it seems

[05:53:01.0000] <annevk>
/me reads logs

[06:15:00.0000] <annevk>
btw, anyone going to be in Cambridge, UK next week who is up for meeting?

[06:15:01.0000] <annevk>
the CSS WG is meeting Wednesday-Friday and I'll be there those days

[06:58:00.0000] <gsnedders>
annevk: AFAIK jgraham and Philip` are the only ones there. Pinging them directly may work better :)

[07:07:00.0000] <annevk>
if this doesn't work out I'll do that

[07:08:00.0000] <annevk>
hmm, no new Ubuntu security updates in a week, impressive

[07:28:00.0000] <Philip`>
/me will most likely be in Cambridge then

[07:29:00.0000] <gsnedders>
/me will be at school :(

[07:30:00.0000] <annevk>
meeting in the evening would be best for me

[07:31:00.0000] <gsnedders>
annevk: You better not ask Philip` where or when to meet :)

[07:31:01.0000] <annevk>
I'm leaving Saturday morning and I believe there's at least some group activity planned for Wednesday but other days work

[08:39:00.0000] <jgraham>
/me is in sunny Cambridge all next week and would quite happily meet up with annevk

[08:39:01.0000] <jgraham>
/me assumes that will get noticed in the logs

[08:49:00.0000] <gsnedders>
/me wishes it was sunny here

[08:50:00.0000] <jgraham>
/me was being slightly ironic (although it isn't too bad today, it did rain most of last week)

[08:50:01.0000] <gsnedders>
/me notes it is normally far more sunny in Cambridge than it is here

[08:53:00.0000] <takkaria>
jgraham: I like your latest pictures on flickr... how did you go about getting http://www.flickr.com/photos/jgraham/2737000236/in/photostream/?

[08:53:01.0000] <jgraham>
/me notes that this is gsnedders fault for living somewhere with such notoriously bad weather as Scotland

[08:53:02.0000] <gsnedders>
jgraham: The east cost has less rain than Rome!

[08:53:03.0000] <gsnedders>
*coast

[08:54:00.0000] <gsnedders>
jgraham: Besides, it's my parents fault!

[08:57:00.0000] <jgraham>
takkaria: The flowers one was just my 60mm macro lens wide open and focussed through some small foreground flowers to give the hazy effect at the front. I had the square crop in mind and used lightroom to do the split toning and to add the vignetting

[08:57:01.0000] <jgraham>
Lightroom is awesome. 

[08:57:02.0000] <jgraham>
The things that it does well it does so easilly that it feels like cheating

[09:03:00.0000] <gsnedders>
/me needs to decide what to do for his computing project!

[09:03:01.0000] <gsnedders>
/me is too indecisive

[09:03:02.0000] <jgraham>
/me thought gsnedders was trying to decide this over a week ago

[09:04:00.0000] <gsnedders>
/me was

[09:04:01.0000] <gsnedders>
/me is still too indecisive :)

[09:05:00.0000] <jgraham>
/me suggests gsnedders decides before the deadline for completing it

[09:05:01.0000] <gsnedders>
:)

[09:20:00.0000] <takkaria>
jgraham: ah, interesting, thanks

[09:21:00.0000] <takkaria>
gsnedders: I suggest nothing that'll ever be useful, because it's kind of heartbreaking to spend months on a computing project just to throw it away

[09:21:01.0000] <jgraham>
takkaria: Why the requirement that you throw it away?

[09:22:00.0000] <takkaria>
well, you don't throw it away, it just never gets used by anyone

[09:22:01.0000] <takkaria>
for my A-level computing project I wrote a web-based accounting package for a debt management company

[09:22:02.0000] <takkaria>
sadly, my brother left the place before I'd finished it

[09:23:00.0000] <takkaria>
and that was that, it never got used

[09:23:01.0000] <jgraham>
I guess it depends what you do. Is there some requirement that you can't release it as open source afterwards

[09:23:02.0000] <jgraham>
for example?

[09:23:03.0000] <takkaria>
I suppose so

[09:23:04.0000] <gsnedders>
jgraham: Here, there isn't.

[09:24:00.0000] <jgraham>
/me didn't do A-level computing but the projects that people did all seemed to be boring database+VBA type jobs

[09:24:01.0000] <jgraham>
s/people/people he knew/

[09:24:02.0000] <takkaria>
but afaik most computing syllabusses tend to make doing anything that's vaguely interesting quite hard

[09:24:03.0000] <takkaria>
at least, all the ones I've seen have

[09:24:04.0000] <jgraham>
Maybe Scotland is better

[09:25:00.0000] <gsnedders>
Oh, the syllabus here bores me

[09:27:00.0000] <jgraham>
Well I guess they have the requirement that unless 30% of people can get an A then its too hard

[09:28:00.0000] <gsnedders>
jgraham: AH is harder than A-Level :P

[09:28:01.0000] <jgraham>
gsnedders: Even so

[10:31:00.0000] <gsnedders>
jgraham: I think I'll just do the variation of your idea, i.e., a feed reader with sorting by interest

[10:39:00.0000] <Philip`>
gsnedders: I suggest not deciding on a project - just develop a framework within which you could develop your project in the hypothetical event that you decide on one

[10:39:01.0000] <Philip`>
Everyone loves frameworks

[10:45:00.0000] <Philip`>
I have an idea for a very useful program: One where you take a photo of a collection of many nearly-identical-looking unpaired socks, and the program analyses the image and tells you how to pair them up correctly

[10:46:00.0000] <jgraham>
Develop a machine learing algorithm that cn deduce whether you are likely to enjoy a particular project based on other projects that you have started

[10:48:00.0000] <jcranmer>
or finished

[11:17:00.0000] <hsivonen>
zcorpan: <svg></body> in live dom fixed. thanks.

[11:17:01.0000] <hsivonen>
(the fix was putting in something that does nothing but that the GWT compiler can't prove does nothing, thus protecting stuff immediately after getting optimized away)

[11:18:00.0000] <hsivonen>
s/getting/from getting/

[15:15:00.0000] <jgraham>
/me wonders why people keep asking for clarifications of the alt section

[15:18:00.0000] <jgraham>
It seems pretty clear at the moment. I can't decide if there is a real lack of understanding or if it is because having a quote from Hixie saying somehing like "images in emails targeted at known recipients are not required to contain alt text" is has greater moral weight than the spec saying it

[15:19:00.0000] <jgraham>
s/is//

[15:31:00.0000] <jgraham>
I guess 4.7.2.1.3. could be clear that logos always require alt text unless the provisions of 4.7.2.1.10. apply

[15:33:00.0000] <jgraham>
It should probabaly also be made clear what happens if the author doesn't think that any of the provisions apply

[15:33:01.0000] <jgraham>
i.e. if they have some use of an image that isn't explicitly mentioned

[16:02:00.0000] <gsnedders>
jgraham: There shouldn't be trailing full stop characters in the section numbers! That's breaking ISO 2145! :P


2008-08-18
[23:48:00.0000] <Dashiva>
/me wonders how othermaciej handled the Garret-Kris spiral of misconceptions

[23:54:00.0000] <jruderman>
Hixie: i have some questions about https://bugzilla.mozilla.org/show_bug.cgi?id=450981

[23:55:00.0000] <jruderman>
Hixie: when the spec says 'meta elements must be inside a head element', does that mean user agents are supposed to ignore them when they appear elsewhere?

[23:55:01.0000] <jruderman>
Hixie: can you cook up some stats on how many sites would break if we made that change?

[23:56:00.0000] <Hixie>
when the spec says "meta elements must be inside a head element" it only means that head elements, to be conforming, have to have a head element ancestor

[23:56:01.0000] <Hixie>
it doesn't mean anything to do with how browsers work

[23:59:00.0000] <jruderman>
thanks

[00:00:00.0000] <Hixie>
commented

[00:00:01.0000] <Hixie>
and invalidated

[00:01:00.0000] <jruderman>
Hixie: it was originally filed as an anti-XSS measure, not an HTML5 conformance thing

[00:01:01.0000] <jruderman>
Hixie: but i guess HTML5 actually requires not fixing it

[00:02:00.0000] <Hixie>
oh, the subject was changed? man i hate it when people do that :-)

[00:02:01.0000] <Hixie>
but yeah, technically html5 requires it to work anywhere

[00:02:02.0000] <Hixie>
maybe that should be changed

[00:02:03.0000] <Hixie>
though i know that at least for the character encoding, we have to support it anywhere

[00:03:00.0000] <jruderman>
if evil people can screw with character encoding they can probably sneak an XSS in through bytes that don't look like '<' but turn into '<'

[00:04:00.0000] <jruderman>
so you've seen a number of documents with meta charset in <body>?

[00:05:00.0000] <Hixie>
meta charset in <body> happens all the time

[00:05:01.0000] <jruderman>
fun

[00:05:02.0000] <Hixie>
thing is the <body> starts as soon as you see any character

[00:05:03.0000] <Hixie>
so e.g. in <html>a<head><meta charset=""></head><body>...</body></html> the meta is in the body

[00:06:00.0000] <jruderman>
interesting. i've seen pages like that.

[00:06:01.0000] <jruderman>
i've also gotten frustrated by how whitespace seems to move around near the head ;)

[00:07:00.0000] <Hixie>
yeah

[00:09:00.0000] <jruderman>
do you want to comment in the bug again, or should i? it should probably be reopened with the summary reverted, with a suggestion of WONTFIX

[00:09:01.0000] <Hixie>
i can

[00:10:00.0000] <jruderman>
thanks

[00:10:01.0000] <Hixie>
done

[00:13:00.0000] <jruderman>
/me changes the summary again

[00:53:00.0000] <hdworak>
hi. does html5lib handle CDATA?

[00:54:00.0000] <hdworak>
I mean CDATA sections, like <![CDATA[<sender>John Smith</sender>]]>

[01:37:00.0000] <othermaciej>
Dashiva: at some point, I stopped reading

[01:55:00.0000] <Hixie>
hdworak: HTML5 doesn't have <![CDATA[]]> blocks (except in embedded mathml, but that's probably not what you're asking about)

[02:07:00.0000] <hsivonen>
at some point, someone needs to define case-sensitivity of role values

[02:08:00.0000] <hsivonen>
even if only to say that they aren't like other enumerated values

[03:07:00.0000] <gsnedders>
/me has proved once again on public-html he's an idiot

[03:19:00.0000] <hdworak>
Hixie: ok, thank you

[03:24:00.0000] <hdworak>
Hixie: so does that mean that from HTML 5 point of view, it is just ill-formed mark-up?

[03:30:00.0000] <Philip`>
hdworak: Yes - it's treated as a (bogus) comment, and the parser just reads up until the next ">" character after the "<!"

[03:30:01.0000] <hdworak>
Philip`: thank you

[03:31:00.0000] <Philip`>
so it should be parsed the same as <!--[CDATA[<sender-->John Smith</sender>]]>

[03:34:00.0000] <hdworak>
Philip`: ok, but I'm using your lib to parse any HTML/XHTML not just HTML 5, so I'm just going to translate all CDATA to text beforehand

[03:35:00.0000] <hdworak>
:)

[03:37:00.0000] <gsnedders>
hdworak: any HTML should be parsed per the HTML 5 rules

[03:37:01.0000] <Philip`>
hdworak: If it's actually really proper XHTML, you should probably use an XML parser instead; otherwise you should probably just use html5lib without any special translations, since it ought to match how browsers parse pages

[03:38:00.0000] <gsnedders>
(where proper XHTML is served as application/xhtml+xml, application/xml, or text/xml, NOT text/html)

[03:39:00.0000] <hdworak>
well, I'm then using a regex to match any <!-- <rdf:RDF ... comments, which are the deprecated way to embedded RDF/XML in the code

[03:39:01.0000] <hdworak>
if it's valid, I'm doing try: dom = minidom.parseString(code)

[03:39:02.0000] <hdworak>
first

[03:39:03.0000] <gsnedders>
hdworak: Where's the data coming from?

[03:39:04.0000] <hdworak>
if it fails, I'm doing html5lib

[03:39:05.0000] <hdworak>
direct input, file upload, remote URI

[03:40:00.0000] <gsnedders>
hdworak: For the first two, I'd make it user selectable. For the third, it should be done from Content-Type, optionally with content-type sniffing

[03:40:01.0000] <hdworak>
so, because I'm using regex to match this commented <rdf:RDF, I do need to take care of CDATA, because if this appears in CDATA, it should not be parsed at all

[03:41:00.0000] <gsnedders>
hdworak: de-facto CDATA doesn't exist in any HTML version, so that's irrelevant. What you need to know is whether the input is HTML or XML, nothing else.

[03:41:01.0000] <hdworak>
gsnedders: I understand, but IMHO the content-type really does not matter for this particular application

[03:41:02.0000] <hdworak>
I'm not doing a general purpose parser, but use it for extracting very specific information

[03:42:00.0000] <gsnedders>
/me notes that feeds got to the stage at one point where you absolutely had to completely ignore any content-type header because someone thought that

[03:42:01.0000] <hdworak>
I know about Content-Type, Content-Location, XHTML 1.1 required MIME, <?xml version="" encoding="", <meta content-type, xml:base, <base href=""

[03:43:00.0000] <hdworak>
I just do not think it does matter at all for this particular application

[03:43:01.0000] <gsnedders>
possible issue of ignoring it: what if I send Content-Type: text/html; charset=gb2312?

[03:43:02.0000] <hdworak>
I'll probably add content selection (HTML, XHTML, RSS) once I introduce RSS parsing

[03:43:03.0000] <hsivonen>
/me mumbles something about Trackback autodiscovery being so, so wrong

[03:44:00.0000] <hdworak>
gsnedders: does html5lib not return utf-8?

[03:44:01.0000] <gsnedders>
hdworak: You _really_ don't want to touch RSS parsing. Srsly.

[03:44:02.0000] <gsnedders>
hdworak: It wouldn't know what the input character encoding is.

[03:44:03.0000] <hsivonen>
I can imagine someone go "this is kinda like metadata. if you have metadata, you MUST use RDF. RDF won't validate. Let's comment it out"

[03:45:00.0000] <hdworak>
so it fails even for proper iso-8859-2 or something?

[03:45:01.0000] <gsnedders>
/me can't remember the details

[03:45:02.0000] <hdworak>
hsivonen: yes, but now they should use RDFa

[03:45:03.0000] <annevk>
Hixie, "requiring browsers to implementing rendering engines"

[03:45:04.0000] <hdworak>
does html5lib not try to guess the encoding?

[03:46:00.0000] <hsivonen>
hdworak: no, the point were the thinking goes wrong is at "if you have metadata, you MUST use RDF"

[03:46:01.0000] <hdworak>
like encutils or chardet?

[03:46:02.0000] <gsnedders>
hdworak: All it does is use meta@charset.

[03:46:03.0000] <gsnedders>
hdworak: and falls back to Windows-1252

[03:46:04.0000] <hdworak>
huh...

[03:46:05.0000] <gsnedders>
hdworak: Oh, it does also use chardet

[03:46:06.0000] <hdworak>
so that's fine for the time being

[03:47:00.0000] <hdworak>
in the next versions I'll add content-type and encoding selection (aside of autodetection)

[03:47:01.0000] <hdworak>
like W3C Validator does

[03:47:02.0000] <hsivonen>
in retrospect, both Trackback and Pingback are more complex and bogus than they'd need to be

[03:47:03.0000] <hdworak>
but really, the first release does not have to be bulletproof

[03:47:04.0000] <hdworak>
pls understand

[03:47:05.0000] <Philip`>
To anyone who cares: http://wiki.whatwg.org/wiki/User_talk:Highhi and http://wiki.whatwg.org/wiki/User_talk:Hijing are spam

[03:48:00.0000] <hsivonen>
Philip`: I'll deal with those

[03:48:01.0000] <gsnedders>
hsivonen: Trackback needs more complexity insofar as to give a character encoding, really

[03:48:02.0000] <gsnedders>
hdworak: Plenty of things have become broken forever because the first release of a reader was broken (e.g., the web)

[03:49:00.0000] <hdworak>
:)

[03:49:01.0000] <hdworak>
I think this is going a bit too far

[03:49:02.0000] <hsivonen>
gsnedders: I think the right way to implement blong pinging would be to collect the link relation from Referer and for the pinged site to dereference the Referer URI and find the relevant <a href> element and take an extract around that

[03:49:03.0000] <gsnedders>
hsivonen: Yeah, that's Hixie's own opinion :)

[03:49:04.0000] <Philip`>
gsnedders: Plenty of things have failed even more badly because the first release of a reader was written too carefully and correctly and took too long to write and didn't have enough exciting features (e.g., the web's competitors) :-p

[03:50:00.0000] <hdworak>
thank you for your time for answering my questions, I do appreciate that, great channel

[03:50:01.0000] <gsnedders>
Philip`: Like me! I'm broken!

[03:50:02.0000] <Philip`>
(Those competitors have failed so badly that I can't even think of any specific examples, so I'm just assuming some must have existed)

[03:51:00.0000] <gsnedders>
Philip`: XHTML?

[03:51:01.0000] <Philip`>
gsnedders: XHTML isn't competing with the web

[03:52:00.0000] <gsnedders>
Philip`: True. But it's competing for the web.

[03:52:01.0000] <hsivonen>
Philip`: done

[04:00:00.0000] <Philip`>
/me finds a shop receipt that says "DVDs Coming Soon: ... NimÆs Island ...", and wonders where the character encoding issue got introduced

[04:02:00.0000] <Philip`>
hsivonen: Thanks

[04:03:00.0000] <virtuelv_>
hsivonen: you can mumble about trackback autodiscovery being wrong

[04:04:00.0000] <virtuelv_>
I'll just mumble about trackbacks being so wrong

[04:04:01.0000] <virtuelv_>
I had to turn support for them off globally on my sites, due to the insane amounts of spam that got through

[04:05:00.0000] <Philip`>
Is it the whole concept of bidirectional links that is wrong?

[04:06:00.0000] <virtuelv_>
no, rather the concept of untrusted bidirectional links

[04:08:00.0000] <Philip`>
/me isn't sure about the concept of trust either

[04:08:01.0000] <hdworak>
bye

[04:12:00.0000] <virtuelv_>
Philip`: there usually is none

[04:12:01.0000] <virtuelv_>
unless the trust is between two endpoints both controlled by the same entity

[04:13:00.0000] <virtuelv_>
or all approval of trust is manual, or (cryptographically) secure

[04:13:01.0000] <virtuelv_>
trackback fails in all such respects

[04:13:02.0000] <annevk>
man, Flickr Uploadr for the Mac sucks

[04:13:03.0000] <annevk>
keeps failing after a few images

[04:25:00.0000] <hsivonen>
/me uses his own upload script

[04:26:00.0000] <hsivonen>
annevk: do you want a copy?

[04:52:00.0000] <annevk>
hsivonen, yeah, why not

[04:52:01.0000] <gDashiva>
Philip`: Some cash registers here in Norway have a weird bug where lowercase norwegian letters become symbols, but uppercase ones work fine

[04:58:00.0000] <hsivonen>
annevk: emailed

[04:59:00.0000] <hsivonen>
annevk: you may treat the script as being under the MIT license

[05:04:00.0000] <annevk>
cheers

[06:45:00.0000] <gsnedders>
I am insane.

[06:46:00.0000] <gsnedders>
Four AHs, and a H.

[06:46:01.0000] <gDashiva>
Almost Honorable and Honorable?

[06:46:02.0000] <gsnedders>
Advanced Higher and Higher

[06:46:03.0000] <Philip`>
Ah

[06:46:04.0000] <gsnedders>
http://en.wikipedia.org/wiki/Advanced_Higher and http://en.wikipedia.org/wiki/Higher_(Scottish)

[06:48:00.0000] <annevk>
/me doesn't quite get the connection between insane and doing ok (at school, anyway)

[06:48:01.0000] <gsnedders>
annevk: No, that's what I'm doing this year. It is an insane amount of work. Nobody else in my year is doing it.

[06:48:02.0000] <gsnedders>
annevk: Very, very, very few people are doing four AHs anyway.

[06:48:03.0000] <gDashiva>
Maybe he's insane because he's not being consistent and doing 5 AH?

[06:50:00.0000] <gsnedders>
No, because I'm doing too much :)

[06:50:01.0000] <gsnedders>
But also because my name is gsnedders, therefore by definition I am insane.

[06:51:00.0000] <annevk>
makes perfect sense indeed

[06:52:00.0000] <gDashiva>
gsnedders: You have to rephrase your statement in lisp to qualify as patently insane

[06:53:00.0000] <gsnedders>
gDashiva: jfhaksdsfjkhdshjkdsf

[06:55:00.0000] <Philip`>
/me wonders who collects the royalties on the insanity patent

[06:55:01.0000] <gsnedders>
Philip`: Nobody — there's too much prior art

[06:57:00.0000] <gDashiva>
Are you implying artists are insane?

[06:58:00.0000] <gsnedders>
All art is insanity.

[07:00:00.0000] <jcranmer>
bah, I'm in my first class 95% of which will be review

[07:01:00.0000] <gsnedders>
/me goes back tomorrow

[07:01:01.0000] <gsnedders>
/me ought to type up a computing project proposal

[10:49:00.0000] <takkaria>
grr, oo.org spreadsheet is so far inferior to excel I don't know how anyone can switch from one to the other

[10:54:00.0000] <hsivonen>
takkaria: VBA in Excel is the big Windows lock-in these days

[10:56:00.0000] <takkaria>
I just mean the UI and graphing tools

[10:56:01.0000] <takkaria>
the kinds of things I used to use VBA for I can normally manage on the commandline

[12:17:00.0000] <gDashiva>
webapps is nick confusion waiting to happen... AnnB, ArtB, AdamB, and anne and arve_ to boot :)

[12:42:00.0000] <Philip`>
gDashiva: You should encourage them to each prepend a random letter to their name for disambiguation

[13:04:00.0000] <jgraham>
Adobe just sent me an email with no text but with a PDF attachment... containing nothing but the adobe logo and some text. 

[15:13:00.0000] <Hixie>
wow

[15:13:01.0000] <Hixie>
just... wow

[15:13:02.0000] <Hixie>
fundamentalism has eached public-html in force.

[15:15:00.0000] <gavin_>
I find it kind of entertaining, actually

[15:15:01.0000] <gavin_>
I enjoy reading boris' arguments

[15:17:00.0000] <Hixie>
james, job, anne, and boris appear to be the voices of sanity in a sea of crazy

[15:17:01.0000] <annevk>
and hsivonen on #html-wg :)

[15:18:00.0000] <roc>
which thread are we talking about?

[15:18:01.0000] <Hixie>
it utterly baffles me that these people are being serious

[15:18:02.0000] <gavin_>
roc: "flickr and alt"

[15:22:00.0000] <Hixie>
oooh, squee! a win for the new event loop stuff. It allows a really easy fix for a problem that might have been a bitch to write up the fix for.

[15:24:00.0000] <roc>
heh

[15:27:00.0000] <roc>
once again I think the confusion is that people think HTML5 has the force of law

[15:28:00.0000] <roc>
maybe one of these people will go to court and get HTML5 non-conformance declared a violation of the ADA. Wouldn't that be exciting

[15:29:00.0000] <Hixie>
i hope i'm a witness in that court case

[15:29:01.0000] <Hixie>
so that i can help make that law pass

[15:29:02.0000] <Hixie>
i'll then carefully change html5 to require everytone to pay me $5 a day

[15:30:00.0000] <annevk>
did anyone see an e-mail about https://bugzilla.mozilla.org/show_bug.cgi?id=448674 ?

[15:31:00.0000] <annevk>
/me didn't

[15:32:00.0000] <roc>
it's completely obvious

[15:32:01.0000] <roc>
I'll send an email about it if you like

[15:33:00.0000] <annevk>
prolly good as it's not part of the spec now

[15:33:01.0000] <Philip`>
It's inconsistent with the spec's requirements for animated images (like animated GIF and APNG)

[15:33:02.0000] <Philip`>
(since the spec requires the poster frame or first frame to be rendered, I think)

[15:35:00.0000] <roc>
was there an email when Opera added SVG element support there?

[15:35:01.0000] <roc>
because I've got some questions about that :-)

[15:35:02.0000] <annevk>
yeah, I e-mailed

[15:35:03.0000] <roc>
ok cool

[15:35:04.0000] <annevk>
to public-html iirc

[15:35:05.0000] <annevk>
so might not have reached you, dunno

[15:35:06.0000] <roc>
oh

[15:35:07.0000] <roc>
dear

[15:36:00.0000] <roc>
hmm

[15:36:01.0000] <roc>
sigh

[15:39:00.0000] <roc>
thank goodness we're open source :-)

[16:02:00.0000] <Philip`>
"what's the poing?" - I have often wondered that

[16:05:00.0000] <annevk>
http://www.google.com/search?q=%22what's+the+poing?%22

[16:08:00.0000] <_Xenos>
http://www.google.com/search?hl=en&hs=SkX&defl=en&q=define:poing

[16:08:01.0000] <_Xenos>
"a closed hand" seems to be the #1 definition

[16:14:00.0000] <gavin_>
heh

[16:14:01.0000] <gavin_>
that's french


2008-08-19
[17:06:00.0000] <Hixie>
i wonder how i should handle the relative order of events fired on a <video> element when it has a pending load() about to start

[17:07:00.0000] <Hixie>
i mean, if it takes 250ms to get around to calling load(), and timeupdate has fired since that load() was queued

[17:07:01.0000] <Hixie>
should i just discard all those events?

[17:08:00.0000] <Hixie>
we can't very well delay the load() until we've dealt with all the timeupdate events, we might just end up generating more

[19:01:00.0000] <Hixie>
wow, this alt discussion just gets more and more asinine

[19:02:00.0000] <Hixie>
luckily since none of the recent posts seem to actually be about the spec, i probably don't have to actually respond to any of the recent ones

[19:02:01.0000] <Hixie>
(the ones steven sent near the start of the thread are actual useful feedback that needs replies, obviously)

[22:10:00.0000] <Hixie>
roc: the idea is that the bounding box height is used to set the height, at which point it can't be scrolled, since all the content is visible

[22:12:00.0000] <roc>
even if, say, the iframe element has a CSS max-height?

[22:13:00.0000] <roc>
I have to admit I'm a bit confused about how to compute this bounding box

[22:17:00.0000] <roc>
it seems we're supposed to compute the bounding box of everything that's rendered, ignoring viewport clipping ... but what about completely transparent elements? elements with width 0 but nonzero height?

[22:17:01.0000] <roc>
Should we just use the area that we would permit scrolling to, assuming the viewport had scrollbars?

[22:18:00.0000] <roc>
that might work, although since we don't allow scrolling up to see content above the top of the viewport, it doesn't quite jive with your answer about having to include content that overflows vertically above in the bounding box

[22:19:00.0000] <Hixie>
i guess it is pretty complicated :-)

[22:19:01.0000] <roc>
using the height of the root element seems a lot simpler

[22:19:02.0000] <Hixie>
originally i went this way because it was the route you suggested in the mozilla bug on the topic :-)

[22:19:03.0000] <roc>
I guess I changed my mind :-)

[22:19:04.0000] <Hixie>
let's use the root element height then

[22:19:05.0000] <roc>
ok cool

[22:20:00.0000] <Hixie>
like i said in my last e-mail, i'm basically just waiting for your experience

[22:20:01.0000] <Hixie>
i don't know what the spec should say really

[22:20:02.0000] <Hixie>
i don't want it to be as complicated as sicking's "let's merge the DOMs" idea though

[22:20:03.0000] <Hixie>
that scares me

[22:20:04.0000] <roc>
me too

[22:20:05.0000] <roc>
 we're not going to be doing anything except the intrinsic width/height stuff initially

[22:21:00.0000] <Hixie>
cool

[22:21:01.0000] <roc>
I don't think that level of incomplete implementation will break the feature going forward, but if that's a concern, we can make it mozSeamless or something

[22:21:02.0000] <Hixie>
what are you not doing?

[22:22:00.0000] <roc>
the stylesheet integration

[22:22:01.0000] <Hixie>
ah, right

[22:22:02.0000] <Hixie>
hmm

[22:22:03.0000] <Hixie>
yeah probably best to call it moz-seamless

[22:22:04.0000] <Hixie>
that's a pretty big difference

[22:23:00.0000] <roc>
ok

[00:58:00.0000] <jmb>
~

[01:21:00.0000] <BenMillard>
for the IRC records, if you haven't seen it on Public-HTML the current progress of my web page research for this year is here: http://projectcerbera.com/web/study/2008/collection

[01:34:00.0000] <webben_>
/me finds the use by http://www.enableholidays.com/ (from BenMillard's list) outline-style: none; mildly depressing

[01:35:00.0000] <BenMillard>
that disables focus outlines, right?

[01:36:00.0000] <BenMillard>
I find it rather unfortunate, too

[01:37:00.0000] <webben_>
BenMillard: Right.

[01:37:01.0000] <Hixie>
http://www.hereticpress.com/Dogstar/Publishing/WebSurvey.html gives me an error message

[01:39:00.0000] <BenMillard>
works for me

[01:40:00.0000] <Hixie>
it says:

[01:40:01.0000] <Hixie>
Your permission to access Heretic Press has been denied.

[01:40:02.0000] <Hixie>
Your browser may be unidentified, you may be downloading too many files for offline viewing or are breaching copyright on digital images. Your IP address might be trying to access password protected files? 

[01:40:03.0000] <Hixie>
You may have been unfairly excluded from access. Sorry.

[01:40:04.0000] <BenMillard>
I'm using Firefox 2.0.0.16 on Windows XP

[01:42:00.0000] <webben_>
Hixie: wfm in Lynx and OmniWeb

[01:42:01.0000] <Hixie>
oh well

[01:43:00.0000] <gDashiva>
At least they acknowledge false positives may happen :)

[01:44:00.0000] <BenMillard>
Hixie, do you have scripting disabled?

[01:44:01.0000] <BenMillard>
looking at the source code, the error you gave is placed in a <noscript> element near the start

[01:45:00.0000] <BenMillard>
the pages are built in a very weird way...maybe I should study them :P

[01:46:00.0000] <BenMillard>
it has 23 <link rel> elements, many seem to be using made-up rel values

[01:47:00.0000] <BenMillard>
23 <link rel> elements pointing to related documents, I mean

[03:29:00.0000] <Hixie>
ok, the spec now defines an event loop mechanism

[03:29:01.0000] <Hixie>
and that is used throughout whenever asynchronous behaviour is involved

[03:33:00.0000] <Hixie>
i wonder how the es-future work is going to do opt-in with javascript: urls

[03:38:00.0000] <hsivonen>
Hixie: do I guess correctly that you are not a fan of opt-in?

[03:38:01.0000] <Hixie>
yes.

[03:38:02.0000] <Hixie>
well no

[03:39:00.0000] <Hixie>
i'm specifically not a fan of out-of-band opt-in

[03:39:01.0000] <Hixie>
that is, i have professional reservations about out-of-band opt-in

[03:39:02.0000] <Hixie>
i also have personal reservations about any kind of opt-in, but that's just a personal preference

[03:40:00.0000] <Hixie>
out-of-band opt-in as far as i am concerned is a complete non-starter. for example, it means that you can never update an existing widely deployed script to use 

[03:40:01.0000] <Hixie>
the new features

[03:41:00.0000] <Hixie>
e.g. google couldn't on-the-fly use new features in the google analytics scripts

[03:45:00.0000] <Philip`>
Gecko supports most of the new JS features without any opt-in; you only need opt-in for new keywords

[03:46:00.0000] <Hixie>
yeah

[03:46:01.0000] <Hixie>
so silly

[03:46:02.0000] <Philip`>
(so it should only change the behaviour of scripts that didn't compile under previous versions of the language)

[03:47:00.0000] <Hixie>
anyway, bed time

[03:48:00.0000] <virtuelv_>
Hixie: any particular reasoning behind the odd camelcasing of navigator.onLine?

[03:53:00.0000] <annevk>
virtuelv_, ask Microsoft

[03:54:00.0000] <virtuelv_>
heh

[03:54:01.0000] <virtuelv_>
that was an answer at least

[04:18:00.0000] <hsivonen>
/me doesn't like it when in a public bug tracker fanboys try to explain how the product doesn't suck and the reporter should jump through hoops instead

[04:18:01.0000] <hsivonen>
(remark not related to any HTML bug tracker bug)

[04:27:00.0000] <gDashiva>
Don't you know they're doing you a service by letting you help improve the product? ;)

[04:28:00.0000] <gDashiva>
/me groans at Dr. Hoffman getting started on poetry markup again

[06:11:00.0000] <jcranmer>
"The W in WHATWG stands for..." shouldn't that be "the first W" ?

[07:16:00.0000] <jgraham>
/me wonders if it's worth dragging up the fact that WCAG is only chartered to address the needs of users with disabilities whereas HTML has to work for a wider range of people

[07:18:00.0000] <hsivonen>
jgraham: yes, if you think it's worthwhile to touch the subject matter in general

[07:19:00.0000] <jgraham>
hsivonen: That's what I can't decide. I don't think I will be very popular...

[07:19:01.0000] <jgraham>
(but perhaps it needs to be said)

[07:21:00.0000] <hsivonen>
PT(W) completely missed the point of my question

[07:22:00.0000] <Philip`>
Then people might say that alt is primarily an accessibility feature so it needs to address the needs of users with disabilities and therefore WCAG is the right place to do that

[07:22:01.0000] <gDashiva>
I am amazed, over and over, at how little faith they have in WCAG

[07:23:00.0000] <hsivonen>
gDashiva: have you read WCAG 2.0? If you have, do *you* have faith in it?

[07:23:01.0000] <gDashiva>
hsivonen: No, but then I don't consider it a good idea either

[07:24:00.0000] <gDashiva>
(My first introduction to it was Joe Clark's article, and I haven't seen anything to dispute it)

[07:29:00.0000] <jgraham>
Philip`: It's not just an accessibility feature though. It's also needed or e.g. text-only browsers

[07:29:01.0000] <jgraham>
s/or/for/

[07:30:00.0000] <gDashiva>
Are balloons 'moored' when landed or is there a different verb?

[07:30:01.0000] <hsivonen>
gDashiva: it's not quite as bad as it used to be

[07:30:02.0000] <hsivonen>
gDashiva: it's still bad, though

[07:30:03.0000] <Lachy>
othermaciej, yt?

[07:31:00.0000] <gDashiva>
hsivonen: Bad enough that you might be better off with the samurai errata instead?

[07:32:00.0000] <Philip`>
gDashiva: http://www.balloonlife.com/publications/balloon_life/9608/images/tetherm.gif shows a "MOORED" balloon, so that sounds reasonable

[07:32:01.0000] <Philip`>
(assuming it is actually moored, not just sitting on the ground because it's too heavy to float)

[07:32:02.0000] <hsivonen>
gDashiva: bad enough that the WG chose to produce exegesis documents themselves, because the spec itself is not suitable for mortals to interpret on their own

[07:33:00.0000] <gDashiva>
Philip`: Good enough, thanks

[07:33:01.0000] <gDashiva>
I'm anticipating a new category of spec writers, the ones sitting in the balloon moored on top of the ivory tower

[07:37:00.0000] <Philip`>
jgraham: Indeed, but it sounds like that'd possibly just shift the arguments to go in a different direction, rather than actually resolving anything

[07:38:00.0000] <Lachy>
http://arstechnica.com/news.ars/post/20080819-mozilla-drags-ie-into-the-future-with-canvas-element-plugin.html

[07:38:01.0000] <Philip`>
How would you make a whole tower out of ivory, unless you poached one of the Discworld elephants?

[07:38:02.0000] <gDashiva>
Philip`: That's what we've been doing all along

[07:39:00.0000] <gDashiva>
One person agrees to one resolution, and then another one picks up a different end and starts it all over again

[07:39:01.0000] <jgraham>
Philip`: Well the point that I may or not make is that you end up with different conformance requirements if your only goal is to support the disabled compared to those you end up with if your goal is to support everyone

[07:39:02.0000] <gDashiva>
I'm sure every single person on the list has said "Yeah, that would work" to at least one suggestion

[07:39:03.0000] <gDashiva>
But they never do it to the same one

[07:40:00.0000] <jgraham>
e.g. WCAG seems to forbit publishing James Joyce's Uylsses in an unaltered form

[07:40:01.0000] <jgraham>
s/forbit/forbid/

[07:42:00.0000] <Philip`>
Lachy: Why do people think Adobe releasing an ActionScript VM as open source has any relation to making it not totally unrealistic that Adobe would distribute Mozilla's canvas plugin with Flash? That part of the article seems totally irrelevant to me...

[07:43:00.0000] <jgraham>
also the WCAG people still seem to have the attitude that people should be prepared to put an unlimited amount of effort into accessibility despite the evidence that this is not true. Designing around this assumption doesn't help accessibility but does have negative consequences on people's perception of the value of conformance requirements

[07:43:01.0000] <Lachy>
Philip`, I have no idea

[07:44:00.0000] <Philip`>
(and the rest of the article is just rephrasing the three-week-old blog post)

[07:44:01.0000] <Lachy>
I didn't see the 3 week old blog post before

[07:44:02.0000] <Philip`>
You need to subscribe to more people's blog feeds :-)

[07:45:00.0000] <Lachy>
I'm probably on too many already. It's difficult to keep up with everything

[07:55:00.0000] <hsivonen>
/me wonders if the whole content area of IE could be taken over by Gecko-as-ActiveX

[07:59:00.0000] <hsivonen>
/me sees http://starkravingfinkle.org/blog/2006/12/xule-what-if/

[08:04:00.0000] <Philip`>
Combining the web-compatibility of Gecko with the UI of IE? That sounds like a fantastic idea

[08:04:01.0000] <gDashiva>
How long does an internet argument have to last before you can stop assuming ignorance and start assuming malice?

[08:07:00.0000] <jgraham>
gDashiva: I thought most people assumed that from the start ;)

[08:12:00.0000] <Lachy>
is there anyone here who understands and can explain how DOMImplementation.getFeature() and Node.getFeature() in DOM3Core are supposed to work?

[08:14:00.0000] <takkaria>
gDashiva: given the length of the alt thread, I'd say at least a few years

[08:20:00.0000] <Lachy>
JohnResig, yt?

[08:21:00.0000] <JohnResig>
Lachy: what's up

[08:21:01.0000] <Lachy>
JohnResig, the selectors api test suite needs tests for hasFeature() and getFeature()

[08:22:00.0000] <Lachy>
I'm not exactly sure what getFeature() is supposed to do, besides return a copy of the object itself if the feature is implemented

[08:22:01.0000] <Lachy>
and if that's all it's meant to do, then it seems quite useless

[08:23:00.0000] <Lachy>
and Node.isSupported() too

[08:24:00.0000] <Lachy>
and probably getDOMImplementation() too

[08:24:01.0000] <Lachy>
and getDOMImplementationList()

[08:24:02.0000] <Lachy>
/me wishes he didn't add a feature string. I didn't realise it would complicate so many things :-(

[08:25:00.0000] <JohnResig>
Lachy: could you come up with a list of what exactly is supposed to be tested?

[08:28:00.0000] <Lachy>
JohnResig, I'm mailing public-webapps about it now

[08:33:00.0000] <Lachy>
JohnResig, http://lists.w3.org/Archives/Public/public-webapps/2008JulSep/0436.html

[08:35:00.0000] <JohnResig>
Lachy: what feature/version strings should I be checking for?

[08:37:00.0000] <Lachy>
see the spec

[08:38:00.0000] <Lachy>
"Selectors-API", "1.0"

[09:24:00.0000] <takkaria>
argh, someone talking about it being a moral duty to the web to produce valid markup

[09:26:00.0000] <Philip`>
We need to define some HyperText Morality Laws to enforce that

[09:28:00.0000] <takkaria>
I'm not sure there's enough demand for that spec :)

[09:29:00.0000] <Philip`>
Lack of demand doesn't mean we don't have a moral obligation to write it

[09:30:00.0000] <takkaria>
true enough

[09:31:00.0000] <hsivonen>
a moral duty to use an html5 validator could be good for me

[09:31:01.0000] <hsivonen>
if others had the duty that is

[09:31:02.0000] <MikeSmith>
hsivonen: fwiw, I notice that the v.nu schemas define ol.elem.phrase, ul.elem.phrase, and dl.elem.phrase patterns -- but those are never actually referenced by any defines

[09:32:00.0000] <takkaria>
hsivonen: it could be good for you too if you used AdSense :)

[09:33:00.0000] <hsivonen>
MikeSmith: I left those around expecting Hixie to allow a <span> hack for doing phrase-level lists

[09:36:00.0000] <MikeSmith>
hsivonen: ah, I see

[10:21:00.0000] <hober>
I wonder why Sunava always says "HTML 5.0". Where did the ".0" come from?

[10:23:00.0000] <gDashiva>
Maybe it's a microsoft thing?

[10:24:00.0000] <gDashiva>
If you're always doing minor releases, it might become a habit to attach .0

[10:24:01.0000] <Lachy>
hober, 5.0 == 5.

[10:25:00.0000] <Lachy>
it's probably just an assumption based on the fact that it was  HTML 2.0, 3.0, 3.2, 4.0 and 4.01 before.

[10:29:00.0000] <gDashiva>
/me wonders about the logic going on

[10:30:00.0000] <gDashiva>
Maybe something like "If it's easy, it's too easy. Too easy is bad. Accessibility isn't bad. Therefore easy means it's not accessible!"

[10:47:00.0000] <othermaciej>
Lachy: yes?

[11:00:00.0000] <Lachy>
othermaciej, someone at Opera was wondering what the licencing is for WebKit's testcases, like these. http://trac.webkit.org/browser/trunk/LayoutTests/fast/css/ Do you know?

[11:01:00.0000] <othermaciej>
I guess it's not always clearly stated in the test case

[11:01:01.0000] <othermaciej>
I think the intent is for the test suite to be BSD licensed, other than test cases from external sources

[11:02:00.0000] <Lachy>
ok, that makes sense, since the rest of webkit is BSD or LGPL

[12:05:00.0000] <anne_mibbit>
jgraham, where shall we meet?

[12:05:01.0000] <jgraham>
anne_mibbit: Dunno. 

[12:05:02.0000] <anne_mibbit>
/me is in cambridge now, at "Cats"

[12:05:03.0000] <jgraham>
Did we decide when we were meeting?

[12:06:00.0000] <anne_mibbit>
I thought today :)

[12:06:01.0000] <jgraham>
Oh, I didn't realise that we had decided that. Does Philip` know?

[12:06:02.0000] <jgraham>
Philip`: ^^ that was aimed at you

[12:07:00.0000] <jgraham>
I can wonder into town, I guess

[12:07:01.0000] <anne_mibbit>
I'm not really sure where the rest of the CSS gang is and I need some food

[12:08:00.0000] <anne_mibbit>
quite a long trip to get here

[12:08:01.0000] <jgraham>
OK I will come in

[12:09:00.0000] <anne_mibbit>
wait, where are you going to?

[12:09:01.0000] <anne_mibbit>
hmm

[12:10:00.0000] <jgraham>
I could meet you outside cats I guess. Oroutside Kings which is a bit closer to where I am going and I guess better for Phillip

[12:10:01.0000] <jgraham>
Only one l ...

[12:10:02.0000] <anne_mibbit>
lets see

[12:10:03.0000] <jgraham>
It will take me about 30 minutes to get there

[12:11:00.0000] <anne_mibbit>
where is kings?

[12:12:00.0000] <anne_mibbit>
oh, Kings College that is a bit further up the road?

[12:12:01.0000] <jgraham>
http://maps.google.co.uk/maps?f=q&hl=en&geocode=&q=kings+cambridge&ie=UTF8&ll=52.207764,0.132179&spn=0.022617,0.06815&z=15&iwloc=A

[12:13:00.0000] <anne_mibbit>
where are looking at the same building :)

[12:13:01.0000] <anne_mibbit>
s/where/we/

[12:13:02.0000] <anne_mibbit>
ok, I'll go there in 10 minutes 

[12:14:00.0000] <jgraham>
OK, I'll meet you outside the front gate, by the odd-shaped post box

[12:14:01.0000] <jgraham>
/me leaves

[12:15:00.0000] <anne_mibbit>
Philip`, ^^

[12:22:00.0000] <gsnedders>
anne_mibbit: I'd say it'd take maybe two or three minutes to walk there from Cat's, FYI

[12:24:00.0000] <anne_mibbit>
anyway, got to go

[13:08:00.0000] <zcorpan>
Hixie: #seekUpdate says The user agent must queue a task to queue a task to ...

[14:05:00.0000] <Philip`>
anne, jgraham: Oh, I thought people had said Thursday or Friday or something, not today

[14:06:00.0000] <Philip`>
and I can't make it now

[14:08:00.0000] <Philip`>
(jgraham: By the way, I've moved house since we last met and now I live on pretty much exactly the opposite side of King's, though still about the same distance away so it wouldn't actually make much difference)C

[14:09:00.0000] <Philip`>
(Um, that's a peculiar sequence of characters on the end there)

[14:09:01.0000] <Philip`>
/me blames his intermittent wireless disconnections

[15:32:00.0000] <Hixie>
/me replies to http://lists.w3.org/Archives/Public/wai-xtech/2008Aug/0114.html

[15:33:00.0000] <Hixie>
i wish i understood more of what Al said

[15:33:01.0000] <Hixie>
http://lists.w3.org/Archives/Public/wai-xtech/2008Aug/0118.html is my reply

[15:47:00.0000] <webben>
Hixie: Isn't Al's point about not bothering referring to mass photo uploaders who don't bother to provide text equivalents? His stuff about the two agents is simply that it's only by considering the mass upload photo site web developers are agents in isolation that one can shunt this not bothering outside the remit of the spec.

[15:48:00.0000] <webben>
Hixie: Seems to me people regular pop up to say they can't be bothered to provide text equivalents for all their photos on Flickr.

[15:48:01.0000] <webben>
(in the various threads about alt)

[15:49:00.0000] <Hixie>
but the author in that case is flickr, not the uploader

[15:50:00.0000] <Hixie>
the uploader is just a user

[15:50:01.0000] <webben>
Hixie: I think that's a distinction Al's questioning.

[15:51:00.0000] <Hixie>
i doubt it, that would be crazy

[15:51:01.0000] <webben>
how would it be crazy

[15:51:02.0000] <webben>
let's say you have an interface that provides a box in which i enter text

[15:51:03.0000] <webben>
and then that text is dumped out in some HTML

[15:51:04.0000] <webben>
are you really saying I'm not the author of the text?

[15:52:00.0000] <webben>
conversely, if the box is left blank, am I not the author of the blankness

[15:52:01.0000] <Hixie>
you're not the author of the web page, certainly. you don't even know what html is, in all likelihood.

[15:52:02.0000] <webben>
I think that's like arguing a publisher is an author of the book.

[15:52:03.0000] <Hixie>
nobody is seriously going to think that flickr users are going to give two hoots about what html5 says about their usage of flickr

[15:52:04.0000] <webben>
Hixie: yes. but that's not the point contended.

[15:53:00.0000] <Hixie>
no, it's like arguing that the person who sends a letter to an editor of a newspaper is responsible for the typography in the newspaper

[15:53:01.0000] <webben>
Hixie: I don't think the absence of a text equivalent is equivalent to the typography of the newspaper.

[15:53:02.0000] <Hixie>
or that a book author is responsible for the placement of the book on the bookshelf in the book shops

[15:53:03.0000] <webben>
I think that's more like leaving your name off the letter.

[15:54:00.0000] <webben>
what the software does with the absence of the text equivalent is perhaps like the typography; but not "authorship".

[15:55:00.0000] <Hixie>
*shrug*

[15:55:01.0000] <Hixie>
it's a moot argument

[15:55:02.0000] <webben>
it might help if "author" wasn't used in isolation but in a phrase like "author of"

[15:55:03.0000] <Hixie>
flickr users aren't going to provide useful text replacements

[15:55:04.0000] <Hixie>
there's really no point discussing whose fault that is

[15:55:05.0000] <webben>
e.g. "author of code to handle some user input" vs. "author of an image and its text equivalent, or lack thereof"

[15:55:06.0000] <Hixie>
the html5 spec talks about the conformance of the web page

[15:55:07.0000] <webben>
Hixie: Al isn't discussing whose fault it is.

[15:56:00.0000] <Hixie>
(regardless of who wrote it)

[15:56:01.0000] <Hixie>
so what is he discussing?

[15:57:00.0000] <webben>
He's saying there's some not bothering going on, and that's clearly true: indeed it's precisely what you're saying.

[15:57:01.0000] <Hixie>
ah

[15:57:02.0000] <Hixie>
well that seems like a moot point too

[15:58:00.0000] <webben>
he's saying that "According to HTML5, if the human didn't bother, the page isn't compliant" isn't true, because it's meaningless to exclude the most important human - the one who actually authors the meaning and would normally provide the text equivalent.

[15:58:01.0000] <webben>
whether it's important whether that's true or not doesn't make it true or not.

[15:58:02.0000] <Hixie>
hm

[16:00:00.0000] <Hixie>
given that flickr can't ever get suitable replacement text for most of these images, and given that we want to cater for flickr since that kind of site is a big part of the web, it's not clear to me how to address that feedback in the spec.

[16:01:00.0000] <webben>
well, Flickr is a largely moot case, because so far neither conformance not text equivalents have proved of any importance to flickr

[16:02:00.0000] <Hixie>
i use flickr as a symbol for any media sharing site

[16:02:01.0000] <webben>
one kind of has to imagine an idealized flickr that cares about conformance (why?) and text equivalents (SEO? accessibility? interoperability?) for it to be even useful as a way to think about this problem

[16:03:00.0000] <Hixie>
we want it to be technically possible to write a media sharing site using conforming html5

[16:03:01.0000] <Hixie>
even if the users don't provider alternative text (since they won't)

[16:03:02.0000] <Hixie>
provide, even

[16:04:00.0000] <webben>
/me isn't really clear on why.

[16:04:01.0000] <Hixie>
why what?

[16:04:02.0000] <gavin_>
why you want that

[16:04:03.0000] <webben>
why we want it

[16:04:04.0000] <Hixie>
media sharing sites are amongst the most important (by usage) sites on the web

[16:04:05.0000] <Hixie>
it would be a pretty big failure if we didn't address them

[16:04:06.0000] <webben>
would it? why?

[16:05:00.0000] <webben>
why isn't enough to define how UAs should interoperate with them?

[16:05:01.0000] <gavin_>
what practical effects would "not addressing them" have?

[16:05:02.0000] <Hixie>
because we're trying to write a language for the web, and they're a big part of the web?

[16:05:03.0000] <Hixie>
i don't understand the question

[16:06:00.0000] <webben>
yes, but the reality is not every document will "conform" (indeed, if history is any guide at all, most won't) and that UAs need to cope with that.

[16:06:01.0000] <Hixie>
sure, but so what?

[16:06:02.0000] <Hixie>
that's already dealt with

[16:06:03.0000] <Hixie>
that's the easy part of this problem, frankly

[16:06:04.0000] <webben>
indeed.

[16:07:00.0000] <gavin_>
you want flickr to be conformant, and accessibility people want conformance to mean accessibility, and flickr will not be accessible

[16:07:01.0000] <gavin_>
right?

[16:07:02.0000] <webben>
I guess it depends on what the point of "conformance" is .

[16:07:03.0000] <Hixie>
i don't know what the accessibility people want, but if they want conformance to mean accessibility, they're doomed.

[16:07:04.0000] <Philip`>
Flickr's developers don't care about conformance, so it's not going to have any practical effect on them at all

[16:07:05.0000] <webben>
I mean, I completely agree one wants authors to write markup with defined effects.

[16:08:00.0000] <webben>
but afaict the only point of conformance is saying: here's a good way to do X.

[16:08:01.0000] <gavin_>
assuming what I said is true, the easy way to resolve the conflict is to stop wanting flickr to be conformant

[16:08:02.0000] <gavin_>
I don't see offhand what the problem with that would be

[16:08:03.0000] <gavin_>
(in practical terms)

[16:08:04.0000] <webben>
and there may be things where one can write markup with defined effects but there's fundamentally no good way to do it.

[16:09:00.0000] <Hixie>
conformance has many points, in particular promoting best practices and trying to keep people away from things that will make it harder to extend the language later

[16:09:01.0000] <webben>
maybe it's worth _thinking_ about seperating the two.

[16:10:00.0000] <Hixie>
however, to be able to convince people to care about conformance, we have to make it possible to be conformant

[16:10:01.0000] <Philip`>
What's the point in trying to keep people away from those things, when most people are going to completely ignore that attempt and will do those things regardless?

[16:10:02.0000] <Hixie>
it's not black and white, it's a matter of degrees

[16:10:03.0000] <webben>
Hixie: The funny thing about that statement, from what I've seen, is that the inability to extend is one of the /drivers/ of non-conformance.

[16:11:00.0000] <webben>
e.g. people wanting to make up their own elements and attributes

[16:11:01.0000] <Hixie>
conformance is like an attractor, by having it as a goal we get people closer to the goal, if not at the goal

[16:11:02.0000] <webben>
you've gone some way to head that off with data-*

[16:11:03.0000] <Hixie>
whereas without it, we'd have people distributed across all of phase space

[16:11:04.0000] <Hixie>
webben: i don't think that extending html is a driver of non-conformance

[16:12:00.0000] <Hixie>
webben: most non-conformance isn't to do things that html can't do

[16:12:01.0000] <Hixie>
webben: it's just mistakes and not understanding html

[16:12:02.0000] <webben>
Hixie: yeah that's accidental non-conformance. I'm talking about deliberate non-conformance. And not making a claim about "most".

[16:13:00.0000] <Hixie>
"most" is what matters

[16:13:01.0000] <Hixie>
it's all a matter of degrees

[16:13:02.0000] <Hixie>
anyway

[16:14:00.0000] <Hixie>
not much i can do to satisfy what al wants if what he wants is what you say he wants

[16:14:01.0000] <webben>
not with a single profile of conformance if you want publishing stuff without text equivalents to be promoted as a good thing to do.

[16:15:00.0000] <Hixie>
the spec says it's a bad thing to do

[16:15:01.0000] <webben>
what's the point of making a bad thing to do conformant again?

[16:15:02.0000] <webben>
to encourage people do other things right?

[16:16:00.0000] <webben>
*to do

[16:16:01.0000] <Hixie>
the point is to get people to do the best thing given their situation

[16:17:00.0000] <webben>
where "people" is basically software developers?

[16:17:01.0000] <Hixie>
anyone writing HTML documents

[16:18:00.0000] <webben>
yeah, the notion "writing" there seems rather problematic

[16:18:01.0000] <Hixie>
only if one is looking for it to be problematic, imho

[16:18:02.0000] <webben>
I'm a web developer on a movies website, but I don't "write" the documents

[16:19:00.0000] <webben>
I guess the point is clearly conformance here is not aimed at getting people to write text equivalents for all their photos: which would be the "best" thing to do.

[16:20:00.0000] <webben>
the conformance is rather aimed at getting software developers to handle the lack of text equivalents sanely.

[16:20:01.0000] <webben>
that goes back to the split Al identifies between two agents.

[16:20:02.0000] <Hixie>
the "best" thing to do would be for the blind people to be able to see.

[16:22:00.0000] <Hixie>
the spec is just giving advice to the people outputting the HTML page in what the best they might be able to do given their constraints is

[16:22:01.0000] <webben>
writing text equivalents is possible with current technology; allowing the blind to see doesn't seem to be (yet) and wouldn't help some of the other use-cases of text equivalents

[16:22:02.0000] <Hixie>
even suggesting that the users uploading their images are html authors in this context is laughable, imho

[16:23:00.0000] <Hixie>
whether writing text equivalents is possible or not is moot if the text equivalents aren't going to be forthcoming

[16:23:01.0000] <Hixie>
it is just as impossible to convince flickr users to provide suitable replacement text as it is to make the replacement text unnecessary

[16:23:02.0000] <webben>
doesn't really see where the laughing matter is actually. I think you're reversing the normal sense of authorship in publishing.

[16:23:03.0000] <Hixie>
(if anything, the latter is likely easier on the long term)

[16:24:00.0000] <webben>
that's not necessarily a big problem, it just makes the whole thing more obfuscated

[16:24:01.0000] <Hixie>
i'm not sure what the antecedent for "that" was

[16:24:02.0000] <webben>
that => "reversing"

[16:25:00.0000] <Hixie>
ah

[16:25:01.0000] <Hixie>
the user has no control over 

[16:25:02.0000] <Hixie>
what flickr does

[16:25:03.0000] <Hixie>
flickr has no control over what the user does

[16:25:04.0000] <Hixie>
the html page is under the control of flickr

[16:26:00.0000] <Hixie>
so flickr is the author

[16:26:01.0000] <webben>
I don't think the "control" thing has much to do with authorship.

[16:26:02.0000] <Hixie>
seems pretty cut and dry to me

[16:26:03.0000] <webben>
I don't control what Penguin does. Penguin doesn't control what I do.

[16:26:04.0000] <svl>
"I wear the cheese. The cheese does not wear me."

[16:27:00.0000] <Hixie>
i just saw that episode

[16:27:01.0000] <svl>
:)

[16:27:02.0000] <Hixie>
webben: control over the html page has everything to do with whose responsibility the conformance of the page is

[16:28:00.0000] <webben>
Hixie: possibly yes. I think "author" is a poor choice of word for the person "whose responsibility the conformance of the page is"

[16:28:01.0000] <webben>
(yes, in rare circumstances, they are the same person)

[16:29:00.0000] <Hixie>
the spec doesn't use the term "author"

[16:29:01.0000] <Hixie>
(for this requirement)

[16:29:02.0000] <webben>
I don't think photo sharing sites can consistently say the lack of text equivalents is both the user's fault and their responsibility however

[16:29:03.0000] <Hixie>
it doesn't matter whose blame it is

[16:30:00.0000] <Hixie>
it is their responsibility because they generate the html

[16:30:01.0000] <Hixie>
maybe a better example would be a webcam

[16:30:02.0000] <Hixie>
or a blind photographer's photo site

[16:30:03.0000] <webben>
I doubt it.

[16:30:04.0000] <webben>
webcam's don't have authors

[16:30:05.0000] <Hixie>
that's the point

[16:30:06.0000] <Hixie>
there is html

[16:30:07.0000] <webben>
flickr does

[16:31:00.0000] <Hixie>
ok let's talk about the webcame case then

[16:31:01.0000] <Hixie>
there is only one human

[16:31:02.0000] <Hixie>
the one who generates the HTML with the <img>

[16:31:03.0000] <Hixie>
he has no idea what the image is of

[16:31:04.0000] <Hixie>
does that make the issue clearer?

[16:31:05.0000] <webben>
is this an important use-case though?

[16:32:00.0000] <Dashiva>
webben: Tried using the internet lately? There are a lot of webcams

[16:32:01.0000] <webben>
Dashiva: aren't they broadcasting in Flash players?

[16:32:02.0000] <Hixie>
it is a use case that i want to make sure we cover.

[16:33:00.0000] <Dashiva>
You'd be surprised how clever people are

[16:33:01.0000] <Hixie>
because i see no reason to not cover it.

[16:33:02.0000] <jruderman>
how about making HTML validators warn when they encounter alt="{User-uploaded image}", saying "it would be great if you could supply real alt text there"?

[16:33:03.0000] <Dashiva>
multipart jpeg, motion jpg, refreshing with script... 

[16:33:04.0000] <Hixie>
jruderman: doesn't the validator.nu validator already do that?

[16:33:05.0000] <jruderman>
ahh :)

[16:33:06.0000] <webben>
jruderman: Doesn't Flickr already know that... ?

[16:34:00.0000] <webben>
when I deal with photos coming in from a feed with failsome or missing text equivalents and need to pump them into a gallery, I certainly know the alts could be better!

[16:34:01.0000] <Hixie>
jruderman: hsivonen certainly has a great image report feature, dunno whether it's been updated to the new text.

[16:34:02.0000] <Hixie>
jruderman: but also... his image report feature itself is a great example of a page where there is an image and no alternative text

[16:34:03.0000] <Philip`>
jruderman: Then people will write alt="User-uploaded image" to get rid of the ugly warning

[16:34:04.0000] <Hixie>
jruderman: and i'm pretty sure we want to make sure the html5 conformance checker can be conforming

[16:34:05.0000] <jruderman>
Philip`: hmm, good point

[16:35:00.0000] <Dashiva>
Hixie: It's an edge case ;)

[16:35:01.0000] <Hixie>
pretty damn important one

[16:35:02.0000] <jruderman>
lol

[16:35:03.0000] <Dashiva>
He can just run it on ftp, can't he?

[16:36:00.0000] <Hixie>
it'd still be html :-)

[16:36:01.0000] <Philip`>
He could remove the HTML output, and just have XML/JSON/text/etc

[16:37:00.0000] <Hixie>
oh yeah, that'd be awesome. html5 is so great, even its conformance checker can't use it.

[16:37:01.0000] <webben>
Hixie: Sure. But I think what's "best" for a validator may be different than what's "best" for a photo uploader.

[16:37:02.0000] <Hixie>
webben: then let's cover both in the spec. what is the best for both?

[16:37:03.0000] <Hixie>
webben: as in, what should the spec recommend in each case?

[16:37:04.0000] <Philip`>
I hope the validator's text output will use AAlib for the image report feature

[16:37:05.0000] <Hixie>
Philip`: doesn't help a blind user. :-)

[16:38:00.0000] <Philip`>
Does anything like WCAG require that text/plain documents be accessible?

[16:39:00.0000] <webben>
Hixie: I think you've already covered that what's best for the person uploading images to do is provide text equivalents.

[16:39:01.0000] <Dashiva>
Imagine if all the energy being used on alt was used to make people consider WCAG conformance equally (not so) important as HTML conformance

[16:39:02.0000] <webben>
and what's best for the person designing a system to house content images is to require text equivalents.

[16:39:03.0000] <webben>
that doesn't mean those systems will require text equivalents

[16:39:04.0000] <jruderman>
Philip`: ASCII art ftw

[16:39:05.0000] <webben>
(and relaxing that recommendation wouldn't _mean_ that those systems would in fact conform)

[16:40:00.0000] <Dashiva>
Philip`: You have to use as simple language as possible

[16:40:01.0000] <Hixie>
webben: but what should happen when they don't have text equivalents?

[16:40:02.0000] <Hixie>
Philip`: common sense does, but maybe that's not "like WCAG"

[16:40:03.0000] <Hixie>
/me ducks

[16:40:04.0000] <webben>
Hixie: but they get into that position by choice

[16:40:05.0000] <webben>
Hixie: i.e. they create a system that doesn't gather the data.

[16:41:00.0000] <webben>
Hixie: whereas a validator cannot gather the data since there is none

[16:41:01.0000] <Hixie>
webben: hsivonen's validator doesn't, nor does a webcam page, nor does a blind photographer, nor does a fractal generator, nor does a ...

[16:41:02.0000] <Hixie>
webben: so what should those cases do?

[16:41:03.0000] <Hixie>
webben: and what should happen if the user doesn't provide alternative text?

[16:42:00.0000] <webben>
Hixie: You can't get into a situation where the user doesn't provide alternative text without choosing that to allow that situation; in which case you've already chosen not to produce the best possible consumer experience.

[16:42:01.0000] <webben>
it might well be a reasonable business tradeoff

[16:43:00.0000] <Dashiva>
webben: What if the user chooses to provide nothing?

[16:43:01.0000] <webben>
Dashiva: Again, it's the business that has chosen to allow the user to provide nothing and still publish.

[16:43:02.0000] <Dashiva>
No, nothing as in nothing useful

[16:44:00.0000] <Dashiva>
A space, asdf, something like that

[16:44:01.0000] <Hixie>
webben: but assuming one has made the "reasonable business tradeoff" of not rejecting 99% of users, what should one do?

[16:44:02.0000] <webben>
Dashiva: Is "nothing useful" machine detectable?

[16:44:03.0000] <Dashiva>
No

[16:44:04.0000] <webben>
Hixie: conform on other points?

[16:44:05.0000] <Dashiva>
If it were, we wouldn't be having this discussion

[16:44:06.0000] <webben>
(if one care's about conformance at all)

[16:44:07.0000] <Hixie>
webben: not good enough

[16:44:08.0000] <Dashiva>
We'd just generate alt text

[16:45:00.0000] <Hixie>
webben: what should the spec say? if you don't have alternative text you... what? omit alt=""? include alt="{image}"? what?

[16:45:01.0000] <webben>
Hixie: what's not good enough? conforming to best practice on some points and not others ? you can't avoid that, since best practice will be avoided on business grounds.

[16:46:00.0000] <webben>
e.g. opening new windows sucks, but the business chooses to do it anyway, or whatever.

[16:46:01.0000] <Dashiva>
webben: So you'd rather have a photo site filled with bogus alt values, than a photo site with some useful and some missing alt values?

[16:46:02.0000] <Hixie>
webben: the spec not giving advice for the case where the html generator doesn't have suitable alternative text is not good enough

[16:46:03.0000] <webben>
Dashiva: Nothing I said implies that.

[16:47:00.0000] <Dashiva>
webben: Sure it did. You said a site has to force its users to provide alt to be conforming.

[16:47:01.0000] <Hixie>
webben: the spec _will_ give advice for the case where suitable alternative text is missing. the question is, what should that advice be?

[16:47:02.0000] <Dashiva>
Forcing users leads to all kinds of "interesting" input

[16:47:03.0000] <webben>
Hixie: I agree the spec probably should give advice. don't see why it needs to be a conformance point.

[16:47:04.0000] <Hixie>
webben: the spec saying "when you are in case X, do Y. Never do Y." is the height of idiocy in a spec.

[16:47:05.0000] <Hixie>
webben: it's what we call a contradiction.

[16:48:00.0000] <webben>
Dashiva: No. I said a site has to require text equivalents in order to provide the best experience.

[16:48:01.0000] <Hixie>
webben: it makes people ignore the spec as a whole

[16:48:02.0000] <Dashiva>
webben: Regardless of the reason, 'require' is the problem

[16:48:03.0000] <webben>
Dashiva: it may be good enough, for a given site, to provide less than the best experience.

[16:50:00.0000] <Dashiva>
webben: So you're saying there's no point in striving for the best

[16:50:01.0000] <webben>
Hixie: No, the spec would be saying if you put yourself in case X, you cannot expect the best experience.

[16:50:02.0000] <webben>
Dashiva: of course there's a point in striving for the best.

[16:51:00.0000] <Hixie>
webben: if the spec says to do X, it can't also say that doing X is not allowed. That's a contradiction.

[16:51:01.0000] <Dashiva>
webben: You said yourself that less than best is good enough

[16:51:02.0000] <jgraham>
webben: I dispute that prviding text equivalents for my flickr images would provide an improved user experience; the point of the images is their aesthetic value as images. Any replacement text (other than something banal like "phtograph") would just highlight my (lack of) writing skills

[16:51:03.0000] <webben>
Hixie: Which is simply true.

[16:51:04.0000] <Hixie>
as far as i can tell, you're just sticking your head in the sand here

[16:51:05.0000] <Hixie>
your position is logically contradictory and you're just denying that that is the case

[16:52:00.0000] <Hixie>
which doesn't make it any less the case

[16:52:01.0000] <Hixie>
if anything, it just makes it metacontradictory as well

[16:52:02.0000] <webben>
jgraham: Your inability to transmute the aesthetic experience into words (and I might well be just as bad) does not mean that such a transmutation would not provide a better user experience.

[16:53:00.0000] <webben>
jgraham: There do seem to be whole reams of writing about aesthetic experiences.

[16:53:01.0000] <Hixie>
webben: what should the spec say the author should do in the case of a page with a webcam?

[16:53:02.0000] <Dashiva>
webben: You can't have it both ways. Either striving for the best is worthwhile, and then people WILL strive for it. Or it's a waste of time, in which case people won't try, and we're wasting time defining it.

[16:54:00.0000] <webben>
Dashiva: Can't see where I'm trying to have it both ways.

[16:54:01.0000] <webben>
Dashiva: I think there are different "bests" in operation.

[16:54:02.0000] <jgraham>
webben: So given that I would be unable to communicate the aesthetic experience, even if I believed it was possible to achieve at all (I don't) who should I get to make my pages containing photos conforming?

[16:54:03.0000] <Dashiva>
webben: Two legs bad

[16:54:04.0000] <webben>
And it's important to clarify which of those "bests" conformance is about (if any).

[16:54:05.0000] <jruderman>
stop using a webcam because webcams are omg inaccessible. porn cam models, in particular, should use text only for cybersex.

[16:55:00.0000] <jgraham>
Or should publishing anything that cannot be made accessible to everyone be non-conforming?

[16:55:01.0000] <Hixie>
webben: sorry to keep hammering on this, but you haven't replied -- what should the spec say the author should do in the case of a page with a webcam?

[16:56:00.0000] <webben>
jgraham: I'm still trying to work out what HTML5 conformance is for, so it's difficult for me to answer. I'm not sure the way lines are being drawn in the sand makes much sense to me.

[16:56:01.0000] <Hixie>
conformance defines what you're allowed to do

[16:56:02.0000] <Dashiva>
webben: Conformance is what we should strive for

[16:56:03.0000] <webben>
Hixie: "allowed" as far as the validator is concerned

[16:56:04.0000] <webben>
Hixie: not "allowed" in terms of a) what works or b) what makes money

[16:57:00.0000] <Hixie>
conformance is the html5's spec way of saying "this is what is ok, and this is what you are not allowed to do"

[16:57:01.0000] <jgraham>
webben: I think the problem is that many people see conformance as a poliical tool that they can use o leverage some ideology

[16:57:02.0000] <Dashiva>
If conformance doesn't allow for making money, it's a pretty useless concept :)

[16:57:03.0000] <jgraham>
webben: for example universal access

[16:57:04.0000] <Dashiva>
jgraham: Indeed.

[16:57:05.0000] <Dashiva>
It's been all but admitted that the WCAG is useless without a law to enforce it, so as much as possible is being piggybacked onto HTML5

[16:58:00.0000] <jruderman>
i wouldn't want a law enforcing HTML5 conformance either...

[16:58:01.0000] <Hixie>
webben still hasn't actually replied to my question

[16:58:02.0000] <Hixie>
jruderman: i would!

[16:58:03.0000] <jgraham>
webben: The problem is that many of these ideologies are not practical or even, frankly, desirable

[16:58:04.0000] <webben>
jgraham: That doesn't really get at the nub of what conformance is for. It just says: conformance is not for universal access.

[16:59:00.0000] <Hixie>
jruderman: i could make the spec say "a page is conforming only if you have paid hixie $5"! :-D

[16:59:01.0000] <webben>
Hixie: I'm not sure why the syntax used is crucial to the discussion of what conformance is for.

[16:59:02.0000] <Dashiva>
Hixie: The google tax!


2008-08-20
[17:00:00.0000] <Hixie>
webben: i'm not discussing what conformance is _for_, i'm asking you for text that i can put in the spec that defines what a web page author should do when he wants to include a webcam

[17:00:01.0000] <jgraham>
webben: Conformance is about defining the technical consraints on what is allowed in a HTML document. It sets the boundaries on the syntax of the document, the ways in which that syntax may be combined and so on

[17:00:02.0000] <webben>
Hixie: Do you mean stills from a webcam when you say a webcam?

[17:01:00.0000] <webben>
jgraham: Yeah that's what conformance is. Not what it's for.

[17:01:01.0000] <webben>
e.g. one of the points Hixie made is about conformance to keep extension points open for future versions of HTML

[17:01:02.0000] <webben>
that makes some sense ... although conformance seems like a very weak stick for achieving that

[17:02:00.0000] <webben>
except in so far as it simply equates to carrots like data-*

[17:02:01.0000] <jgraham>
webben: I agree that conformance is a weak stick for achieving that 

[17:02:02.0000] <Hixie>
webben: i mean a webcam page like, for instance, this one: http://www.menet.umn.edu/coffeecam/

[17:02:03.0000] <Dashiva>
And we make it weaker the more we load onto it

[17:03:00.0000] <jgraham>
Dashiva: You type faster than me :)

[17:03:01.0000] <Dashiva>
I'm going to sleep now, so the stage is all yours in a bit

[17:03:02.0000] <Hixie>
webben: a page that shows an image that updates in realtime

[17:03:03.0000] <webben>
Hixie: If one were to try and establish best practice syntax for that with current tech, it would probably involve getting all the information you can get + indicating that that information may not be a "full" alternative.

[17:04:00.0000] <Hixie>
webben: what information could you get?

[17:05:00.0000] <Hixie>
webben: and how would you indicate it is not a "full" alternative?

[17:05:01.0000] <Hixie>
webben: could you show the markup you had in mind?

[17:05:02.0000] <jgraham>
webben: The inverse correlation (after a point) beween the difficulty of achieving confomance and the value that conformance is perceived to have is a real problem for proposals that seeks to impose stringent conformance requirements

[17:05:03.0000] <webben>
Hixie: I prefer the not-full-text-equivalent attribute type markers as a syntax.

[17:05:04.0000] <webben>
Hixie: what information you have depends on how frequently the scene changes

[17:06:00.0000] <webben>
Hixie: from the options I've seen so far.

[17:07:00.0000] <webben>
sorry "from the options I've seen so far", I prefer <alt="Webcam" no-text-equivalent>, but ideally with a bit fuller alt if you have more info.

[17:07:01.0000] <Hixie>
so if you have no information at all (e.g. it's a cam on someone's head) then you'd have:  <img src=cam.jpeg not-full-text-equivalent>  ?

[17:07:02.0000] <webben>
Hixie: No, I'd always include an alt.

[17:07:03.0000] <Hixie>
so  <img src=cam.jpeg alt="Webcam" not-full-text-equivalent>  ?

[17:07:04.0000] <Hixie>
how would you distinguish that from an image that just contained the text "Webcam" with fancy typography?

[17:08:00.0000] <jgraham>
Hixie: It has an extra attribute

[17:08:01.0000] <webben>
Hixie: you'd need fixed categories to do that.

[17:08:02.0000] <jgraham>
You could use user CSS or something to add a marker in the source

[17:08:03.0000] <Hixie>
jgraham: presumably the text "Webcam" wouldn't be a _full_ text equivalent for a fancy typographic effect of the text "Webcam"

[17:09:00.0000] <webben>
<img src="cam.jpeg" alt="Webcam on my head" not-full-text-equivalent type="webcam">

[17:09:01.0000] <Hixie>
webben: how about instead of "not-full-text-equivalent type" and duplicating the word "webcam", we just use a {...} indicator? as in <img src=cam.jpeg alt="{Webcam on my head}"> ?

[17:09:02.0000] <webben>
or  maybe the thing to do would be to mark images used for fancy typography instead!

[17:09:03.0000] <jgraham>
Hixie: Really? Did webben argue that or did someone else?

[17:10:00.0000] <Hixie>
(that's semantically equivalent, it's just a syntactic difference)

[17:10:01.0000] <webben>
Hixie: I don't think that's as understandable or as backwards compatible.

[17:10:02.0000] <Hixie>
jgraham: i'm arguing that :-)

[17:10:03.0000] <jgraham>
/me has pointed out why he doesn't like the curly braces before

[17:10:04.0000] <webben>
Hixie: but I agree, at that point it's a question of what syntax to use

[17:10:05.0000] <Hixie>
webben: it's actually more backwards-compatible, since it works in existing UAs without change. :-)

[17:11:00.0000] <webben>
Hixie: I don't count inserting {} as "working"

[17:11:01.0000] <jgraham>
Hixie: no-text-equivalent works in any UA that accepts user CSS

[17:11:02.0000] <Hixie>
anyway my point is that if that is acceptable, then we're basically (with minor syntactic differences maybe) at the point the spec is at

[17:11:03.0000] <jgraham>
It also doesn't change the semantics of existing documents

[17:11:04.0000] <jgraham>
Which isn't very "compatible"

[17:12:00.0000] <Hixie>
{..} is used rarely enough that it doesn't affect enough documents to matter either

[17:12:01.0000] <Hixie>
i checked

[17:12:02.0000] <Hixie>
before proposing it

[17:12:03.0000] <webben>
Hixie: yes, it's more a question of what use-cases it's meant to address. i.e. the statements about "best practice" the spec makes or doesn't make.

[17:13:00.0000] <Philip`>
Hixie: They're not equivalent forms with just syntactic differences, since you can't losslessly map <img alt={}> (in the no-text-equivalent attribute world) into the alt-surrounded-by-braces-means-no-text-equivalent world

[17:13:01.0000] <webben>
"enough documents to matter" is a worrying concept to me.

[17:13:02.0000] <jgraham>
Hixie: We went round this but I think the fact that it is used legitimatley makes it a bad idea. I also think it makes writing authouring tools unnecessarilly difficult

[17:13:03.0000] <webben>
especially when we're discussing conformance requirements.

[17:13:04.0000] <Hixie>
Philip`: ?

[17:13:05.0000] <webben>
"Conform to this to achieve the best effects ... until some future editor changes the spec and your particular communication isn't judged important."

[17:14:00.0000] <Hixie>
jgraham: *shrug*

[17:14:01.0000] <Hixie>
jgraham: i'm not convinced

[17:14:02.0000] <jgraham>
e.g. if flickr want to have a page shoing thumbnails with alt text title+username they need to check for the case that title tarts ith { and username ends with }

[17:14:03.0000] <Hixie>
jgraham: but i guess i'll see what evidence was put forward when i get around to the alt text feedback

[17:14:04.0000] <Hixie>
yeah but adding a whole attribute for this is just silly

[17:15:00.0000] <webben>
/me thinks that statement would be more persuasive if the underlying premises were clear.

[17:15:01.0000] <Hixie>
so i'm not convinced that the occasional clash is a problem really

[17:15:02.0000] <Philip`>
Hixie: If you have a document with "<img alt={} no-text-equivalent><img alt={}>", from a version of HTML that has the no-text-equivalent attribute, then you can't syntactically transform it into an equivalent document under the current HTML5 brace rules

[17:15:03.0000] <webben>
well, it's really a problem when there's really a clash.

[17:16:00.0000] <jgraham>
Hixie: I think the whole thing is rather silly and likely to be widely ignored but I think that designing somthing that is likely to be reliable when it is used is better than designing a system that we kno to be fragile if we bother designing anything at all

[17:16:01.0000] <Hixie>
Philip`: i don't understand why "<img alt={} no-text-equivalent>" would ever be valid in the version of HTML that has the no-text-equivalent attribute

[17:16:02.0000] <Hixie>
Philip`: what image would that be suitable for?

[17:17:00.0000] <webben>
didn't we establish images of some mathematical notation would be an example?

[17:17:01.0000] <webben>
or was that 100% speculative?

[17:18:00.0000] <jgraham>
Hixie: I am quite happy with making alt optional and forcing people who want their page to be accessible to actually put the work into checking that they have meaningful alt text here needed rather than just looking for a meaningless syntactic green light

[17:18:01.0000] <webben>
i.e. does { } not mean anything in maths?

[17:18:02.0000] <jgraham>
webben: There is a legitimate use case for putting TeX in alt

[17:18:03.0000] <webben>
jgraham: Yep, I'm not even talking about TeX though.

[17:19:00.0000] <Hixie>
jgraham: yeah maybe the kind of photo just isn't enough to make enough of a difference. but then people like webben will have a wobbly.

[17:19:01.0000] <webben>
e.g. { 5, 6, 7 } could be set notation maybe

[17:19:02.0000] <Hixie>
webben: such examples wouldn't have no-text-equivalent

[17:19:03.0000] <webben>
/me doesn't know enough about all the possible varieties of mathematical notation to make any clear guess.

[17:19:04.0000] <Philip`>
Hixie: Hmm, probably bad example - try "<img alt=Photo no-text-equivalent><img alt={Photo}>" (where the second image is the text "{Photo}" in a fancy font, hence that being the real text equivalent), which has no equivalent in the curly-braces-in-alt-are-special world

[17:19:05.0000] <webben>
Hixie: Then I don't understand your question. {} and  no-text-equivalent are alternatives aren't they?

[17:20:00.0000] <Hixie>
Philip`: yes, if your legitimate alternative text is "{photo}", then you can't include it in the current way html is specified.

[17:20:01.0000] <jgraham>
webben: The current rule seems to be "if you want to be robust against curly bracs always add whitespace in your alt attribute value"

[17:20:02.0000] <webben>
Hixie: how is a ua supposed to distinguish between { 5, 6, 7 } and {photo} ?

[17:21:00.0000] <jgraham>
i.e. alt=" { 5, 6, 7 } " vs alt={Photo}

[17:21:01.0000] <Hixie>
webben: per the spec right now, you can't include alternative text {5,6,7}. that's the problem jgraham mentioned.

[17:22:00.0000] <webben>
well, also, it would mess up previously create alt text like that

[17:22:01.0000] <jgraham>
webben: That is another problem I mentioned

[17:22:02.0000] <jgraham>
:)

[17:22:03.0000] <Hixie>
adding an attribute for this is something that i don't think makes sense, because this isn't an important enough case to have a whole attribute. so that leaves special syntax, or making alt="" optional altogether.

[17:22:04.0000] <Philip`>
Hixie: That's also a problem if you know you have legitimate alternative text (e.g. you have a script that transforms text into images in a special font) and can't be sure it's never going to have curly braces, so it's a problem regardless of how frequently curly braces occur in practice, so claims that they are rare are irrelevant

[17:23:00.0000] <Philip`>
Attributes are cheap - you've already added an infinite number of them, via data-* :-)

[17:23:01.0000] <Hixie>
attributes aren't cheap

[17:23:02.0000] <webben>
/me really doesn't understand why unimportant case => special freaky syntax

[17:23:03.0000] <webben>
important case => normal attribute

[17:23:04.0000] <hober>
/me preferred omitting @alt entirely for this case

[17:23:05.0000] <Hixie>
Philip`: the point is that if the clashes occur rarely, then all that will happen is that occasionally, the UA will say "image" when it doesn't have to -- which, relative to the number of pages that don't include alt="" at all, will hardly be noticed.

[17:24:00.0000] <jgraham>
I think unusual syntaxes are more expensive than atributes

[17:24:01.0000] <webben>
they're more expensive for aspiring developers at any rate

[17:24:02.0000] <Hixie>
that's possible

[17:24:03.0000] <webben>
case in point: most HTML4 authors don't understand &amp; in href

[17:24:04.0000] <webben>
s/most/many/

[17:25:00.0000] <Philip`>
Hixie: The point is that (some) people will want to avoid any clashes at all - they want to write code that works properly, not code that works properly except in some rare cases - so they'll have to deal with this new complexity even if there's little practical effect

[17:25:01.0000] <webben>
newbies don't understand why some attribute values are space separated, some comma separated

[17:25:02.0000] <webben>
syntax is hard to learn

[17:25:03.0000] <jgraham>
(because most systems for processing and consuming HTML are set up to deal with attributes already but fewer can deal with arbitary microsyntaxes in attributes)

[17:26:00.0000] <Hixie>
ok so if we go back to allowing alt="" to be omitted, who wants to go back to the public-html list and let people know?

[17:27:00.0000] <webben>
/me doesn't see the advantage of omitting alt.

[17:27:01.0000] <Hixie>
brb

[17:28:00.0000] <webben>
in that not-fully-equivalent alts are (I think) much less of a problem that missing text for image links

[17:28:01.0000] <jgraham>
Philip`: By the way Anne and I talked about meeting up later in the week if ou are still free

[17:28:02.0000] <jgraham>
/me -> bed

[17:28:03.0000] <webben>
/me -> bus

[17:28:04.0000] <Philip`>
jgraham: How much later in the week?

[17:29:00.0000] <jgraham>
Philip`: Thursday or Friday

[17:29:01.0000] <jgraham>
probably

[17:29:02.0000] <Philip`>
I ought to be free around then

[17:29:03.0000] <jgraham>
I guess Anne has more constraints than me

[18:44:00.0000] <Hixie>
how on earth do you manage the caret with contentEditable

[18:44:01.0000] <tantek>
cursor property?

[18:45:00.0000] <Hixie>
i mean the position

[18:45:01.0000] <tantek>
you mean the selection range?

[18:45:02.0000] <tantek>
flashing caret is just a degenerate selection case

[18:45:03.0000] <Hixie>
is it?

[18:45:04.0000] <tantek>
this is a solve problem in HyperCard

[18:45:05.0000] <tantek>
solved

[18:46:00.0000] <Hixie>
i'm not using hypercard :-)

[18:46:01.0000] <Hixie>
i mean in WinIE and other contentEditable-supporting browsers

[18:49:00.0000] <Hixie>
apparently webkit's Selection object does indeed contain a selection when there is a caret position to report

[18:49:01.0000] <Hixie>
of course IE doesn't suppot getSelection()

[18:49:02.0000] <tantek>
I meant as a scripting language construct

[18:49:03.0000] <tantek>
the essential schema of caret+selectedText

[18:50:00.0000] <tantek>
Hixie, IE does support a form of GetSelection - I've written favelets that use it.

[18:52:00.0000] <Hixie>
yeah, it has some weird range object thing instead of hte DOM Range object

[18:52:01.0000] <Hixie>
ok well anyway

[18:52:02.0000] <Hixie>
looks like we could do with a better caret api

[18:53:00.0000] <tantek>
as a previous work reference, HyperTalk had/has "selectedChunk" - http://www.mactech.com/articles/mactech/Vol.04/04.06/HyperCard1.2/index.html

[18:56:00.0000] <Hixie>
why is window.getSelection().getRangeAt(0).setStart(document.getElementsByTagName('p')[0].firstChild, 2) not working

[18:56:01.0000] <Hixie>
grr

[18:56:02.0000] <Hixie>
(webkit)

[18:57:00.0000] <Hixie>
it kinda works in firefox

[18:57:01.0000] <Hixie>
doesn't work in opera or webkit

[18:57:02.0000] <Hixie>
sigh

[20:49:00.0000] <takkaria>
grr, JSON gets a scheme language: http://json-schema.org/

[20:49:01.0000] <roc>
Hixie: you probably figured this out already, but the caret is just a degenerate ("collapsed") selection in Gecko

[00:58:00.0000] <MikeSmith>
hsivonen: v.nu still reports "Bad value “Content-Type” for attribute “http-equiv” on XHTML element “meta”"

[00:59:00.0000] <MikeSmith>
er, wait, oops

[00:59:01.0000] <MikeSmith>
/me does :set ft=html

[01:32:00.0000] <Lachy>
One more reason to hate Flash! http://blogs.zdnet.com/security/?p=1733

[01:34:00.0000] <Dashiva>
haha

[01:34:01.0000] <Dashiva>
Highlighted user comment is WorkForMe(tm)

[01:35:00.0000] <Lachy>
Dashiva, I don't see that anywhere

[01:40:00.0000] <Dashiva>
Lachy: The talkback section

[01:42:00.0000] <Lachy>
Dashiva, I checked there. I didn't find any "WorkForMe" in there

[01:42:01.0000] <Dashiva>
"nothing wrong with firefox and yahoo. they work fine together. must be your computer. "

[01:44:00.0000] <Lachy>
oh, that's what you're referring to? ok

[01:44:01.0000] <Dashiva>
Oh, I see I left out the s in Works

[02:53:00.0000] <anne_>
mibbit.com is nice

[03:34:00.0000] <anne_>
Philip`, we can prolly meet Thursday / Friday still

[03:49:00.0000] <Hixie>
/me watches the csswg spend multiple people's time discussing issues that they should just have someone read all the arguments for and just pick the best decision

[03:54:00.0000] <hsivonen>
I don't really believe that my editorial general comment about WCAG leads to anything, but at least from now on when people ask if I've taken my concern to the WG, I have.

[03:55:00.0000] <hsivonen>
more influential people have complained about the same thing for years to no avail

[03:57:00.0000] <virtuelv>
Hixie: weird stuff happens here: http://software.hixie.ch/utilities/cgi/unicode-decoder/character-identifier?keywords=right

[03:57:01.0000] <virtuelv>
do an inline find for "hamza"

[03:57:02.0000] <virtuelv>
notice anything special?

[03:58:00.0000] <virtuelv>
hm, different behavior in Opera and Firefox

[03:58:01.0000] <virtuelv>
wth

[04:03:00.0000] <anne_>
Hixie, several are just reading e-mail

[04:16:00.0000] <hsivonen>
is there any reasonable documentation about feeding script-generated subtrees to Renesis or MathPlayer in IE?

[04:17:00.0000] <hsivonen>
is that even possible?

[04:38:00.0000] <Lachy>
LOL, this letter the pirate bay wants people to send to the sweedish minister of justice if funny. http://thepiratebay.org/lovetpb.php

[04:49:00.0000] <anne_>
/me just realized that target=_blank is important for Web IRC and e-mail clients

[04:50:00.0000] <anne_>
(especially IRC)

[04:53:00.0000] <hsivonen>
all the bad stuff has good use cases

[04:55:00.0000] <gDashiva>
Maybe it's time for a change of attitude, and say "All the good stuff is bad for some use cases" :)

[04:55:01.0000] <Lachy>
I suppose, if you consider web based IRC to be a good idea

[04:56:00.0000] <hsivonen>
Lachy: shouldn't it be a good idea to provide Web-based implementations of anything?

[04:57:00.0000] <Lachy>
not everything is well suited to it. Some things really are better with native client side applications

[04:57:01.0000] <hsivonen>
better, sure

[04:58:00.0000] <hsivonen>
/me doesn't use a client side IRC app

[04:58:01.0000] <Lachy>
what do you use?

[04:58:02.0000] <hsivonen>
irssi over ssh

[04:59:00.0000] <anne_>
/me uses mibbit at the moment as the default IRC port is blocked

[04:59:01.0000] <Lachy>
that's useful for maintaining a persistent connection. Though I'd rather do the same thing with a BNC server and a regular client

[04:59:02.0000] <hsivonen>
irssi over ssh works from both Mac and N800

[05:00:00.0000] <Philip`>
/me used miau+mIRC for a while, but then gave up because it was rubbish and switched to ssh+irssi+screen

[05:00:01.0000] <hsivonen>
well, it kinda works from S60, too, but the text is tiny

[05:01:00.0000] <Lachy>
I tried psybnc on the weekend, but it did some things I didn't like, such as munging nicknames with the server prefix

[05:02:00.0000] <Lachy>
I'm going to try ezbounce later, and might try miau

[05:03:00.0000] <hsivonen>
the weirdest spec thing I've learned today: \u0000 is a permitted character in Java identifiers

[05:33:00.0000] <BenMillard>
zcorpan, do you still feel <dd> elements are better as alternatives to each other?

[05:34:00.0000] <BenMillard>
rather than a more loose interpretation, such as <li> has in <ul>

[05:40:00.0000] <jgraham>
Lachy: did you actually try mibbit? It's pretty good. 

[05:40:01.0000] <Lachy>
jgraham, no

[05:40:02.0000] <jgraham>
/me has started using screen+irssi+ssh too fwiw

[05:43:00.0000] <wilhelm>
You should add mutt and vim to that list, and you'll never have to leave your terminal again. (c:

[05:44:00.0000] <jgraham>
heh

[05:44:01.0000] <jgraham>
vim is too confusing for me :)

[05:45:00.0000] <Philip`>
Vim is easy :-)

[05:45:01.0000] <Philip`>
Just press 'i' and then use it like a sensible text editor

[05:46:00.0000] <jgraham>
Philip`: I learnt to type :q and export EDITOR=emacs and after that I stopped learning anything about vim

[05:47:00.0000] <Philip`>
/me uses the mswin.vim thing to get sensible keybindings for clipboards etc since he can never remember the proper Vim ways to access those features

[05:48:00.0000] <jgraham>
/me wishes emacs used either perl or python like regular expressions

[05:48:01.0000] <zcorpan_>
BenMillard: dunno, do you have data that suggests otherwise?

[05:49:00.0000] <zcorpan_>
BenMillard: i looked at your collection btw, pretty impressive

[05:49:01.0000] <BenMillard>
zcorpan, thanks...it's very rough at the moment :P

[05:50:00.0000] <zcorpan_>
yeah

[05:50:01.0000] <BenMillard>
I haven't studied it in detail, but I'm wondering about the markup I should use for the collection

[05:50:02.0000] <BenMillard>
many of the entries will have more than one piece of description

[05:51:00.0000] <BenMillard>
the the tables study, I took your interpretation and used <dd><ul><li>: http://projectcerbera.com/web/study/2007/tables/

[05:52:00.0000] <BenMillard>
but I'm not sure if that nesting is really helpful

[05:52:01.0000] <zcorpan>
probably isn't

[05:53:00.0000] <Lachy>
wtf? crazy people using terminal based text editors like vim?!

[05:54:00.0000] <BenMillard>
zcorpan, ok, I'll use whatever looks best when styling is disabled :P although that might well be the nested version

[05:54:01.0000] <Lachy>
/me uses TextMate

[05:54:02.0000] <virtuelv>
ed is good for you

[05:54:03.0000] <BenMillard>
I wanted to ask you first to ensure I wouldn't doing something incredibly dumb either way

[05:56:00.0000] <Philip`>
Lachy: I actually use Gvim instead, so it isn't terminal-based :-)

[05:56:01.0000] <Philip`>
It's got menu bars and toolbar buttons and everything

[05:57:00.0000] <zcorpan>
BenMillard: hey you're probably wiser than me about these issues :P

[05:57:01.0000] <BenMillard>
zcorpan, lol :D

[05:57:02.0000] <virtuelv>
jgraham: you want <ESC> :q!

[05:58:00.0000] <BenMillard>
zcorpan, I saw your Standards Suck interview about XHTML...you speak exactly as I imagined you would

[05:58:01.0000] <Lachy>
wtf? crazy people using non-terminal based text editors like Gvim?! :-)

[05:59:00.0000] <zcorpan>
BenMillard: cool

[06:00:00.0000] <virtuelv>
my text editor uses gecko

[06:00:01.0000] <virtuelv>
how crazy is that?

[06:00:02.0000] <Philip`>
virtuelv: Very

[06:00:03.0000] <BenMillard>
will HTMLWG be having a meeting somewhen? I can only see stuff about last year's one one http://www.w3.org/html/wg/

[06:01:00.0000] <BenMillard>
as in an F2F

[06:04:00.0000] <virtuelv>
Philip`: doing XHR from an editor macro is sweet, though

[06:04:01.0000] <virtuelv>
actually, just being able to write macros in JS is sweet

[06:05:00.0000] <BenMillard>
aha, I see HTML WG listed in a table here for October 2008: http://www.w3.org/2008/10/TPAC/Schedule.html#Thurs

[06:07:00.0000] <jgraham>
virtuelv: I tried using Komodo Edit for a while but kept doing ctrl-x-ctrl-s to try and save stuff which would save it but with the current line missing...

[06:07:01.0000] <Philip`>
/me thought the dated w3.org URLs were meant to give the date that the content was first published, not some arbitrary future date like 2008/10

[06:08:00.0000] <virtuelv>
jgraham: Hehe

[06:08:01.0000] <virtuelv>
you can write an extension that remaps all the keys

[06:08:02.0000] <svl>
"just as public internet facilities are responsible for porn, they ar also responsible for accessibility" - I wonder if he really meant that the way it reads...

[06:09:00.0000] <BenMillard>
Philip`, I guess they think a more guessable URL for that is more valuable.

[06:17:00.0000] <Philip`>
http://blog.mozilla.com/standards/2008/08/19/fear-and-loathing-on-the-standards-trail-with-an-upbeat-coda/

[06:48:00.0000] <billyjack>
Kuruma: お久しぶり

[07:06:00.0000] <hsivonen>
whee! the spec has again grown larger than the validator.nu max file size limit

[07:09:00.0000] <hsivonen>
I replaced assertions.sch with a custom Java class

[07:09:01.0000] <hsivonen>
I've tested it, but I may have missed something

[07:17:00.0000] <MikeSmith>
hsivonen: hmm, I've been relying on assertions.sch for something

[07:19:00.0000] <hsivonen>
MikeSmith: it still exists. but the deployment now fakes it

[07:19:01.0000] <hsivonen>
MikeSmith: so it looks like Validator.nu is using it but it's really doing something more efficient

[07:20:00.0000] <hsivonen>
MikeSmith: I intend to keep maintaining assertions.sch

[07:32:00.0000] <MikeSmith>
hsivonen: great that you will keep maintaining assertions.sch

[07:35:00.0000] <MikeSmith>
hsivonen: I've been meaning to ask you, would it be possible to create a set of schematron assertions that capture the same logic as your custom table-integrity checking code?

[07:36:00.0000] <hsivonen>
MikeSmith: not possible as far as I can tell

[07:36:01.0000] <MikeSmith>
OK

[07:40:00.0000] <MikeSmith>
hsivonen: where in svn is the source for the table-checker error messages?

[07:44:00.0000] <MikeSmith>
/me reads through org/whattf/checker/table source

[07:59:00.0000] <zcorpan>
hsivonen: http://www.htmlvalidator.com/CSEForum/viewtopic.php?p=2371#2371

[08:20:00.0000] <Lachy>
http://www.nytimes.com/2008/08/10/technology/10digi.html?_r=2&scp=1&%23038;sq=password&%23038;st=cse&%23038;oref=slogin&oref=slogin

[08:22:00.0000] <Lachy>
what they say about openid is interesting. OpenID was never really designed for logging on to sites like Yahoo or MySpace. It seems like it was more designed as a way to easily identify yourself in blog comments or forums, where there's little to no personal information about you

[08:22:01.0000] <Philip`>
Requires registration :-(

[08:23:00.0000] <Lachy>
ah, yeah, NYTimes sucks. Sometimes they say you need to register, other times they just let you see it

[08:23:01.0000] <takkaria>
Lachy: the idea was that openid would provide identity and then yahoo/myspace/whoever would use your ID as the key to your user data on their system

[08:23:02.0000] <takkaria>
at least, AIUI

[08:24:00.0000] <Lachy>
takkaria, that's not how I understood it when I first heard of it

[08:24:01.0000] <Lachy>
it's a totally insecure system for that

[08:24:02.0000] <Lachy>
it should be ditched immediately by all major sites that use it

[08:25:00.0000] <takkaria>
how is it any more insecure than making people sign up at each site?

[08:26:00.0000] <Lachy>
takkaria, because with OpenID, once you've phished the password once, you've got it for every site

[08:27:00.0000] <Lachy>
from wikipedia: "The original OpenID authentication protocol was developed in May 2005[6][7] by Brad Fitzpatrick, creator of popular community website LiveJournal, while working at Six Apart.[8] OpenID support was soon implemented on LiveJournal and fellow LiveJournal engine community DeadJournal for blog post comments,"

[08:28:00.0000] <Philip`>
When you make people sign up with separate accounts at every site, they'll still use the same password for all of them

[08:28:01.0000] <takkaria>
Lachy: but the majority of people use the same password for everything anyway

[08:28:02.0000] <takkaria>
and this way, you can change your openid password once and all your sign-ins are secure, there are none to miss

[08:28:03.0000] <Lachy>
"... a malicious relying party may forward the end-user to a bogus identity provider authentication page asking that end-user to input their credentials. On completion of this, the malicious party (who in this case also control the bogus authentication page) could then have access to the end-user's account with the identity provider, and as such then use that end-user’s OpenID to log into other services."

[08:28:04.0000] <takkaria>
yeah, that can happen, and the way to fix it was with client-side certificates

[08:29:00.0000] <Lachy>
takkaria, yeah, but not necessarily the same username though

[08:29:01.0000] <Philip`>
You should use an identity provider that provides better security than just asking for a password

[08:31:00.0000] <Lachy>
you would have to have a way to guarantee that the site you're logging into isn't simply proxying a copy of the identity provider's login page to you, and intercepting whatever login credentials you provide

[08:31:01.0000] <takkaria>
browser extensions are one way

[08:31:02.0000] <Philip`>
That's what TLS is for

[08:32:00.0000] <Lachy>
you can't fix a fundamentally insecure protocol by relying on the user to take care of security themsevles with browser extensions

[08:32:01.0000] <Philip`>
assuming you check the site's certificate is what you expect it should be, each time you log in

[08:32:02.0000] <Lachy>
user's don't check certificates. They see the padlock and think everything is ok cause its secure.

[08:33:00.0000] <Lachy>
EV certificates have improved the situation a little but, but it wouldn't help too much if the identity provider's login page was presented to the user in an iframe, where there is no address bar

[08:34:00.0000] <Philip`>
So how do you propose to solve phishing?

[08:34:01.0000] <Lachy>
I didn't say I had a solution to the phishing problem. Just that OpenID makes the problem worse

[08:34:02.0000] <Philip`>
(since it seems that's the problem, and it's not specific to OpenID at all)

[08:35:00.0000] <Lachy>
openid encourages users to use type exactly the same login credentials into any login box on any random site, which makes it worse

[08:36:00.0000] <Philip`>
OpenID allows a solution to the problem, since competition between identity providers should result in some providing an actually secure service, and then it'll be secure across all sites that use OpenID; otherwise you'd have to wait forever until each of those sites had upgraded to an unphishably secure system

[09:12:00.0000] <anne_>
/me wonders if the access folks actually read the new HTML5 proposal

[09:12:01.0000] <anne_>
/me thinks that most of them didn't

[09:13:00.0000] <zcorpan>
anne_: why do you think that?

[09:15:00.0000] <anne_>
because several talk or imply talking about it being optional

[09:17:00.0000] <Philip`>
They might just not be aware that HTML5 now has a new proposal

[09:26:00.0000] <gDashiva>
So are spammers hosting their own openid providers yet?

[09:27:00.0000] <zcorpan>
anne_: it is optional in one case (iirc)

[09:28:00.0000] <jgraham>
anne_: I think the spec doesn't explicitly say that alt is required except in one case

[09:29:00.0000] <jgraham>
To be clear, I mean it often says it is explicitly required 

[09:29:01.0000] <jgraham>
in one case it doesn't say it is explicitly required

[09:29:02.0000] <jgraham>
argh

[09:29:03.0000] <jgraham>
that was wrong

[09:30:00.0000] <jgraham>
in one case it says it is explicitly not required

[09:30:01.0000] <jgraham>
but it doesn't say that the cases metioned are supposed to cover all possible cases

[09:31:00.0000] <zcorpan>
there should be a default: useCommonSense();

[09:31:01.0000] <anne_>
oh it still is in some cases? interesting

[09:32:00.0000] <jgraham>
It should be reorganised like if foo then: bar otherwise if baz otherwise do the default thing

[09:32:01.0000] <zcorpan>
/me wonders if anne_ actually read the new HTML5 proposal :P

[09:32:02.0000] <jgraham>
if foo then: bar; otherwise if baz then: foobar; otherwise do the default thing would be closer to what I meant

[09:34:00.0000] <jgraham>
The only case when it's not required is in HTML email and communications intended for the exclusive use of a known audience or words to that effect

[09:34:01.0000] <jgraham>
(you could, perhaps, argue that is a non-Web use case)

[09:44:00.0000] <gsnedders>
ergh.

[09:44:01.0000] <gsnedders>
57 unread emails in the Flickr v. alt thread

[09:45:00.0000] <Philip`>
Your mail client should have a button to quickly rectify that situation

[09:46:00.0000] <gsnedders>
It indeed does.

[09:46:01.0000] <gsnedders>
But not a "read all and store summary of thread in brain" button.

[09:47:00.0000] <Philip`>
The summary should already be in your brain since it's pretty much the same as the previous three thousand alt emails

[09:48:00.0000] <gsnedders>
I am presuming this, but I am seeing more rational emails than I'd expect

[09:56:00.0000] <gDashiva>
Philip`: New project for you: Count the number of alt mails, cluster them, and graph it

[09:57:00.0000] <Philip`>
gDashiva: I think you are vastly overrating how much I care about it :-p

[09:58:00.0000] <MikeSmith>
graph the output to the null device

[10:02:00.0000] <gDashiva>
Philip`: What if you care about making me happy? :D

[10:04:00.0000] <anne_>
zcorpan, what's this thing you call HTML5?

[10:06:00.0000] <Philip`>
gDashiva: I think you are vastly overrating how much I care about enhancing your happiness ;-)

[10:11:00.0000] <gDashiva>
Philip`: You leave me with no option but... the dare. I bet you couldn't do it anyway, that's why you're avoid it.

[10:11:01.0000] <gDashiva>
-'re

[10:16:00.0000] <Philip`>
gDashiva: A cunning plan but one that is doomed to failure, since I really don't care, and also I must spend all my time playing TF2 instead of working on such things

[10:17:00.0000] <gDashiva>
/me gives up

[12:20:00.0000] <gsnedders>
takkaria: Are you really that recent around here? Oh well.

[12:21:00.0000] <takkaria>
gsnedders: no.  I've been hanging around since 2007, possibly somewhat earlier

[12:21:01.0000] <takkaria>
but I was lurk-only

[12:22:00.0000] <gsnedders>
/me wonders whether to bother to change it

[12:22:01.0000] <gsnedders>
/me decides he is too lazy

[12:47:00.0000] <hsivonen>
getting rid of Schematron reduce the RAM required for validating the spec by 12 MB

[12:47:01.0000] <hsivonen>
reduced even

[14:29:00.0000] <MikeSmith>
hsivonen: has getting rid of Schematron also reduced the time measurably?

[14:32:00.0000] <hsivonen>
MikeSmith: it depends mainly of how much memory is available

[14:32:01.0000] <hsivonen>
MikeSmith: in low memory conditions, it seemed to cut time to 25%

[14:32:02.0000] <MikeSmith>
wow

[14:32:03.0000] <hsivonen>
I haven't measured it in conditions where ample memory is available

[14:33:00.0000] <MikeSmith>
how low is low memory? I usually run my Debian VM with only 512MB

[14:33:01.0000] <MikeSmith>
is 512MB low?

[14:33:02.0000] <hsivonen>
low is so low that it barely has space to run

[14:33:03.0000] <hsivonen>
like 31...43 MB

[14:34:00.0000] <hsivonen>
with the new code in 32 bit mode, I can't go lower than 31 MB and validate the spec

[14:34:01.0000] <hsivonen>
with Schematron, I can't go lower than 43 MB

[14:35:00.0000] <MikeSmith>
I'm surprised to know that a Java app like this will even run with that little memory

[14:36:00.0000] <MikeSmith>
I mean of this kind (HTML/XML processing) and complexity

[14:36:01.0000] <MikeSmith>
but I guess it's all SAX-based, right?

[14:36:02.0000] <MikeSmith>
or buffered SAX?

[14:36:03.0000] <hsivonen>
now without Schematron, it's all streaming SAX

[14:37:00.0000] <MikeSmith>
ah, then I can see that would speed it up for sure

[14:37:01.0000] <MikeSmith>
so the buffered-SAX thing was only needed for Schematron processing?

[14:38:00.0000] <hsivonen>
but the 25% figure means nothing in practice. it just means that GC runs a lot when the app is almost out of memory all the time

[14:38:01.0000] <hsivonen>
MikeSmith: no, buffered SAX is for the benefit of other uses of the parser

[14:38:02.0000] <MikeSmith>
I see

[14:39:00.0000] <hsivonen>
MikeSmith: the Schematron validator has an internal tree

[14:39:01.0000] <hsivonen>
MikeSmith: so it doesn't use the HTML parser's buffered SAX code

[14:41:00.0000] <MikeSmith>
OK. so, about the code you've written to replace the Schematron step, this is probably a dumb question, but I'll ask anyway: Is it generalized or generalizable such that is could be used to replace Schematron processing for other arbitrary sets of assertions?

[14:45:00.0000] <hsivonen>
MikeSmith: it's not generalized

[14:45:01.0000] <MikeSmith>
OK

[14:45:02.0000] <hsivonen>
MikeSmith: it has some semi-general parts but those parts aren't paremetrized by the Schematron XML

[14:46:00.0000] <hsivonen>
since I figured that the more specialized case would have to be specific anyway

[14:46:01.0000] <hsivonen>
because I didn't want to write a streaming XPath implementation

[14:46:02.0000] <hsivonen>
it would be cool if someone else wrote one, though :-)

[14:48:00.0000] <MikeSmith>
heh, yeah, I can imagine you not wanting to write one but I don't think I'll hold my breath waiting for anybody to write one..

[14:48:01.0000] <hsivonen>
I think Michael Kay has written one, but it's not Open Source

[14:49:00.0000] <MikeSmith>
ah

[14:49:01.0000] <MikeSmith>
too bad

[14:50:00.0000] <gsnedders>
MikeSmith: On a totally different matter, do you have any opinion on sharing a room with me during the TPAC?

[14:51:00.0000] <MikeSmith>
gsnedders: I think I may be splitting an apartment for the week with Schepers

[14:51:01.0000] <gsnedders>
ah.

[14:52:00.0000] <gsnedders>
/me needs to find someone

[14:53:00.0000] <MikeSmith>
gsnedders: I would think asking others here on #whatwg could be good

[14:54:00.0000] <takkaria>
I would like to come to the TPAC but don't think my funds can stretch to it

[14:54:01.0000] <gsnedders>
Heh. I'm in France anyway, and I now have a flight back from Nice, so I've rather committed myself to going :)

[14:56:00.0000] <takkaria>
I have this suspicion I should be in uni at some point whilst it's on, too

[14:57:00.0000] <gsnedders>
takkaria: Excuses, excuses, excuses!

[15:02:00.0000] <jgraham>
I would but I can't go either because I'm in NZ

[15:03:00.0000] <MikeSmith>
hsivonen: I want to ask another dumb question, which is: For the non-schema processing bits of the v.nu code, I think it would embedding annotations in the code that provide assertion-like statements (in RFC 2119 language) about the particular constraints that part of the code is checking?

[15:03:01.0000] <MikeSmith>
e.g., "A table cell must not be overlapped by a later table cell."

[15:04:00.0000] <hsivonen>
MikeSmith: yeah, doing that would probably be good

[15:06:00.0000] <MikeSmith>
that way, it would be possible for me to extract those statements/assertions as least, if not an actual formal expression of the constraint being checked

[15:07:00.0000] <MikeSmith>
And would be helpful potentially to other implementors of conformance checkers

[15:10:00.0000] <MikeSmith>
hsivonen: if you do add something like that, I think that where possible, the annotation should include the associated element or element@attribute names it's checking

[15:10:01.0000] <MikeSmith>
e.g., "td|tr: A table cell must not be overlapped by a later table cell."

[15:11:00.0000] <MikeSmith>
or whatever

[15:37:00.0000] <gsnedders>
Anyone who vaguely knows me want to share a room with me at the TPAC (I'll be there, FWIW, from the 19th till the 25th, leaving in the evening, even though that is a rather useless detail)?


2008-08-21
[20:13:00.0000] <takkaria>
bloody hell, why does CC feel the need to use RDF for metadata?

[23:42:00.0000] <hsivonen>
takkaria: http://intertwingly.net/blog/2008/02/09/Mashups-Smashups#c1202870522

[01:27:00.0000] <zcorpan>
Hixie: what's the point in making xml:lang non-conforming in text/html?

[01:27:01.0000] <zcorpan>
Hixie: (i.e. lang in the XML namespace)

[01:28:00.0000] <Dashiva>
zcorpan: How would you ever get that?

[01:28:01.0000] <zcorpan>
Dashiva: with scripting or by moving a node from an xml document to an html document

[01:28:02.0000] <Dashiva>
But then you're working on the DOM form, not text/html anymore, aren't you?

[01:29:00.0000] <zcorpan>
Dashiva: when i say text/html i mean the spec's definition of "HTML document"

[01:37:00.0000] <hsivonen>
zcorpan: presumably, scripts written for both HTML and XHTML wouldn't cause the DOM to throw if allowed

[01:38:00.0000] <hsivonen>
/me isn't sure if the DOM throws now

[01:55:00.0000] <Hixie>
zcorpan: because you couldn't serialise it, so it would just get you in more trouble than it would be worth

[01:55:01.0000] <Hixie>
oh he left

[01:56:00.0000] <Hixie>
zcorpan needs to be online more. :-P

[01:56:01.0000] <philipj>
he's likely on the way to the office

[01:58:00.0000] <MikeSmith>
I wish I were on the way to the office there.. I'm sure it's nicer today than muggy and smoggy Tokyo in summertime

[01:59:00.0000] <MikeSmith>
rather, I wish I were in Bergen

[02:00:00.0000] <hsivonen>
Schematron sees to be one of those things whose design works nicely on today's desktop in non-batch mode but that causes grief when offered as an online service with concurrent users.

[02:01:00.0000] <hsivonen>
that is, 12 MB means nothing if you need 12 MB * 1 at a time on a desktop with 4 GB of RAM

[02:01:01.0000] <hsivonen>
s/sees/seems/

[02:04:00.0000] <Dashiva>
Hixie: So it's possible for a DOM to "be" text/html, and therefore unconforming, but you'd need a DOM validator to "prove" it?

[02:07:00.0000] <MikeSmith>
hsivonen: true perhaps of a lot of XML technologies in particular 

[02:08:00.0000] <hsivonen>
looks like news gets distorted by the time it hits slashdot: http://tech.slashdot.org/article.pl?sid=08/08/20/218240&from=rss

[02:11:00.0000] <Hixie>
Dashiva: DOMs are the only thing you ever conformance check, and they are always either HTML Documents or XHTML Documents

[02:12:00.0000] <Hixie>
Dashiva: search for "flagged as being HTML documents"

[02:14:00.0000] <Dashiva>
Hixie: But since you can't serialize it, you wouldn't be able to send it to the validator. So you'd need a UA-local validator (what I meant by DOM validator). Right?

[02:18:00.0000] <Philip`>
You could serialise to a special non-HTML format that is capable of representing any DOM and that is understood by your favourite validator

[02:19:00.0000] <anne_>
Hixie, all the "through the element" needs to be changed to "through the element or object"

[02:19:01.0000] <anne_>
Hixie, currently it makes no sense for Window etc.

[02:19:02.0000] <anne_>
Hixie, just "through the object" would work too probably

[02:21:00.0000] <hsivonen>
Dashiva: if someone contributes a streaming SAX parser for SDF... http://simon.html5.org/specs/sdf

[02:21:01.0000] <hsivonen>
or we get an XML5 that can serialize all DOMs

[02:23:00.0000] <Philip`>
/me wouldn't suggest using SDF, since it can have O(n^2) length compared to the XML-serialised length

[02:24:00.0000] <hsivonen>
Philip`: Namespaces strike again?

[02:25:00.0000] <Hixie>
Dashiva: you don't serialise it, you pass the serialised document to the validator and they turn it into a dom and check that

[02:25:01.0000] <Philip`>
hsivonen: No, just "<x><x><x><x><x><x>..." will result in loads of indentation

[02:26:00.0000] <Hixie>
anne_: yeah, i think i have a note about that somewhere. feel free to file a bug if you care enough. :-)

[02:29:00.0000] <hsivonen>
DOM Level 1 could be relatively reasonably by represented as JSON

[02:29:01.0000] <hsivonen>
too bad that Namespaces make even that inefficient

[02:39:00.0000] <Dashiva>
Hixie: The case we were talking about was invalid because of a script creating xml:lang, the validator might not run that script (if it runs scripts at all)

[02:40:00.0000] <Hixie>
Dashiva: ah yeah, checking such cases would need a full browser-like environment and a validator to be integrated

[03:32:00.0000] <Hixie>
well

[03:32:01.0000] <Hixie>
this is annoying

[03:33:00.0000] <Hixie>
TAB, CR, and LF aren't in Zs.

[03:33:01.0000] <anne_>
you lose!

[03:33:02.0000] <anne_>
and will not collect 200$

[03:34:00.0000] <Hixie>
i guess i should use White_Space instead

[03:34:01.0000] <Lachy>
/me wonders why people keep putting the $ on the wrong end. it goes at the beginning.

[03:34:02.0000] <Hixie>
though that includes U+000B too

[03:35:00.0000] <anne_>
Lachy, cause that's how you say it

[03:43:00.0000] <zcorpan>
Hixie: there are other things in the dom that can't be serialized but are still allowed

[03:44:00.0000] <anne_>
who is w?

[03:44:01.0000] <anne_>
(re: recent commits)

[03:44:02.0000] <zcorpan>
hsivonen: setting the lang attribute in the xml namespace in html documents doesn't throw

[03:45:00.0000] <Hixie>
anne_: wakaba

[03:45:01.0000] <Hixie>
zcorpan: so?

[03:45:02.0000] <Hixie>
zcorpan: (like what?)

[03:45:03.0000] <zcorpan>
Hixie: like a textarea containing a comment node

[03:45:04.0000] <zcorpan>
Hixie: or a comment containing -->

[03:46:00.0000] <Hixie>
true

[03:46:01.0000] <Hixie>
that's mostly just because it'd be a pain for me to specify otherwise, to be honest

[03:47:00.0000] <zcorpan>
ok

[04:15:00.0000] <Philip`>
Hixie: How would an SVG file not labelled as SVG being sniffed as a valid image in <img> allow privilege escalation, given that scripts in SVG in <img> must not be run?

[04:19:00.0000] <Hixie>
Philip`: it wouldn't, but if scripts ever were to run, it would

[04:19:01.0000] <Hixie>
Philip`: it's more defence in depth than anything really

[04:20:00.0000] <anne_>
I think those rules match what we implement (though we also fail if the root element is not <svg> in the right namespace)

[04:20:01.0000] <Philip`>
Oh, okay

[04:21:00.0000] <Hixie>
anne_: that matches "User agents must not support non-image resources with the img element (e.g. XML files whose root element is an HTML element)."

[04:21:01.0000] <anne_>
ah

[04:28:00.0000] <Hixie>
ok bed time

[04:28:01.0000] <Hixie>
nn

[04:36:00.0000] <zcorpan>
it seems karl misunderstood the spec change wrt xml:lang

[04:46:00.0000] <anne_>
seems he did

[04:46:01.0000] <anne_>
admittedly it's quite hard

[04:47:00.0000] <anne_>
especially with the {xml:lang} versus xml{lang} nonsense

[04:51:00.0000] <hsivonen>
Namespaces make things hard.

[04:52:00.0000] <anne_>
I suppose this could be added to the anecdotical evidence thing

[04:53:00.0000] <hsivonen>
I made the Validator.nu error messages strictly correct but not necessarily obvious on this point as a juvenile act of rebellion against the badness of namespaces

[04:53:01.0000] <anne_>
maybe I should add the professor from my XML class who thought namespaces inherited into attributes

[04:53:02.0000] <anne_>
hsivonen, :p

[04:53:03.0000] <anne_>
lol, mibbit has a weird smiley for that one

[04:54:00.0000] <hsivonen>
anne_: your professor is in good company, since it seems that a W3C spec editor can have that misconception

[05:00:00.0000] <zcorpan>
hsivonen: <html xml:lang='en' lang='sv'> validates but <html lang='sv' xml:lang='en'> doesn't

[05:00:01.0000] <hsivonen>
hmm. that's weird

[05:07:00.0000] <zcorpan>
hsivonen: event-source was renamed to eventsource

[06:32:00.0000] <hsivonen>
zcorpan: <html xml:lang='en' lang='sv'> fixed

[06:32:01.0000] <hsivonen>
thanks

[06:48:00.0000] <hsivonen>
hmm. perhaps the spec check-in for renaming event-source wasn't properly annotated...

[06:48:01.0000] <hsivonen>
fixing anyway :-)

[06:51:00.0000] <anne_>
ah hsivonen has a flickr backup utility, good

[07:02:00.0000] <hsivonen>
hmm. I haven't updated Validator.nu's copy of the spec since March 20...

[07:11:00.0000] <hsivonen>
MikeSmith: do you need assertions.sch to work in an XPath 1.0 environment?

[07:12:00.0000] <hsivonen>
MikeSmith: that is, would your setup break if I start using XPath 2.0?

[07:43:00.0000] <Philip`>
/me notices that he failed to notice that the OpenGL 3.0 spec was released last week, after months of silence and delays, seemingly without any of the interesting features that had been promised earlier

[07:43:01.0000] <Philip`>
It seems the most radical change they've made is "In OpenGL 3.0 we have introduced a deprecation mechanism for the first time in OpenGL’s 16-year history."

[07:47:00.0000] <Philip`>
It doesn't seem entirely clear how the deprecation thing is meant to work - GL 3.0 implementations let you choose a "forward compatible" mode, which disables all the deprecated features, but also have to have a "full" mode which still provides them and is compatible with all old (GL <= 2.1) software

[07:48:00.0000] <Philip`>
so I guess the idea is that the features would be removed from the GL 3.1 "full" mode too, or maybe they're just going to hang around until GL 4.0, or until forever, since nobody's actually going to bother rewriting their code to not use the deprecated features

[08:10:00.0000] <MikeSmith>
hsivonen: using XPath 2.0 would not break anything for me

[08:19:00.0000] <Lachy>
http://ejohn.org/blog/queryselectorall-in-firefox-31/

[08:21:00.0000] <Lachy>
JohnResig, are the performance tests you used for that available? I'd like to test in them in Opera

[08:45:00.0000] <BenMillard>
each heading in my web page collection now provides an anchor point (via ID) and states how many entries are in that section: http://projectcerbera.com/web/study/2008/collection

[08:46:00.0000] <BenMillard>
Hixie, ^ is that a useful change?

[09:29:00.0000] <hsivonen>
MikeSmith: ok

[09:30:00.0000] <hsivonen>
Philip`: are the deprecated features conflicting with some insanely great improvements or are they just "bad"?

[09:31:00.0000] <anne_>
/me thought 1 and 2 were already incompatible

[09:31:01.0000] <anne_>
/me thus wonders what deprecation means in the context of OpenGL

[09:36:00.0000] <zcorpan>
http://tinyurl.com/559u23 (live dom viewer)

[09:37:00.0000] <zcorpan>
um, you need to insert a character before the script in firefox to work around the parsing bug...

[09:38:00.0000] <MikeSmith>
if there are cases where

[09:38:01.0000] <MikeSmith>
oops

[09:38:02.0000] <MikeSmith>
(ignore that)

[09:44:00.0000] <zcorpan>
http://tinyurl.com/5rj9l5 works in fx and saf

[09:44:01.0000] <zcorpan>
is there a spec that covers this?

[09:45:00.0000] <anne_>
ECMAScript?

[09:45:01.0000] <anne_>
though maybe I'm misunderstanding what you're testing

[09:51:00.0000] <Philip`>
hsivonen: They're not inherently bad ideas, they're just decade-old ideas and there are much better (more powerful, more efficient) alternative features that achieve the same effect on modern hardware

[09:52:00.0000] <Philip`>
like the whole glBegin();glVertex();glEnd() thing is deprecated, in favour of vertex arrays

[09:53:00.0000] <Philip`>
(The problem with all the legacy features is that they make drivers extremely complex, and make it hard to add new features since you have to consider interactions with all the old ones)

[09:54:00.0000] <Philip`>
anne_: OpenGL 1 and 2 aren't incompatible, in the sense that a program using the OpenGL 1.x API will still work in a GL 2.1 implementation (and will still work in GL 3.0 in 'full' mode)

[09:54:01.0000] <gDashiva>
zcorpan: What anne said, what is it that's working?

[09:56:00.0000] <anne_>
philip, interesting

[09:56:01.0000] <zcorpan>
anne_: ok

[09:57:00.0000] <gavin_>
/me just read http://lists.w3.org/Archives/Public/public-html/2008Aug/0510.html and couldn't agree more

[09:57:01.0000] <gavin_>
jgraham++

[10:06:00.0000] <gDashiva>
gavin_: Now watch it be completely ignored, or if we're lucky, marked as a troll.

[10:06:01.0000] <gsnedders>
/me has a lucky feeling about this one

[10:10:00.0000] <BenMillard>
I set my e-mail client so that "alt" in a subject line gets the message automatically marked as "Read"

[10:10:01.0000] <gavin_>
http://lists.w3.org/Archives/Public/public-html/2008Aug/0517.html is mind boggling

[10:11:00.0000] <BenMillard>
there's actually very little which gets through as unread in my e-mail client

[10:17:00.0000] <anne_>
gavin_, yeah, there's an enormous gap between the two camps

[10:23:00.0000] <gDashiva>
BenMillard: I hope it's more like \balt\b :)

[10:23:01.0000] <Philip`>
Some subject lines just say "... alternative text ...", so you don't want the trailing \b

[10:24:00.0000] <BenMillard>
gDashiva, the less I see the more I know. ;)

[10:25:00.0000] <gDashiva>
We need a neural network to determine if an occurence of 'alt' refers to alternate text for HTML or not

[10:26:00.0000] <Philip`>
A much simpler heuristic is to look at the size of the thread - if the subject contains the substring 'alt' and there are at least 20 messages, it must be about <img alt>

[10:26:01.0000] <BenMillard>
the only things that interest me on Public-HTML are links to use cases and the results of research...maybe I should use a whitelist instead of a blacklist?

[10:27:00.0000] <gDashiva>
BenMillard: Nah, there's plenty of namedropping in the alt threads about research (or the need for it, or the non-need for it)

[10:28:00.0000] <BenMillard>
ok, I'll stick with what I've got...it ain't broke, so don't fix it

[10:29:00.0000] <BenMillard>
I've decided to do a bit of analysis of the collection before my midpoint comes up

[10:30:00.0000] <BenMillard>
this means figuring out things are useful to count

[10:30:01.0000] <BenMillard>
here's what I think for headings: http://projectcerbera.com/web/study/2008/collection#headings

[10:31:00.0000] <BenMillard>
Philip`, I think some of the heading analysis could be automated

[10:31:01.0000] <BenMillard>
like checking when heading levels get skipped

[10:32:00.0000] <BenMillard>
other things need judgement, like figuring out when the headings should have changed level but didn't

[10:35:00.0000] <jgraham>
BenMillard: Are you planning to look at table summaries? contrary to Al's opinion I think it would be exceedingly useful to have data here like 10% of @summaries are accurate descriptions of the table, 30% are SEO keywords 20% are descriptions of a table with a different structure etc.

[10:36:00.0000] <Philip`>
..., 90% are saying that it's a layout table, etc

[10:36:01.0000] <BenMillard>
jgraham, yes

[10:36:02.0000] <jgraham>
BenMillard: Cool.

[10:36:03.0000] <BenMillard>
although I thought summary had already been studied in quite large numbers by Mark Pilgrim or someone?

[10:37:00.0000] <anne_>
that was longdesc

[10:37:01.0000] <BenMillard>
anne_, ah, that must be it

[10:37:02.0000] <anne_>
(if we remember the same thing, yes)

[10:38:00.0000] <BenMillard>
anne_, yes, I just found it and that jogged my memory: http://blog.whatwg.org/the-longdesc-lottery

[10:38:01.0000] <BenMillard>
jgraham, the collection currently mentions the summary attribute 6 times at the moment

[10:39:00.0000] <jgraham>
BenMillard: OK. I guess it is also something that might be worth looking at from a different same e.g. a random sample from google to see how that influences the results

[10:40:00.0000] <hober>
hmm.

[10:40:01.0000] <hober>
awfully tempted to feed the trolls.

[10:40:02.0000] <hober>
WDYT? http://edward.oconnor.cx/tmp/alt-reply.txt

[10:40:03.0000] <BenMillard>
jgraham, yes that would be good. here's ~800 from Philip`: http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/tagattr/table/summary

[10:41:00.0000] <Philip`>
http://canvex.lazyilluminati.com/misc/summary.html

[10:41:01.0000] <gDashiva>
hober: Seems fine to me

[10:41:02.0000] <BenMillard>
Philip`, ooh that's saved me some work :P

[10:42:00.0000] <Philip`>
hober: Judging by the URL, it's talking about alt, which is a bad sign :-p

[10:43:00.0000] <hober>
Philip`: well, that may have been a poor choice of names. it's more of a side issue from that interminable thread

[10:44:00.0000] <BenMillard>
Philip`, is that summary page using the 799 pages in your 2007-07-17 collection? If not, can you tell me how many pages it covers?

[10:45:00.0000] <Philip`>
hober: I think you typoed "editor" as "editors" :-)

[10:47:00.0000] <Philip`>
BenMillard: http://krijnhoetmer.nl/irc-logs/whatwg/20070513#l-348 - it looks like that came from a list of pages from Yahoo (passing it common words ("a", "the", etc) as search terms)

[10:48:00.0000] <BenMillard>
Philip`, so it's covering 105 pages?

[10:49:00.0000] <hober>
Philip`: :)

[10:49:01.0000] <Philip`>
(It'd be fairly trivial to get a list of all the summary values used my newer collection of ~130K pages, though it'd take significant effort to categories them)

[10:49:02.0000] <Philip`>
BenMillard: According to myself (who I wouldn't necessarily trust), it is

[10:50:00.0000] <BenMillard>
yes, I wouldn't like to categories ~1,000 uses of summary

[10:50:01.0000] <Philip`>
It'd be easy to count how many contain the string "layout", though

[10:50:02.0000] <BenMillard>
indeed

[10:51:00.0000] <BenMillard>
if you randomly take summary strings from ~105 pages from that ~130,000, I could add those to study 210

[10:51:01.0000] <BenMillard>
as this seems to be a hot topic which reccurs

[10:55:00.0000] <BenMillard>
Philip`, could you count the total number of tables in the big and then see what proportion of them used summary?

[10:55:01.0000] <BenMillard>
*big sample

[10:59:00.0000] <JohnResig>
is dev.whatwg down?

[11:05:00.0000] <heycam>
JohnResig, it's very slow today

[11:07:00.0000] <BenMillard>
jgraham & Philip`, just noticed my collection contains "Summary="a summary of ..." where ... is the <caption> text." for this: http://dev.opera.com/articles/view/19-html-tables/#morestructure

[11:08:00.0000] <BenMillard>
should I studying each summary value in the context of the page it came from?

[11:08:01.0000] <BenMillard>
then I could check if it repeats the previous sentence, or is the previous heading rephrased slightly, etc

[11:13:00.0000] <Philip`>
BenMillard: Hmm, I can't look at anything now, since the university's internet connection has died, so I can't access all my files

[11:13:01.0000] <BenMillard>
Philip`, that's ok, I'm not your boss :)

[11:14:00.0000] <BenMillard>
you can do it whenever is convenient for you

[11:14:01.0000] <Philip`>
You might want to remind me later because otherwise I'll just forget :-)

[11:14:02.0000] <BenMillard>
Philip`, I have the same affliction :P

[11:14:03.0000] <BenMillard>
I'll put a note in the collection, then I'll stumble upon it from time to time anyway

[11:15:00.0000] <anne_>
jgraham, Philip`, WG dinner tonight, maybe tomorrow

[11:15:01.0000] <anne_>
/me has to go

[11:15:02.0000] <Philip`>
Hopefully it won't be long until somebody notices the whole university is offline and plugs the cable back in

[11:15:03.0000] <Philip`>
anne_: Tomorrow sounds reasonable to me

[11:18:00.0000] <jgraham>
Philip`: Someone just plugged me back in :)

[11:19:00.0000] <Philip`>
BenMillard: I see 3301 pages using @summary (on any element, but I assume pretty much all on table), out of 92066 pages using table

[11:19:01.0000] <Philip`>
jgraham: Indeed, I just noticed it started working too :-)

[11:20:00.0000] <Philip`>
/me likes how the CL has a totally independent open wireless network that just uses standard ADSL, so he can still connect when the proper network is down

[11:22:00.0000] <Philip`>
Oh, I've already got that data online - http://philip.html5.org/data/tag-count-pages.txt and http://philip.html5.org/data/attr-count-pages.txt

[11:22:01.0000] <Philip`>
(Excludes attributes with fewer than five occurrences, because there were zillions of them)

[11:23:00.0000] <BenMillard>
Philip`, awesome. :)

[11:25:00.0000] <Philip`>
(Oops - when I say "totally independent", I don't mean that at all, since it uses the same wireless access points and the internal CL network; but it's independent in its connection to the outside world)

[11:26:00.0000] <Philip`>
Oh, they broke the university again

[11:26:01.0000] <Philip`>
/me goes home

[11:28:00.0000] <hsivonen>
/me wonders if key compiler devs are on board with C++0x

[11:29:00.0000] <Philip`>
GCC and MSVC have already shipped parts of it, so I assume they are

[11:29:01.0000] <hsivonen>
ok

[11:31:00.0000] <BenMillard>
Philip`, can you count the number of summary attributes and the number of table elements rather than the number of pages?

[11:31:01.0000] <jcranmer>
well, boost provides most of the C++0x libraries

[11:31:02.0000] <hsivonen>
clearly, C++ embodies drastically non-Exupérian design principles

[11:31:03.0000] <Philip`>
BenMillard: http://philip.html5.org/data/tag-count-total.txt and http://philip.html5.org/data/attr-count-total.txt

[11:31:04.0000] <BenMillard>
Philip`, aha!

[11:31:05.0000] <Philip`>
BenMillard: (940942 table, 12834 @summary)

[11:32:00.0000] <Philip`>
/me fails to find helpful definitions of Exupérian on Google

[11:33:00.0000] <Philip`>
Aha, "Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away."

[11:34:00.0000] <hsivonen>
Philip`: that's the one

[11:34:01.0000] <BenMillard>
I prefer: "Perfection is having just enough."

[11:34:02.0000] <Philip`>
C++ has never had perfection as a design principle - it's just designed to be useful :-)

[11:57:00.0000] <BenMillard>
Philip`, 940,942 <table> elements with 12834 summary attributes means summary attribute appears on 1.3% of all <table> elements in that sample of 130,000 pages, right?

[11:57:01.0000] <BenMillard>
(maths isn't my strong point, even basic maths)

[11:58:00.0000] <Philip`>
BenMillard: Yes

[11:58:01.0000] <Philip`>
BenMillard: (Well, 1.4% if you round-to-nearest)

[11:58:02.0000] <Philip`>
BenMillard: (And not precisely 130,000, but close enough)

[11:58:03.0000] <BenMillard>
LOL, I've forgotten how to round numbers :D

[11:59:00.0000] <BenMillard>
940,942 <table> elements with 12,834 summary attributes means summary attribute appears on 1.4% of all <table> elements in that sample of ~130,000 pages?

[12:00:00.0000] <Philip`>
Yes

[12:00:01.0000] <jcranmer>
Philip`: round half-up, half-down, half-even, half-odd?

[12:00:02.0000] <Philip`>
jcranmer: Doesn't matter, since it's 1.36...% so it's not a half :-p

[12:17:00.0000] <BenMillard>
Philip`, the summary proportion looks about right compared with this (bottom of the <table> attributes list): http://code.google.com/webstats/2005-12/tables.html

[12:40:00.0000] <BenMillard>
Philip`, thanks for help with that. I'm off for dinner now, cya.

[13:06:00.0000] <hsivonen>
jgraham++

[13:07:00.0000] <hsivonen>
jgraham: there seems to be no documented reasoning why action 72 is a better solution to issue 20 than doing the kind of tweaks you mentioned

[13:13:00.0000] <virtuelv>
/me thinks that include proposal would mostly be solved with an equivalent to -o-content-size

[13:14:00.0000] <virtuelv>
http://lists.w3.org/Archives/Public/www-style/2008Jul/0380.html

[13:16:00.0000] <virtuelv>
the security implications you get in combination with legacy cmses and clients supporting something like <include /> just strikes me as "potentially ugly"

[13:33:00.0000] <hsivonen>
what does Dreamweaver generate into summary='' by default?

[14:07:00.0000] <jgraham>
hsivonen: I have a feeling that Dreamweaver generates the dimensions of the table, which could clearly just be determined on the client side (I might be wrong about what it does though; this is just a memory of the last time this discussion came up)

[14:09:00.0000] <hsivonen>
jgraham: sad if that's the case

[16:47:00.0000] <rubys>
anybody know the current implementation status of web sockets?  A pointer to a webpage with this answer would be fine...

[16:53:00.0000] <Hixie>
i believe no browser implements it, but mcarter has a js-based shim implementation iirc

[16:54:00.0000] <rubys>
http://www.orbited.org/

[16:57:00.0000] <rubys>
That looks to be the one.  Thanks!  Two more (hopefully quick) questions.  Is the video element compatbile with rtsp?  Is there anything in HTML5 or related work that addresses video capture?

[16:58:00.0000] <Hixie>
<video> is intended to be compatible with any streaming protocol, though if there are any issues please do raise them and i'll try to address them

[16:59:00.0000] <rubys>
fair enough

[16:59:01.0000] <Hixie>
regarding video capture, do you mean streaming video from a camrea, or taking stills from a video stream?


2008-08-22
[17:00:00.0000] <rubys>
Currently, there are plugins that do capturing and serving of video in a separate process.  I have no idea whether there is anything in HTML5 which would enable this to be moved into the browser.

[17:02:00.0000] <roc>
there is an old W3C draft that specifies <input type="file" device="camera"> or something like that

[17:04:00.0000] <Hixie>
rubys: if you mean camera capture, then no, there currently isn't, though many people have requested it and i imagine that in due course (when browsers have implemented more of html5) it would be something we would want to add

[17:04:01.0000] <rubys>
thanks!

[17:09:00.0000] <roc>
Hixie: Brad Lassey (I think) has been looking at doing camera integration using something like that W3C draft

[17:12:00.0000] <Hixie>
roc: cool

[17:22:00.0000] <franksalim>
rubys: like Hixie said, orbited has a js based shim implementation. kaazing has a flash/js shim that exposes the websocke api and tries to speak the wire protocol when flash is available

[17:23:00.0000] <franksalim>
this would be a good time to point out that my words do not represent my friends or my employers :-p

[17:23:01.0000] <franksalim>
*websocket

[17:48:00.0000] <rubys>
franksalim: thanks!

[17:54:00.0000] <franksalim>
rubys: np! feel free to pick my brain if you have any questions about that sort of thing

[17:56:00.0000] <rubys>
at the moment, I'm just relaying questions from somebody who is trying to decide if html5 is worth exploring.  Having two sets of "shims' that allow you to get by until the browsers catch up is a good thing.

[01:34:00.0000] <Hixie>
hsivonen: yt?

[01:50:00.0000] <hsivonen>
Hixie: yes

[01:58:00.0000] <Hixie>
hsivonen: i assume you don't want ftp:// but https:// is ok?

[02:02:00.0000] <hsivonen>
https works

[02:02:01.0000] <hsivonen>
ftp not

[02:02:02.0000] <hsivonen>
I'm using Commons HttpClient

[02:05:00.0000] <Hixie>
right-o

[02:05:01.0000] <Hixie>
/me updates his script and reruns it to filter out ftp://

[02:06:00.0000] <BenMillard>
publishing the progress of my collection isn't major enough to get into Mark Pilgrim's summary :( http://blog.whatwg.org/this-week-in-html-5-episode-3

[02:06:01.0000] <Hixie>
BenMillard: i don't think mark is reading the mailing list. however, you should feel free to publish your own blog post announcing it, i think that would be great actually

[02:08:00.0000] <Hixie>
hsivonen: so i am getting you 1,000,000 urls taken from an arbitrarily sized subset of google's index that represents what google thinks are the most important paegs (for some vague definition of "important" that really gets kinda hard to pin down when you're talking about such large samples)

[02:09:00.0000] <Hixie>
hsivonen: the random sample is biased as follows:

[02:09:01.0000] <Hixie>
hsivonen: i dropped all pages that we didn't determine were HTML when we fetched them

[02:09:02.0000] <Hixie>
hsivonen: i dropped all pages whose urls didn't start with "http"

[02:10:00.0000] <Hixie>
hsivonen: i dropped 80% of pages whose urls only had 3 "/" characters in them and whose urls' last character was a "/"

[02:11:00.0000] <Hixie>
hsivonen: and i only took one page per cluster of pages that seemed to consist a single "site"

[02:12:00.0000] <BenMillard>
Hixie, I made a little one on my own website the other day: http://projectcerbera.com/blog/2008/08/midpoint

[02:13:00.0000] <BenMillard>
so I guess a WHATWG entry for my research would be redundant? although I would've liked a paragraph in Mark's entry :P

[02:14:00.0000] <hsivonen>
BenMillard: the JS live DOM didn't make it in last week, either.

[02:15:00.0000] <hsivonen>
BenMillard: it's a summary of spec checkins

[02:15:01.0000] <hsivonen>
BenMillard: but you can write a post yourself

[02:15:02.0000] <hsivonen>
(for people who subscribe to the blog in general instead of the markp-only feed)

[02:16:00.0000] <BenMillard>
ah, ok

[02:16:01.0000] <Hixie>
BenMillard: don't worry, he got some of my changes wrong anyway :-) (i added a comment to the blog entry correcting it, feel free to add a comment if you don't want to add a whole blog post)

[02:16:02.0000] <BenMillard>
Hixie, oh yeah, a comment is a good idea

[02:21:00.0000] <Hixie>
man, backing up half a terabyte over USB 2.0 is not fast

[02:21:01.0000] <BenMillard>
I set my Name to be "Ben 'Cerbera' Millard" and each time a Preview I get a \ before each '. So, after a few previous I see "Ben \\\\'Cerbera\\\\' Millard". would be nice if that didn't happen

[02:21:02.0000] <BenMillard>
*a few previews

[02:22:00.0000] <Hixie>
probably a wordpress bug

[02:22:01.0000] <Hixie>
does wordpress use php?

[02:22:02.0000] <Hixie>
i've found php apps uniformly get ' escaping wrong

[02:22:03.0000] <hsivonen>
it does

[02:22:04.0000] <Hixie>
dunno why

[02:22:05.0000] <BenMillard>
yes, it does look like a magic quote time thing

[02:22:06.0000] <BenMillard>
*type (yikes, my typing is groggy this morning)

[02:22:07.0000] <hsivonen>
the magic quote thing is in the worst ideas ever category

[02:23:00.0000] <BenMillard>
I've made the comment but it hasn't appeared...there's no message but I *assume* it's in a moderation queue of some sort and hasn't been lost?

[02:23:01.0000] <BenMillard>
I used some HTML and some links

[02:23:02.0000] <hsivonen>
I'll check the queue

[02:24:00.0000] <hsivonen>
BenMillard: now approved

[02:25:00.0000] <BenMillard>
yay, thanks! for the IRC records, the comment is precisely here: http://blog.whatwg.org/this-week-in-html-5-episode-3#comment-26873

[02:25:01.0000] <Hixie>
http://%3cbr%3etyxynola.newmail.ru/_visits_26.html is a weird url

[02:25:02.0000] <Hixie>
but it seems to work in safari

[02:25:03.0000] <Hixie>
i don't understand

[02:26:00.0000] <Hixie>
hsivonen: do you want these in lexical order or random order?

[02:26:01.0000] <hsivonen>
Hixie: random is better

[02:27:00.0000] <hsivonen>
either works

[02:27:01.0000] <Hixie>
i have them in random order

[02:27:02.0000] <Hixie>
so that works

[02:27:03.0000] <Hixie>
that url doesn't work in firefox

[02:27:04.0000] <Hixie>
and it gives a different page in IE than WebKit

[02:28:00.0000] <Hixie>
doesn't help that i can't speak russian

[02:28:01.0000] <Hixie>
opera doesn't like that url either

[02:28:02.0000] <Hixie>
oh well

[02:29:00.0000] <Hixie>
this URL list is 52MB uncompressed

[02:29:01.0000] <Hixie>
20MB compressed

[02:32:00.0000] <BenMillard>
which flavour of compression?

[02:33:00.0000] <hsivonen>
I guess a URI list in random order doesn't have too much common substring locality apart from "http://www.", ".com" and ".html"

[02:36:00.0000] <roc>
sigh

[02:37:00.0000] <hsivonen>
roc: ?

[02:37:01.0000] <Hixie>
BenMillard: gzip

[02:37:02.0000] <Hixie>
hsivonen: surprisingly many sites don't have www., .com, and .html, too

[02:37:03.0000] <roc>
making the intrinsic size of an <iframe> depend on the layout of its container isn't really a good i dea

[02:37:04.0000] <Hixie>
hsivonen: sent e-mail

[02:38:00.0000] <Hixie>
um

[02:39:00.0000] <Hixie>
not sent

[02:39:01.0000] <Hixie>
my SMTP server said no to a 20MB attachment

[02:41:00.0000] <hsivonen>
Hixie: email now received. thanks

[02:53:00.0000] <gDashiva>
Matt's reply to James strikes me as rather lacking

[02:54:00.0000] <gDashiva>
Making a painting is a lot more work than adding <img> to a HTML page, yet apparently making alternate text for a painting is infeasible

[02:54:01.0000] <gDashiva>
And uploading a video to youtube is limited by physical space?

[02:56:00.0000] <gDashiva>
And the synthesia example talks explicitly about a webpage

[02:56:01.0000] <Philip`>
Presenting alternate text for a painting is infeasible (you don't want people rubbing their hands all over your priceless paintings trying to find a braille description)

[02:57:00.0000] <Philip`>
but for <img> it's trivial since you just add an attribute and the user has tools that let them read it

[02:57:01.0000] <Hixie>
i love how everyone and their brother has now pointed out that there are references to RFC3066 in HTML5

[02:59:00.0000] <gDashiva>
Philip`: Imagine if the alt text was embedded in the original painting, then you wouldn't need to write alt text for the <img>

[02:59:01.0000] <gDashiva>
They could just use an automated extract-alt-text-from-image AT program

[03:02:00.0000] <Philip`>
gDashiva: I can imagine such a situation but I'm not sure what relevance that has

[03:03:00.0000] <Hixie>
i wonder if al plans to reply to http://lists.w3.org/Archives/Public/wai-xtech/2008Aug/0132.html

[03:04:00.0000] <hsivonen>
Hixie: there's a chance that your assuption that alt text should make sense when flattened as part of surrounding prose isn't the right assumption

[03:05:00.0000] <hsivonen>
Hixie: and that users would be better off getting cued that a piece of text is an alt digression from the surrounding prose and doesn't need to flow nicely into the paragraph

[03:05:01.0000] <Hixie>
there's always a chance that i'm wrong on many things :-)

[03:06:00.0000] <Hixie>
well i don't know about AT users, but at least from the point of view of a Lynx/Links user (which I am regularly), I certainly wouldn't want to be told whenever there's an image, i just want the image replaced with the text

[03:07:00.0000] <gDashiva>
Philip`: one painter writing alt, or X web page authors doing the same work over and over :)

[03:07:01.0000] <gDashiva>
But yes, it's not that relevant in today's digital age

[03:09:00.0000] <Hixie>
ok bed time

[03:09:01.0000] <Hixie>
nn

[03:10:00.0000] <gDashiva>
m

[03:10:01.0000] <Philip`>
gDashiva: Maybe the painter could embed the alt text as a kind of watermark in their painting, so even if someone takes a photo of it and uploads it to their web page then the text will still be extractable and the image will still be accessible

[03:11:00.0000] <gDashiva>
Philip`: Kinda like all that pirate-tracing tech being proposed for DRMed media?

[03:14:00.0000] <zcorpan>
Hixie: "well i don't know about AT users, but at least from the point of view of a Lynx/Links user (which I am regularly), I certainly wouldn't want to be told whenever there's an image, i just want the image replaced with the text" -- that'd work fine when reading your blog but not so fine for web pages in general since alt is so often misused

[03:23:00.0000] <BenMillard>
zcorpan, here are the interesting uses of alt I collected: http://projectcerbera.com/web/study/2008/collection#alt

[03:24:00.0000] <BenMillard>
(it's not a study of how many are good versus how many are bad; it focusses on the ones I found interesting or considered wrong)

[03:24:01.0000] <BenMillard>
duplication of nearby text and filenames seem like a common mistake, although I haven't crunched the numbers yet

[03:25:00.0000] <BenMillard>
sometimes it looks like the alt was written for one version of the image, then the image changed and the alt wasn't updated

[03:29:00.0000] <BenMillard>
missing the alt in places it was needed happened in quite a few cases, too

[04:25:00.0000] <zcorpan>
hsivonen: <meta charset=asdfasdf> validates

[04:27:00.0000] <anne>
I heard yesterday namespaces could've been worse. Philippe Le Hégaret detailed a huge device at the W3C (took eight months to resolve and many many hours) as to whether a namespace should be a URI reference or a string.

[04:54:00.0000] <webben>
Philip`: note that some image formats allow you to embed text metadata : http://www.mirrorservice.org/sites/www.libpng.org/pub/png/spec/1.2/PNG-Chunks.html#C.tEXt for example

[04:55:00.0000] <hsivonen>
zcorpan: asdfasdf meets the syntactic requirements for a charset name

[04:55:01.0000] <hsivonen>
zcorpan: should I check the value against a list of actual names?

[04:55:02.0000] <hsivonen>
zcorpan: if so, which list?

[04:56:00.0000] <hsivonen>
I don't like the theoretical open-endedness of charset

[04:56:01.0000] <hsivonen>
I want UTF-8 and a closed list of grandfathered legacy names

[04:57:00.0000] <hsivonen>
anne: do you meant like doing relative URI processing on NS URIs?

[04:58:00.0000] <hsivonen>
s/meant/mean/

[05:01:00.0000] <zcorpan>
hsivonen: in xml you warn that it doesn't match external encoding information

[05:02:00.0000] <zcorpan>
hsivonen: in text/html it's an all clear

[05:02:01.0000] <anne>
hsivonen, yes

[05:02:02.0000] <anne>
s/device/devide/

[05:02:03.0000] <anne>
too

[05:03:00.0000] <zcorpan>
hsivonen: you could use the list that your validator supports

[05:03:01.0000] <hsivonen>
zcorpan: hmm. somehow I thought I had code for checking that meta matched the encoding in use

[05:03:02.0000] <hsivonen>
weird

[05:04:00.0000] <zcorpan>
hsivonen: yeah i remember it whining about it before

[05:04:01.0000] <zcorpan>
hsivonen: i tested with textarea input

[05:06:00.0000] <hsivonen>
/me finds // XXX NOP

[05:12:00.0000] <hsivonen>
zcorpan: fixed. thanks

[05:29:00.0000] <hsivonen>
hmm. using real web content is almost like fuzzing

[05:33:00.0000] <wilhelm>
It is. Our stability testing tools use URLs from the wild or fuzzers interchangeably. (c;

[07:15:00.0000] <anne>
hsivonen, anonymous suggests maybe making a list of namespace benefits for browser vendors (not suggesting it's non-empty)

[07:17:00.0000] <hsivonen>
anne: namespace or Namespace?

[07:19:00.0000] <anne>
XML namespace, if that helps

[07:19:01.0000] <anne>
namespaces*, meh

[07:21:00.0000] <anne>
should we update Wikipedia to mention Web Workers?

[07:23:00.0000] <anne>
I guess it will happen eventualy

[07:23:01.0000] <anne>
lly

[07:23:02.0000] <hsivonen>
I can't think of any benefits from Namespaces to browser vendors where the benefit couldn't have been obtained by a mechanism that doesn't use the kind of indirection Namespaces use

[07:25:00.0000] <hsivonen>
CSS extensions, DOM extensions and MathML all involve putting a short prefix into the opaque name

[07:26:00.0000] <anne>
same here

[07:30:00.0000] <hsivonen>
I can't think of cases where a browser vendor is itself using Namespaces except XUL

[07:30:01.0000] <hsivonen>
though the benefit could be that if others use Namespaces, they keep out of the browser-sensitive name turf

[07:30:02.0000] <anne>
even there you could just have moz-button and such

[07:30:03.0000] <hsivonen>
right

[07:31:00.0000] <hsivonen>
oh, and then there's VML

[07:32:00.0000] <hsivonen>
which for practical purposes uses the short prefix approach

[07:37:00.0000] <hsivonen>
the other day, I was pondering if HTML5 could make createElementNS fade into obscurity by making createElement recognize MathML and SVG names and the SVG/HTML name-colliding elements implementing the interfaces for both HTML and SVG

[07:38:00.0000] <hsivonen>
after all, SVG rendering is already sensitive to the tree position of the element

[07:39:00.0000] <anne>
maybe

[07:41:00.0000] <anne>
grmbl, I e-mailed ietf-types, but http://www.alvestrand.no/pipermail/ietf-types/ doesn't reflect that

[10:34:00.0000] <Philip`>
/me can't remember whether he's already commented that his objection to the alt="{...}" proposal (in that it makes it hard to write reliably-correct HTML-generating tools) could be fixed by permitting it in the alt-is-actually-a-proper-equivalent cases as well as the alt-is-just-a-description cases

[10:34:01.0000] <Philip`>
which would mean conforming pages using alt="{...}" would be ambiguous, but that's okay because it's going to be ambiguous in reality anyway and this won't make it any worse

[10:35:00.0000] <Philip`>
So, anyway, if I hadn't commented on that, then at least I have now

[10:58:00.0000] <krijnh>
/me adds a new subtitle to his logs

[10:59:00.0000] <krijnh>
jgraham: thanks for the inspiration :)

[11:03:00.0000] <Lachy>
krijnh, which one was added?

[11:04:00.0000] <anne__>
/me hits F5

[11:04:01.0000] <krijnh>
Something about an eval overlord

[11:04:02.0000] <krijnh>
*evil

[11:04:03.0000] <anne__>
"Because, Seriously, Dmitry Can't Do It All By Himself" hehe

[11:04:04.0000] <Lachy>
that won't work too well for me, since I don't konw most of the existing ones

[11:04:05.0000] <jgraham>
Lachy: http://krijnhoetmer.nl/irc-logs/?gimme-a-hobby

[11:05:00.0000] <anne__>
/me finds "Where The Evil HTML 5 Overlord And His Band Of Minions Chat, In Public"

[11:05:01.0000] <anne__>
/me lols

[11:06:00.0000] <jgraham>
anne__: In the middle of a WG meeting?

[11:06:01.0000] <anne__>
jgraham, at the end actually, maybe you can join for dinner

[11:07:00.0000] <anne__>
jgraham, you can

[11:07:01.0000] <anne__>
/me figures out where we are going

[11:07:02.0000] <anne__>
Philip`, you can too, but it's Indian and I heard you might not enjoy that

[11:08:00.0000] <jgraham>
anne__: OK

[11:08:01.0000] <jgraham>
cool

[11:08:02.0000] <anne__>
jgraham, http://www.pipasha-restaurant.co.uk/?id=contact

[11:08:03.0000] <anne__>
we have a reservation at 6:45

[11:08:04.0000] <anne__>
Philip`, there is place for you fwiw

[11:09:00.0000] <jgraham>
Interesting. How are you getting there

[11:09:01.0000] <anne__>
(also, it's a subset of the WG)

[11:09:02.0000] <jgraham>
/me can't walk that far by 6:45

[11:10:00.0000] <jgraham>
You are just across the road, right?

[11:10:01.0000] <anne__>
http://maps.google.com/maps?f=q&hl=en&geocode=&q=cambridge+pipasha&ie=UTF8&ll=52.213353,0.154431&spn=0.004944,0.009871&t=h&z=17

[11:11:00.0000] <anne__>
we are at Microsoft still

[11:11:01.0000] <anne__>
and going by car

[11:11:02.0000] <anne__>
if you come a little later that's ok

[11:12:00.0000] <anne__>
or grab a cab, not sure if that's expensive or not though

[11:12:01.0000] <jgraham>
I guess I might just stay here and do some work since it's so far away and I've got so little done

[11:13:00.0000] <anne__>
is it really that far?

[11:13:01.0000] <anne__>
ok

[11:13:02.0000] <jgraham>
anne__: I guess 45-60 minutes walk

[11:14:00.0000] <anne__>
arriving at 7, 7:15 is fine, but decide for yourself

[11:14:01.0000] <anne__>
/me has to pack his laptop now

[11:36:00.0000] <Lachy>
/me wrote a script to extract all subtitles from krijnh's logs

[11:36:01.0000] <Lachy>
well, I hope I got them all. I relied on the probability of all of them being sent at least once within 100 separate requests for the page

[11:37:00.0000] <krijnh>
Hah

[11:37:01.0000] <krijnh>
Don't DOS my server please :)

[11:37:02.0000] <Lachy>
I didn't. I did: wget http://.../; sleep 1; in a loop

[11:37:03.0000] <krijnh>
How many did you find?

[11:37:04.0000] <Lachy>
<em>This</em> Line Is My Opinion, Not The Other(')s

[11:37:05.0000] <Lachy>
<em>What working group</em> is going to work on extending HTML...

[11:37:06.0000] <Lachy>
Actually, We Need A Solid Use Case For These Logs

[11:37:07.0000] <Lachy>
And <code>X-UA-Compatible</code> Is A Great Idea!

[11:37:08.0000] <Lachy>
And Indeed, <code>#xhtml</code> Wasn't Kick Ass Enough

[11:37:09.0000] <Lachy>
And RSS Is A Success As Well

[11:38:00.0000] <Lachy>
And You Can <code>/join</code> Too

[11:38:01.0000] <Lachy>
Because A Mailinglist Alone Isn't Enough

[11:38:02.0000] <Lachy>
Because All Those Web 2.0 Widgets Should Be Accessible

[11:38:03.0000] <Lachy>
Because XHTML Is The Future!

[11:38:04.0000] <Lachy>
Because, Seriously, Dmitry Can't Do It All By Himself

[11:38:05.0000] <Lachy>
Could You Elaborate?

[11:38:06.0000] <Lachy>
Great Minds Think Alike. And Sick Minds, Too!

[11:38:07.0000] <krijnh>
Damn you! Now I have to find new ones

[11:38:08.0000] <Lachy>
In Search Of A Better Title

[11:38:09.0000] <Lachy>
Including Offtopic WHATWG Cabal Ramblings

[11:38:10.0000] <Lachy>
Issue Tracking Tools Will Save Us!

[11:38:11.0000] <Lachy>
OMG! We Can't Break The Web!

[11:38:12.0000] <Lachy>
Opinions Expressed Here Are Those ... Bla Bla Bla

[11:38:13.0000] <Lachy>
Public-html: A Mailinglist Apart

[11:38:14.0000] <Lachy>
So, When Can We Use <code>application/xhtml+xml</code> Again?

[11:38:15.0000] <Lachy>
Somebody Needs To Do To The Semantic Web What The WHATWG Did To HTML

[11:38:16.0000] <Lachy>
Sorry, But Who Asked For This?

[11:38:17.0000] <Lachy>
We Cannot Definitively Say Why XHTML Has Not Been Successful On The Web

[11:38:18.0000] <Lachy>
Web 2.0, Evolved (Still In Beta Though)

[11:38:19.0000] <Lachy>
What Else Do I Need To Do To Get Hired By Opera? Or Google?

[11:38:20.0000] <Lachy>
Where The Evil HTML 5 Overlord And His Band Of Minions Chat, In Public

[11:38:21.0000] <Lachy>
While We All Agree Flash Is Still A Superior Technology ;)

[11:38:22.0000] <Lachy>
Yeah, And Standards Suck, Right

[11:38:23.0000] <Lachy>
Your Ad Here? Let's See. Nope, It Isn't.

[11:38:24.0000] <Lachy>
htmllogs4all!

[11:38:25.0000] <Lachy>
you could always erase them from your log, and it'll be as if I never posted them :-)

[11:38:26.0000] <krijnh>
:)

[11:39:00.0000] <krijnh>
So now it's pretty obvious: sometimes I just don't have anything to do

[11:42:00.0000] <krijnh>
Same thing for you btw

[11:50:00.0000] <Lachy>
woah, I did miss one on the first attempt. there was 30 in that list, this time I got 31

[11:50:01.0000] <Lachy>
"And SVG Needs Some Love As Well" was missed

[11:51:00.0000] <krijnh>
Oh dear

[11:51:01.0000] <Lachy>
btw, if anyone else wants to participate in a DDoS against krijnh, here's the script :-)

[11:51:02.0000] <Lachy>
for ((i=0;$i<100;i=$i+1));

[11:51:03.0000] <Lachy>
 do wget -O $i.html http://krijnhoetmer.nl/irc-logs/?gimme-a-hobby;

[11:51:04.0000] <Lachy>
 sleep 1;

[11:51:05.0000] <Lachy>
done

[11:51:06.0000] <Lachy>
cat * | grep ^\<p | sort | uniq > all.txt

[11:51:07.0000] <Lachy>
cat all.txt | wc -l

[11:51:08.0000] <krijnh>
Thanks for ruining my fun ;)

[11:52:00.0000] <gDashiva>
See

[11:52:01.0000] <gDashiva>
You need to add a captcha or something

[11:52:02.0000] <krijnh>
I seriously hope nobody is interested in this as much as you.. And not because of the DOS stuff ;)

[11:53:00.0000] <Lachy>
I was just bored, and wanted to practice my skills with writing bash scripts, which I started learning recently

[11:53:01.0000] <jcranmer>
Lachy: the bash script I wrote to help teach myself is ugly

[11:53:02.0000] <jcranmer>
but I have proven that it is possible to write a BFS in bash

[11:54:00.0000] <gDashiva>
It's also possible to bang your head against the wall

[11:54:01.0000] <Lachy>
BFS? first result was http://en.wikipedia.org/wiki/Breadth-first_search

[11:54:02.0000] <Lachy>
is that right?

[11:54:03.0000] <jcranmer>
yes

[11:54:04.0000] <krijnh>
gDashiva: Good point

[11:55:00.0000] <Lachy>
the bug with my script is that it doesn't stript <p><b> tags from the lines. Is that possible to do with grep, or some other utility?

[11:55:01.0000] <krijnh>
sed ?

[11:56:00.0000] <Lachy>
ok, I'll have to learn sed later

[11:56:01.0000] <krijnh>
No idea if it's possible, I don't use it that much

[11:56:02.0000] <krijnh>
Anyway, done reading up on HTML WG mail

[11:56:03.0000] <Lachy>
anything interesting worth reading?

[11:57:00.0000] <krijnh>
Anybody in here going to Fronteers 2008, to see anne speak? :)

[11:57:01.0000] <Lachy>
no

[11:57:02.0000] <krijnh>
Tsk

[11:57:03.0000] <Lachy>
when and where is that?

[11:57:04.0000] <krijnh>
http://fronteers.nl/congres/2008/speakers

[11:57:05.0000] <Lachy>
anne isn't listed as a speaker on that page

[11:58:00.0000] <krijnh>
There are 3 speakers to be annoucned

[11:58:01.0000] <krijnh>
*announced

[11:58:02.0000] <krijnh>
Anne is one of those 3

[11:58:03.0000] <Lachy>
I'm going back to Australia on September 20 instead, and I have my own presentation to give on the Oct 7

[12:05:00.0000] <krijnh>
But Dean Edwards is coming!

[12:25:00.0000] <gsnedders>
/me notes he doesn't have 200 EUR

[12:26:00.0000] <jgraham>
/me notes that stripping stuff from lines ia a major use case for sed

[12:30:00.0000] <jgraham>
sed -e "s|<p><b>\(.*\)<\/b><\/p>|\1|"

[12:30:01.0000] <jgraham>
for example

[12:30:02.0000] <jgraham>
obviously Philip` will have an easier way :)

[12:31:00.0000] <Philip`>
perl -pe's~<p><b>(.*)</b></p>~$1~'

[12:32:00.0000] <jgraham>
krijnh: I don't know how the subtitle gets chosen but you should put in a few with a low probability like 1 in 1000

[12:32:01.0000] <krijnh>
Why exactly? )

[12:33:00.0000] <jgraham>
To annoy Lachy 

[12:33:01.0000] <jgraham>
:)

[12:33:02.0000] <krijnh>
Good reason :)

[12:33:03.0000] <Philip`>
anne__: Indeed, I'm not much of a fan of any non-trivial food :-(

[12:34:00.0000] <jgraham>
/me notes that he has empirical evidence that Philip` eats more types of food than the guy in his department who eats only processed chicken and white bread

[12:35:00.0000] <Philip`>
I eat unprocessed chicken too :-)

[12:40:00.0000] <gsnedders>
Not spinach though.

[12:43:00.0000] <gsnedders>
(or at least I think that's what Philip` didn't eat on his pizza back when myself and jgraham went out to supper with him in May)

[12:45:00.0000] <gsnedders>
krijnh: I have to admit I think one or two are a bit mean to some people, even if they are funny :)

[12:46:00.0000] <gsnedders>
Oh, BTW, I did decide on the project that I came up with based on jgraham's idea.

[13:03:00.0000] <krijnh>
gsnedders: which ones?

[13:03:01.0000] <gsnedders>
krijnh: Oh like, Dimitry

[13:03:02.0000] <gsnedders>
though that is funny.

[13:06:00.0000] <krijnh>
Hope so, not meant to be offensive or something :)

[13:12:00.0000] <Hixie>
hsivonen: just make the list of encodings you consider valid be the list of encodings you support

[13:15:00.0000] <Hixie>
Kristof sends weird e-mails

[13:15:01.0000] <Hixie>
i don't really know what to ask him to change

[13:16:00.0000] <Hixie>
but like his latest e-mail about GNU libmagic... I mean, I'm sure some people don't know about file(1) yet, but so what?

[13:16:01.0000] <Lachy>
jgraham, thanks for the help with sed

[13:17:00.0000] <Hixie>
karl sent a similar e-mail (about new york)

[13:42:00.0000] <Hixie>
i asked al if he was going to reply to my last e-mail on the alt thread

[13:42:01.0000] <Hixie>
he replied that i was asking the wrong person

[13:42:02.0000] <Hixie>
i'll take that as a no then...

[13:46:00.0000] <krijnh>
(Meanwhile, the bomb keeps ticking)

[15:33:00.0000] <Hixie>
"why would Flickr decide to violate [the requirement that alt not be bogus] rather than the one that says 'authors must provide suitable alt text'"

[15:33:01.0000] <Hixie>
seriously?

[15:34:00.0000] <Hixie>
we're still at that point in the debate?

[15:34:01.0000] <Hixie>
jesus

[15:34:02.0000] <hober>
sigh.

[15:35:00.0000] <Dashiva>
On one hand they want people to be conformant. On the other hand they expect people to ignore conformance for arbitrary reasons

[15:36:00.0000] <roc>
"why would Flickr decide to violate the alt requirement rather than the one that says Flickr should stay in business"

[15:37:00.0000] <roc>
although of course Flickr probably would stay in business, users would quickly find the shortest alt string that passes filters

[15:37:01.0000] <Dashiva>
And the AT could just hardcode that to mean 'no alt'. "Brillant"

[15:38:00.0000] <Hixie>
hey what are some photo sites like picasa and flickr

[15:38:01.0000] <roc>
photobucket

[15:38:02.0000] <Dashiva>
my opera has galleries

[15:38:03.0000] <hober>
photo.net

[15:38:04.0000] <Hixie>
i need one more with a single word name and then google sets will give me the rest

[15:38:05.0000] <Dashiva>
all the photobucket-like sites, imageshack etc

[15:39:00.0000] <Hixie>
cool thanks

[15:39:01.0000] <Hixie>
http://labs.google.com/sets?hl=en&q1=flickr&q2=picasa&q3=photobucket&q4=imageshack&q5=&btn=Large+Set

[15:39:02.0000] <roc>
very cool

[15:49:00.0000] <Hixie>
holy crap there are lot of these

[15:50:00.0000] <Dashiva>
And then you have image boards

[16:42:00.0000] <Lachy>
OMG, Stargate Atlantis is ending at the end of season 5 and move on to straight-to-DVD movies

[16:42:01.0000] <Hixie>
wow, i hadn't realised how much research we actually had here for this alt="" thing

[16:42:02.0000] <Lachy>
and Stargate Universe has been confirmed for next year

[16:42:03.0000] <Hixie>
sweet

[16:42:04.0000] <Lachy>
http://gateworld.net/news/2008/08/wright_iatlantisi_is_going_out_o.shtml

[16:43:00.0000] <Hixie>
i can't go to gateworld.net

[16:43:01.0000] <Lachy>
SPOILERS: http://gateworld.net/news/2008/08/istargate_universei_has_a_go.shtml

[16:43:02.0000] <Lachy>
oh

[16:43:03.0000] <Hixie>
i get spoilt every single time

[16:43:04.0000] <Lachy>
the first link doesn't contain any spoilers about the plot, it just talks about why they're ending the series now and changing to movies

[16:44:00.0000] <Xenos>
Half the heroes die

[16:45:00.0000] <Lachy>
liar!

[16:45:01.0000] <Hixie>
Lachy: i bet there's something in some image in a sidebar somewhere that's a spoiler for something (e.g. future episode titles, guest star for next week, etc)

[16:48:00.0000] <Lachy>
it also says there's also going to be a new computer game called Stargate Worlds

[16:49:00.0000] <Lachy>
yeah, it has the name of a future episode in the sidebar

[16:49:01.0000] <Hixie>
see :-)

[16:49:02.0000] <Lachy>
it's tonghts episode title

[16:49:03.0000] <Hixie>
oh hey today is friday!

[16:49:04.0000] <Hixie>
woot!

[16:51:00.0000] <Lachy>
yep, I just waiting for it to air, so I can download the episode in about 4 hours

[16:52:00.0000] <Hixie>
i'll have to see it tomorrow, sadly, iTunes doesn't let me download it until about 3am

[16:52:01.0000] <Hixie>
very annoying

[16:52:02.0000] <Lachy>
iTunes doesn't have stargate available in either Australia or Norway, and they're the only stores I can buy from

[16:52:03.0000] <Lachy>
so I have to use usenet

[16:53:00.0000] <Lachy>
but I get to see it in 1080p

[16:53:01.0000] <Lachy>
I found that itunes quality is a bit too low

[16:53:02.0000] <Hixie>
yeah

[16:53:03.0000] <Hixie>
well sadly my mac mini couldn't handle HD anyway :-)

[16:54:00.0000] <Lachy>
don't you have a faster machine to watch it on?

[16:55:00.0000] <Lachy>
btw, do you strip the DRM from your itunes purchases so you can watch them wherever you like?

[16:55:01.0000] <Hixie>
i could watch it on the laptop, but connecting the laptop to the cinema display and the audioengine speakers is a bitch, so i don't bother

[16:55:02.0000] <Hixie>
i haven't found a way to losslessly strip the DRM sadly

[16:55:03.0000] <Lachy>
Requiem 1.7.3 works

[16:56:00.0000] <Lachy>
do you want a copy?

[16:57:00.0000] <Lachy>
whoever wrote it actually reverse engineered the way iTunes decrypts them, so it's completely lossless

[16:57:01.0000] <Hixie>
oooh, never heard of it! sure!

[16:57:02.0000] <Hixie>
that would be awesome

[16:57:03.0000] <Hixie>
ian⊙hc


2008-08-23
[17:02:00.0000] <Lachy>
if you need any help with it, read this forum thread http://hymn-project.org/forums/viewtopic.php?t=2494 and the FAQ http://hymn-project.org/forums/viewtopic.php?t=2615

[22:31:00.0000] <Hixie>
heh maybe we should adopt role="" for <img> just to make it harder to adopt over the role spec should anyone propose we do that later :-P

[05:45:00.0000] <hsivonen>
hmm. I think either my parser or Jing has a memory leak...

[09:10:00.0000] <Dashiva>
I see nothing has changed on public-html today either

[09:10:01.0000] <gsnedders>
Change is over-rated.

[09:11:00.0000] <hsivonen>
I hope Change Congress doesn't depend on RDF

[11:55:00.0000] <gsnedders>
jgraham: Nice email :)

[11:55:01.0000] <gsnedders>
jgraham: (your latest on public-html)

[11:57:00.0000] <Philip`>
Nicer if it didn't end with a paragraph saying "I" and look like it was sent prematurely :-)

[11:57:01.0000] <gsnedders>
That's true :)

[11:58:00.0000] <jgraham>
Philip`: I just noticed that. I wonder what I was going to write there

[12:00:00.0000] <jgraham>
I wasn't sure if I had made my point clearly so I kept it as a draft for a while. Then I realised that I may as well just send it

[12:22:00.0000] <webben_>
/me is still keen on seeing a clear explanation of what the goal of HTML5 conformance is beyond weakly protecting future extension points. (The main purpose of HTML 4.01 conformance seems to have been a common core of interoperability.)

[12:25:00.0000] <Philip`>
webben_: Do you mean UA conformance, or author conformance, or both?

[12:25:01.0000] <webben_>
author conformance

[12:28:00.0000] <jgraham>
webben_: I think the goal is a) increasing interoperability (also in the face of future changes) b) discouraging things that are known to have bad cost/benefit 

[12:29:00.0000] <jgraham>
But I admit I find it hard to pin down in simple terms

[12:29:01.0000] <webben_>
jgraham: a) is the same as protecting future extension points, or more than that?

[12:29:02.0000] <jgraham>
webben_: a) is also about working with existing UAs

[12:30:00.0000] <webben_>
hmm

[12:30:01.0000] <webben_>
I see.

[12:31:00.0000] <webben_>
so preferring features that work with Internet Explorer over features that don't?

[12:32:00.0000] <webben_>
roughly speaking

[12:32:01.0000] <webben_>
*equivalents that don't

[12:33:00.0000] <webben_>
Why do things that have bad cost/benefit ratios need to be discouraged by conformance rules?

[12:33:01.0000] <webben_>
assume it's not an interoperability issue

[12:33:02.0000] <webben_>
*assuming

[12:34:00.0000] <Philip`>
Perhaps so that validators will alert authors when they make a mistake and do those bad cost/benefit things

[12:34:01.0000] <jgraham>
(for an example of b, discouraging presentational markup seems to be one of the goals of HTML5 conformance yet, for an author aware of the issues presentational markup need not have an adverse effect. It's just that sometimes it's easier to say "don't play with guns" than to stand by and watch people blow their own feet off)

[12:35:00.0000] <webben_>
so the cost/benefit ratio changes from one authoring situation to another?

[12:35:01.0000] <webben_>
but the conformance should be geared to help newbie authors?

[12:37:00.0000] <jgraham>
webben_: Not necessarily just things that work in IE. Consider an attribute that takes a date. It's good for consistency and readability if everyone adopts the same syntax without relying  on the funky parsing that browsers have to do; allowing any random string that could be converted into a date would have no benefits

[12:37:01.0000] <jgraham>
So sometimes it's about reducing choice

[12:38:00.0000] <jgraham>
(which is usually a good thing as long as you're not taking away functionality. indeed it often is if you are)

[12:38:01.0000] <webben_>
jgraham: Sounds very much like the arguments for requiring attributes to be quoted.

[12:39:00.0000] <jgraham>
webben_: Sure. I don't think those arguments are unreasonable. OTOH, I don't think they're necessarily strong enough to warrant making the change 

[12:40:00.0000] <jgraham>
I guess if there was a lot of evidence of authors doing the wrong thing because they weren't quoting attributes, it would be a good change

[12:40:01.0000] <webben_>
jgraham: you mean because there's a counterbalancing benefit (slightly fewer bytes/less typing)?

[12:40:02.0000] <jgraham>
webben_: Yeas

[12:41:00.0000] <jgraham>
s/Yeas/Yeah/ :)

[12:41:01.0000] <Philip`>
jgraham: There is evidence - see <meta charset>, and the very common garbage attributes that come from unquoted keyword/description strings

[12:42:00.0000] <webben_>
isn't there a danger in making validators throw errors over mere consistency and readability?

[12:42:01.0000] <webben_>
(in that it becomes, perhaps, harder to migrate old content and maintain new conforming content, given common authoring practices atm)

[12:43:00.0000] <webben_>
to put it another way, is it a problem to lump in "stuff that arguably makes HTML maintenance harder" with stuff that harms end-users?

[12:43:01.0000] <jgraham>
Who gains if the date 2008-08-23 can also be conformingly written aa2008--08:23--?

[12:44:00.0000] <webben_>
in the same category of error.

[12:44:01.0000] <jgraham>
(I have no idea if that would parse in HTML5 btw; it was random :) )

[12:44:02.0000] <gsnedders>
jgraham: No, it wouldn't.

[12:44:03.0000] <webben_>
jgraham: If browsers /can/ parse that, anyone whose code already relies on it.

[12:45:00.0000] <jgraham>
webben_: I think you are underestimating the value of consistency and readability

[12:46:00.0000] <jgraham>
Also some things have knock on consequences

[12:46:01.0000] <webben_>
Just to be clear, I like consistency and readability and prefer to follow coding conventions in my own work.

[12:46:02.0000] <webben_>
I'm just trying to tease out the aims of conformance and validity here.

[12:46:03.0000] <webben_>
and whether they really make sense.

[12:47:00.0000] <webben_>
both in terms of end-goals and in terms of author psychology/pressures.

[12:48:00.0000] <jgraham>
e.g. people who have systems that depend on presentational markup are probably making their users' experience worse (usually slower to download, no caching, harder to adapt the style to different media, harder to restyle on the client side, harder to extract information from programmatically) 

[12:48:01.0000] <jgraham>
So making those systems non conforming can be a win

[12:48:02.0000] <jgraham>
even thoough in the short term the author losees out

[12:57:00.0000] <Dashiva>
Related question: Is conformance a carrot or a stick?

[12:59:00.0000] <gsnedders>
Dashiva: A pineapple.

[13:00:00.0000] <Dashiva>
I'm not familiar with that metaphor

[13:00:01.0000] <gsnedders>
Nor am I.

[13:01:00.0000] <Dashiva>
/me fills gsnedders' tub with glow-in-the-dark hubcaps

[13:01:01.0000] <jgraham>
Dashiva: It's an organic carrot, an non conforming web page is a non-organic carrot, and there is no stick.

[13:01:02.0000] <gsnedders>
Ooooo! Glow-in-the-dark hubcaps! Awesome!

[13:03:00.0000] <jgraham>
(by organic, I mean grown to organic food standards without pesticides and so on, not "made from organic molecules")

[13:04:00.0000] <Dashiva>
jgraham: Sure, okay. But I was talking about the concept of conformance, not a specific page. :)

[13:06:00.0000] <jgraham>
Dashiva: My point was that the concept of conformance is similar to the difference between an organic and a non-organic carrot

[13:06:01.0000] <jgraham>
Functionally either will work as a carrot

[13:06:02.0000] <Dashiva>
Yeah, I got that

[13:07:00.0000] <Dashiva>
But I'm asking if conformance is something we offer, or something we deny.

[13:35:00.0000] <jgraham>
Dashiva: In practice it is only something we offer

[13:36:00.0000] <Dashiva>
jgraham: But we'd rather it be something to deny?

[13:44:00.0000] <jgraham>
Dashiva: I don't think so. But I'm not quite sure what "deny" means. I understood carrot and stick.

[13:47:00.0000] <jgraham>
(obviously I know what "deny" means. but I don't understand what the difference between allow and deny in this context is. I was assuming allow == carrot deny == stick)

[13:48:00.0000] <Dashiva>
"Congratulations, you get to call yourself conforming." vs "You have failed us, so we will label you as a failure."

[13:49:00.0000] <gsnedders>
EPIC FAIL.

[13:52:00.0000] <jgraham>
Dashiva: I think the answer is "none of the above"

[13:53:00.0000] <jgraham>
But offer is much closer because you have to choose to test yourself for conformance

[13:54:00.0000] <jgraham>
We don't come round, check conformance and go "sorry you don't conform we're shutting you down"

[13:54:01.0000] <Dashiva>
Except if you're Target

[13:55:00.0000] <gsnedders>
/me shoots an arrow

[13:55:01.0000] <jgraham>
Dashiva: That was failing to comply with a11y legislation, not faliure to comply with HTML 

[13:55:02.0000] <jgraham>
HTML could have said nothing about alt and they still could have been sued if blind users couldn't use their site

[13:56:00.0000] <jgraham>
(I guess XML does go "you don't conform we're shutting you down" but that's a huge mistake)

[13:57:00.0000] <Dashiva>
heh

[13:59:00.0000] <gsnedders>
/me shuts jgraham down for not conforming to Man 1.0, requirement 3B: MUST have short hair.

[14:04:00.0000] <benoitc>
hi

[14:06:00.0000] <jgraham>
benoitc: Hi

[14:06:01.0000] <webben_>
jgraham: "people who have systems that depend on presentational markup are probably making their users' experience worse (usually slower to download, no caching, harder to adapt the style to different media, harder to restyle on the client side, harder to extract information from programmatically)" ... absolutely, but those are all side issues to readability/consistency.

[14:07:00.0000] <krijnh>
jgraham: wouldn't your points remain the same for http://juicystudio.com/wcag/tables/complexdatatable.html ?

[14:07:01.0000] <jgraham>
krijnh: I hope so, otherwise I'm going to have it pointed out to me in excrutiating detail just how idiotic I am :)

[14:07:02.0000] <krijnh>
(I would also markup those <td class="header"> as <th>'s)

[14:09:00.0000] <krijnh>
http://lists.w3.org/Archives/Public/public-html/2008Aug/0633.html

[14:09:01.0000] <jgraham>
webben_: I agree. I guess those are examples of conformance requirements designed to make end user's lives better by increasing the probability the the author does the thing that works best for the user

[14:09:02.0000] <jgraham>
krijnh: I replied

[14:10:00.0000] <krijnh>
That example has way too complex markup, for such a simple table, imho

[14:11:00.0000] <jgraham>
krijnh: Seriously

[14:12:00.0000] <krijnh>
Yeah :)

[14:12:01.0000] <jgraham>
:)

[14:13:00.0000] <krijnh>
<td id="col8-2a">Budgeted</td> is interesting though

[14:13:01.0000] <krijnh>
Should <th scope="row">Budgeted</th> work on that?

[14:17:00.0000] <benoitc>
mmm what is the current status of web forms 2 ? I just saw that dev of http://code.google.com/p/webforms2/ has been suspended

[14:17:01.0000] <jgraham>
krijnh: See http://tinyurl.com/624mko for my take on how it should work. That has just <th>Budgeted</th> (no @scope)

[14:17:02.0000] <benoitc>
i would like to make my website in html5 if possible so I don'tt know if it's better to use current forms standard or trying web forms 2 ?

[14:18:00.0000] <webben_>
benoitc: well, WF2 is designed to degrade gracefully.

[14:18:01.0000] <webben_>
(not that it necessarily achieves that 100%)

[14:18:02.0000] <jgraham>
benoitc: WF2 will be merged into HTML5 soon (the forms task force kind of died)

[14:18:03.0000] <jgraham>
Only Opera has any support atm though

[14:18:04.0000] <gsnedders>
It never came into life.

[14:19:00.0000] <Lachy>
LOL, Google translate is broken. I translated a page with a price list on it, and it chagned "NOK" to "U.S. $", but left the value unchanged.

[14:19:01.0000] <webben_>
benoitc: If you're going to try and conform to the changing requirements of HTML5, you'd want to use web forms 2.

[14:19:02.0000] <krijnh>
jgraham: cool!

[14:20:00.0000] <benoitc>
webben_: ok thanks

[14:20:01.0000] <benoitc>
this lib is the  only one to do it or ther is another ? I found one on sf 

[14:21:00.0000] <webben_>
benoitc: Using a library isn't the same as using web forms 2.

[14:21:01.0000] <webben_>
benoitc: you can use web forms 2 without using a library

[14:21:02.0000] <webben_>
like I said, it's designed to degrade gracefully

[14:22:00.0000] <krijnh>
jgraham: can your tool also show the inspected html source?

[14:22:01.0000] <webben_>
benoitc: all a library would do is try to provide the same functionality as a wf2 implementation when js is available.

[14:22:02.0000] <Philip`>
Lachy: I'd guess that's a danger of Google's apparent statistical model of translation, since it recognises when languages use different words for the same concept but fails to understand the more subtle differences in meanings, so it gets more false positives than a more human-based approach

[14:23:00.0000] <benoitc>
ok i see thanks

[14:23:01.0000] <benoitc>
will try to get further before coming with any question :)

[14:24:00.0000] <benoitc>
I would like to use any js lib to emulate behaviour off new features anyway

[14:27:00.0000] <jgraham>
krijnh: Not at the moment. 

[14:28:00.0000] <krijnh>
jgraham: would be handy, imho

[14:28:01.0000] <Philip`>
/me was just about to ask precisely the same question, having failed to notice krijnh ask it

[14:28:02.0000] <jgraham>
Well it should be trivial to add but on the other hand I should be spending less time at the moment doing this not more :)

[14:28:03.0000] <krijnh>
Would also easily show "that we're able to markup data tables accessibly"

[14:29:00.0000] <krijnh>
Where "we" includes more people than with http://juicystudio.com/wcag/tables/complexdatatable.html

[14:29:01.0000] <jgraham>
/me should note the algorithm in the tool is not the same as the one in the spec

[14:29:02.0000] <krijnh>
I don't care

[14:29:03.0000] <jgraham>
(the one in the spec doesn't allow headers to have headers)

[14:30:00.0000] <krijnh>
Just show that it's possible, with whatever markup

[14:37:00.0000] <Philip`>
/me wonders how to be sure that the table headers algorithm isn't just overfitted to the finite number of examples it was designed for

[14:40:00.0000] <Hixie>
i looked at a few tables, distilled the requirements, designed the algorithm, then checked the algorithms on the remainder of ben's data tables

[14:40:01.0000] <Hixie>
and it seemed to work ok


2008-08-24
[22:27:00.0000] <Hixie>
wow, david is arguing not only for alt= being required but for it being required to never be empty even for decorative images

[22:27:01.0000] <Hixie>
and he's blind, as far as i can tell

[22:27:02.0000] <Hixie>
so presumably would be aware of the crazyness of that position

[22:27:03.0000] <Hixie>
/me is baffled

[04:36:00.0000] <Hixie>
Philip`: i would be interested in hearing your opinion of http://lists.w3.org/Archives/Public/public-html/2008Aug/0254.html

[05:04:00.0000] <Hixie>
http://lists.w3.org/Archives/Member/w3c-html-cg/2008JulSep/0073.html sugests that the HCG call two weeks ago consisted almost entirely about talking about me

[05:05:00.0000] <Dashiva>
Too bad it's a treehouse-archive

[05:07:00.0000] <Philip`>
I think the described "gain for one segment of the population (those needing alt text)" comes from having any mechanism that gives a textual description of an image when no equivalent is available (and that requires authors to always write something, and hence to think about alt and (with non-zero probability) write something that's better than nothing);

[05:07:01.0000] <Philip`>
and this particular "small pain for another (tool writers)" is specific to the curly-brace syntax and doesn't seem to be a problem in any other proposed mechanism, because the other mechanisms don't add any complexity to the otherwise-pleasant situations where you really do have equivalent text

[05:07:02.0000] <Philip`>
I think the gain is worthwhile, so I wouldn't want to go back to alt being optional; but I think the pain is real, so I'd prefer some other solution that has the same gain without the pain

[05:07:03.0000] <Philip`>
but that involves tradeoffs against the pains (of varying levels of theoreticality) of other solutions, which is not trivial (if it's even possible) :-(

[05:09:00.0000] <Hixie>
k

[05:11:00.0000] <Philip`>
(I guess the desire to be able to distinguish equivalent text from descriptive text, so that <img alt="equivalent"> can be rendered the same as <span>equivalent</span>, seems like a mostly theoretical concern to me, since in reality nobody distinguishes alt that way, so a UA will have to cope with the indistinguishability anyway)

[05:12:00.0000] <Hixie>
that's basically the argument which is leading me back to alt="" being optional -- whatever we do the data won't be reliable, so UAs are going to have to do heuristics in all cases anyway, so why not just have the simplest syntax

[05:12:01.0000] <Hixie>
it's not like categories/roles like the spec suggests using now are actually going to be useful

[05:13:00.0000] <Hixie>
ok bed time

[05:13:01.0000] <Hixie>
nn

[05:13:02.0000] <Philip`>
That's far more of a change than just a simpler syntax, since it deprives the UA/user of even descriptive text about the image

[05:13:03.0000] <Hixie>
in practice i doubt that we'd be able to get descriptive text about the image

[05:14:00.0000] <Hixie>
it's not like flickr knows what the images are

[05:14:01.0000] <Hixie>
and in the case of a computer-generated fractal view, it's not like saying "it's a fractal" is gonna help anyone

[05:14:02.0000] <Hixie>
same with a webcam

[05:15:00.0000] <Hixie>
or a blind user's photos

[05:26:00.0000] <Philip`>
Even if Flickr just said something like <img alt="Main photo">, that would help users tell apart the main photo from e.g. the "next in photostream" button where Flickr forgot to put any alt attribute, so it's better than nothing

[05:27:00.0000] <Hixie>
so you want to require that flickr be conforming for one <img> so that the usability of their site is improved because you think they won't be conforming for another one?

[05:28:00.0000] <Dashiva>
So once again, we're back at something like notalt="text that isn't alternate"

[05:28:01.0000] <Hixie>
if we are assuming we can get them conforming for any images, why would we assume the main image (which is the hardest to get right) would be where they would start in becoming conforming?

[05:30:00.0000] <Philip`>
Hixie: The evidence seems to back up that assumption, because Flickr does have alt on its main image whereas it doesn't on some of its other images

[05:31:00.0000] <Hixie>
the alt on the main image is pretty bogus right now in most cases

[05:32:00.0000] <Hixie>
i would argue that the flickr site right now shows that which images get alt="" attribute is more likely to be near-random than anything else

[05:32:01.0000] <Philip`>
Presumably the main image has been there since Flickr first existed, so there's plenty of time for someone who knows 'HTML4 says I need an alt attribute' to have edited the code to generate an alt attribute, while the other images are more recent and nobody has cared to 'fix' them yet

[05:32:02.0000] <Philip`>
"pretty bogus" is still better than nothing :-)

[05:32:03.0000] <Hixie>
maybe

[05:32:04.0000] <Hixie>
and maybe

[05:32:05.0000] <Hixie>
or maybe not

[05:33:00.0000] <Philip`>
and when our choices seem to be between "nothing" and "better than nothing", the latter seems a better choice

[05:33:01.0000] <Hixie>
if it was clearly better i'd agree

[05:34:00.0000] <Philip`>
Dashiva: Or just alt="text that isn't alternate" :-)

[05:35:00.0000] <Hixie>
we could also require that all such images be in a link pointing at the image itself

[05:36:00.0000] <Dashiva>
Philip`: I'm cool with it. You just have to convince public-html

[05:36:01.0000] <Hixie>
actually he just has to convince me, and, if it is controversial, the chairs :-)

[05:37:00.0000] <Dashiva>
No, he has to convince public-html so my inbox can get peace and quiet at last ;)

[05:37:01.0000] <Hixie>
i think it's pretty obvious that there are a number of people who are unwilling to move from their positions and who have positions that are unlikely to ever gain wide acceptance

[05:38:00.0000] <Hixie>
e.g. the people who think that even decorative images should have non-empty alt

[05:40:00.0000] <Hixie>
Philip`: so what text would you suggest for the following cases? a generated image for a fractal explorer; the images in google maps in map view and satellite view; the street view images in google maps; an image uploaded as part of a batch upload with no information; a webcam

[05:42:00.0000] <Hixie>
ok going to bed for real now. will read responses tomorrow.

[05:42:01.0000] <Hixie>
nn

[05:43:00.0000] <Philip`>
Dashiva: There's an easier solution to the inbox problem: create a new mailing list for alt discussion

[05:43:01.0000] <Philip`>
Even better, create a Task Force for it

[05:43:02.0000] <Dashiva>
Philip`: Oh boy

[05:43:03.0000] <Dashiva>
Hixie: 04:45 on a sunday? That's dedication :)

[06:02:00.0000] <krijnh>
http://fronteers.nl/congres/2008/speakers#anne-van-kesteren

[07:26:00.0000] <malde_>
Hi all, is there a recommended way to use the database api to retrieve the field names of a table without selecting any data?

[07:28:00.0000] <Dashiva>
describe?

[07:29:00.0000] <Dashiva>
Maybe that's a mysql thing

[07:29:01.0000] <malde_>
describe might work. I'll take a look at the result set.

[07:31:00.0000] <malde_>
With the Gears API I can do SELECT * FROM tableName WHERE 0 = 1

[07:34:00.0000] <Philip`>
SQLite doesn't have "describe" or anything equivalent, I believe

[07:35:00.0000] <Philip`>
but it has a table named "sqlite_master" that contains the database schema, apparently

[07:41:00.0000] <Dashiva>
SQL-92 seems to require a COLUMNS table

[07:42:00.0000] <malde_>
Is sqlite_master a table in every database?

[07:46:00.0000] <malde_>
Ah, seems like it is

[07:58:00.0000] <Philip`>
Dashiva: But that's just a standard, you can't expect it to work :-p

[07:59:00.0000] <gsnedders>
malde_: yes

[07:59:01.0000] <gsnedders>
Dashiva: What actually implements any SQL standard?

[08:00:00.0000] <Philip`>
The SQL standard has things like column types, which is a feature SQLite seems quite happy to ignore

[08:00:01.0000] <Dashiva>
gsnedders: Most of them implement parts of it

[08:00:02.0000] <gsnedders>
"parts" :)

[08:00:03.0000] <Philip`>
so I'd be surprised if it implemented all the other features exactly right :-)

[08:37:00.0000] <gsnedders>
/me looks up that Varsity survey with medicine at the top and maths at the bottom

[08:39:00.0000] <Philip`>
http://www.varsity.co.uk/archive/669.pdf ?

[08:43:00.0000] <Philip`>
(Correlation is not causation, so it would be unwise to base your application decisions on such a survey :-p )

[08:43:01.0000] <gsnedders>
Philip`: Yeah, exactly :P

[08:43:02.0000] <Philip`>
/me wonders how much correlation there is between correlation and causation

[08:43:03.0000] <gsnedders>
Philip`: So, I'll apply for Clare because it has the highest recreational drug use

[08:44:00.0000] <gsnedders>
Hmm, interesting: a greater number of people identify as bi than gay.

[08:44:01.0000] <gsnedders>
/me notes that is unusual

[08:44:02.0000] <gsnedders>
/me has now dragged this channel totally off-topic :D

[08:45:00.0000] <Philip`>
It's fortunate that nobody ever lies in online surveys

[08:45:01.0000] <gsnedders>
Of couse.

[08:45:02.0000] <gsnedders>
*course

[09:52:00.0000] <zcorpan>
Hixie: "A corollary to this is that the alt attribute's value should never contain text that could be considered the image's caption, title, or legend." -- is that intended to be a SHOULD?

[10:25:00.0000] <PHPechowiec>
hi

[10:25:01.0000] <PHPechowiec>
<meta http-equiv=Creation-Date content= "Sun, 24 Aug 2008 10:27:54 GMT"> ?

[10:26:00.0000] <PHPechowiec>
can i use this line in html5?

[10:33:00.0000] <webben>
PHPechowiec: based on http://www.w3.org/html/wg/html5/#pragma , I do not believe that would conform.

[10:33:01.0000] <gsnedders>
It wouldn't.

[10:54:00.0000] <gsnedders>
Dashiva: Are you one of the people under your name on Facebook?

[10:54:01.0000] <Dashiva>
No

[10:54:02.0000] <gsnedders>
Dashiva: Are you someone on Facebook not under your name? :D

[10:54:03.0000] <Dashiva>
No

[10:55:00.0000] <gsnedders>
(Yes, I am bored)

[10:55:01.0000] <gsnedders>
(Yes, I am stalking you all)

[10:56:00.0000] <Dashiva>
I was on orkut for a few weeks before realizing it was a mistake

[10:56:01.0000] <Dashiva>
And I'm on myopera out of necessity

[10:56:02.0000] <gsnedders>
Why is it needed?

[10:57:00.0000] <Dashiva>
I did work for opera in a user-facing role :)

[10:57:01.0000] <gsnedders>
user-facing?

[10:57:02.0000] <Dashiva>
Widgets

[10:57:03.0000] <gsnedders>
No wonder I only ever saw your back.

[10:58:00.0000] <gsnedders>
(I know that is only a very vaguely witty remark, but it'll suffice)

[10:58:01.0000] <Dashiva>
Maybe you should get glasses

[10:59:00.0000] <gsnedders>
I have some in my pocket, and I'm getting new ones next week.

[11:00:00.0000] <Dashiva>
You can find me on mixi.jp though :)

[11:03:00.0000] <Dashiva>
/me notices another friend invite on myopera

[11:03:01.0000] <Dashiva>
I don't get these people who want several thousand friends they never even talk to

[11:09:00.0000] <jgraham>
/me doesn't understand gsnedders joke

[11:09:01.0000] <jgraham>
Presumably if Dashiva was in a user facing role and you only ever saw his back that would mean that you were inside Opera

[11:10:00.0000] <gsnedders>
Or I'm just not a user

[11:13:00.0000] <jgraham>
Oh I see. 

[11:13:01.0000] <Dashiva>
gsnedders isn't a user, he's an addict. *rimshot*

[11:13:02.0000] <jgraham>
I don't think that's true though. Presumably you jut wouldn't see him at all

[11:14:00.0000] <gsnedders>
Well, I'm around enough to be reverse-engineering Opera, even if I'm not really using it

[11:14:01.0000] <jgraham>
Hmm John seems to have missed my point somewhat

[11:14:02.0000] <Dashiva>
Why am I not surprised?

[11:14:03.0000] <gsnedders>
http://media.tumblr.com/1vnWCPZWQ9fd1s1k1zTvXUJM_500.png

[11:15:00.0000] <Dashiva>
4/10

[11:15:01.0000] <Dashiva>
gsnedders: You forgot to supply alternate text

[11:18:00.0000] <gsnedders>
alt="A line, labelling you, heading in a curve away from the marked point"

[11:19:00.0000] <Dashiva>
So the initial heading being towards the point is not significant?

[11:19:01.0000] <gsnedders>
/me shrugs

[11:26:00.0000] <Dashiva>
"The issue boils down to this: should conformance be worth paying any attention to without the law requiring it?  We argue no."

[11:50:00.0000] <webben>
/me thinks that this is a bit of a storm in a teacup giving the lack of general interest in conformance, and authors' happiness to ignore conformance problems they judge irrelevant to them.

[11:50:01.0000] <webben>
*given

[11:51:00.0000] <webben>
i'd be much more worried about trying to explain weirdnesses of HTML to people trying to validate their content.

[11:51:01.0000] <webben>
at least "provide a text equivalent" can be explained in human language

[11:52:00.0000] <webben>
unlike say entity errors in current HTML validation, which needs to be explained in geekspeak :)

[11:54:00.0000] <webben>
/me wonders if HTML5 make & in URLs in href conforming or still requires conversion to &amp;

[11:54:01.0000] <webben>
newbies ever understands that.

[11:54:02.0000] <webben>
*never understand

[12:02:00.0000] <Philip`>
webben: That's still non-conforming, because it's likely to result in errors (e.g. when someone wants to add a "gt" parameter to their query and writes <a href="foo.cgi?a=b&gt=c"> and it all goes horribly wrong), and so it's still worthwhile repeatedly telling authors not to do that

[12:05:00.0000] <Philip`>
Hopefully the conformance checker would detect when you have that error in a URL attribute, and would say something relatively friendly like "It looks like you are using '&' to separate values in this URL - you have to write them as '&amp;' instead", rather than saying something confusing and meaningless like "Error: Entity reference was not terminated by a semicolon."

[12:05:01.0000] <webben>
Hopefully.

[12:06:00.0000] <Dashiva>
Philip`: Only if it outputs the message as RDF

[12:08:00.0000] <webben>
apparently, however, http://www.flickr.com/ don't bother encoding to &amp; even now.

[12:09:00.0000] <Dashiva>
webben: Considering all the hate they get on public-html, maybe they decided conformance wasn't worth trying for :)

[12:09:01.0000] <webben>
which I suppose points at the problem of requiring things that only help sometimes.

[12:09:02.0000] <webben>
Dashiva: Flickr's non-conformance is long-standing; I'm not sure the public-html discussions about Flickr have very much visibility.

[12:11:00.0000] <webben>
I see people regularly use & instead of &amp; because they think it will work, even when they care about other aspects of valid HTML.

[14:10:00.0000] <jgraham>
Sigh

[14:11:00.0000] <annevk>
/me notices people have been having fun over the weekend

[14:11:01.0000] <annevk>
/me played Super Mario World and Call of Duty 4

[14:12:00.0000] <jgraham>
heh

[14:13:00.0000] <annevk>
and I used my friends high upload bandwidth to get my Cambridge photos on flickr

[14:13:01.0000] <Dashiva>
annevk: I'm sure you could make a game out of public-html. They managed to come up with buzzword bingo, after all :)

[14:14:00.0000] <jgraham>
/me wonders if the a11y people only play text based adventures

[14:16:00.0000] <Dashiva>
jgraham: Making textual alternates available would make most games very prone to macroing/botting/etc

[14:16:01.0000] <jgraham>
annevk: Did you have fun punting in the rain?

[14:17:00.0000] <zcorpan>
hsivonen: fwiw, i now didn't get that error message when filing a bug

[14:18:00.0000] <Philip`>
I like how some games (at least Portal, and probably the Half Life 2 games too) have closed captioning, so you can get all the audio cues even without any audio

[14:19:00.0000] <hsivonen>
webben: There's indeed tension between saying "this stuff sucks so you shouldn't use it for new content" and "this stuff sucks so badly that you should go and remove if from the content you are upgrading"

[14:19:01.0000] <hsivonen>
webben: I think we don't have a good solution that respects both the time of upgraders and of authors starting from a clean slate

[14:20:00.0000] <webben>
well, one could actually differentiate between the two

[14:20:01.0000] <annevk>
btw, if you like Mario, http://www.youtube.com/watch?v=in6RZzdGki8 is quite funny

[14:20:02.0000] <webben>
whether it's good to do so or not I don't know.

[14:20:03.0000] <annevk>
jgraham, there wasn't too much rain actually

[14:21:00.0000] <webben>
hsivonen: for example one could error in response to &gt= but only warn for &name=

[14:21:01.0000] <annevk>
jgraham, it was fun to do

[14:21:02.0000] <jgraham>
Ah, I saw Hakon had his umbrella out

[14:22:00.0000] <annevk>
it's from Bert iirc

[14:22:01.0000] <annevk>
but the drops were hardly noticable for most of it

[14:28:00.0000] <jgraham>
Sigh**2

[14:30:00.0000] <annevk>
maybe you should set up a proxy that filters out WHATWG and HTML5 related e-mail until your thesis is done :)

[14:30:01.0000] <annevk>
though I guess it gives some distraction

[14:34:00.0000] <hsivonen>
webben: the problem is how to distinguish them in a way that doesn't fail like HTML 4.01 Strict vs. Transitional?

[14:34:01.0000] <hsivonen>
webben: it seems a bit pointless to make the validator emit messages at all where the messages are meant to be ignored

[14:34:02.0000] <webben>
hsivonen: warnings aren't "meant to be ignored".

[14:35:00.0000] <jgraham>
annevk: Not such a bad idea I guess. But I think my thesis will get done. It's just annoying to try and explain something clearly and then realise that you didn't get your message across at all

[14:35:01.0000] <hsivonen>
webben: I'm getting into the game of analyzing millions of pages in order to get data about what kind of errors are the ones that would particularly waste the time of upgraders

[14:36:00.0000] <webben>
hsivonen: I think the "failure" (if there was one) had to do with choices about what went into strict that disagreed with what authors wanted to use and that worked in the browsers they cared about.

[14:36:01.0000] <annevk>
jgraham, maybe you should try to enforce your points by law

[14:37:00.0000] <webben>
hsivonen: I suppose target and iframe being the most important examples.

[14:37:01.0000] <annevk>
jgraham, they'll probably be familiar with that

[14:37:02.0000] <annevk>
/me hides

[14:37:03.0000] <hsivonen>
webben: as far as HTML tables and IE installed base are concerned, we have that kind of problem with HTML5

[14:37:04.0000] <Dashiva>
webben: Didn't stop target from being a point of contention in html5 as well :)

[14:38:00.0000] <webben>
Dashiva: Well of course not. Authors wanting to use doesn't make it a great idea.

[14:38:01.0000] <Dashiva>
Whether it's a good idea or not doesn't mean much if it leads to scrapping your strict version by authors

[14:39:00.0000] <webben>
depends on what the point of the versioning is

[14:39:01.0000] <webben>
if the point of conformance is to get "good practice" but you can only get conformance by sanctioning "bad practice" that's a largely pointless exercise.

[14:40:00.0000] <hsivonen>
fwiw, I'm not so much concerned about pushing an ideal of good practice

[14:40:01.0000] <Dashiva>
webben: Not if allowing one bad practice means they fix a dozen others

[14:40:02.0000] <hsivonen>
I'm more interested in producing a tool that allows authors catch mistakes they didn't make

[14:41:00.0000] <hsivonen>
although Validator.nu does have political good practice aspects

[14:41:01.0000] <webben>
what are "mistakes they didn't make"?

[14:41:02.0000] <hsivonen>
oops

[14:41:03.0000] <hsivonen>
didn't *intend* to make

[14:42:00.0000] <webben>
how can you intend to make a mistake?

[14:42:01.0000] <Dashiva>
E.g. if you're using a bad practice on purpose

[14:42:02.0000] <webben>
is that good or bad? using a "bad practice" on purpose.

[14:42:03.0000] <hsivonen>
webben: if you use <embed> in HTML 4.01 on purpose, for example

[14:43:00.0000] <webben>
that's not a "mistake". that's a conformance error.

[14:43:01.0000] <webben>
or an instance of deliberate non-conformance

[14:44:00.0000] <webben>
validators can't really do much about instances of deliberate non-conformance

[14:44:01.0000] <annevk>
/me wonders what the point of this conversation is

[14:44:02.0000] <webben>
the point for me, is increasing puzzlement about what conformance in HTML5 is, and its integrity as a single thing.

[14:45:00.0000] <hsivonen>
annevk: I guess the point on my side is that I don't believe as much in the wishful part of conformance as I think Hixie does

[14:45:01.0000] <hsivonen>
annevk: when it comes to removing "bad" legacy features from the conforming language

[14:46:00.0000] <webben>
it seems to me the one bit of conformance everyone (?) would buy into is where you recommend features that work in lots of user agents over features that do the same thing but in fewer user agents.

[14:48:00.0000] <webben>
Hixie seems keen to use conformance to protect future extension points, which is a good idea, but of less widespread appeal.

[14:48:01.0000] <annevk>
works for CSS

[14:49:00.0000] <hsivonen>
webben: that's not what making align on td non-conforming is about, though

[14:49:01.0000] <webben>
annevk: CSS has extension points built into the language.

[14:49:02.0000] <webben>
annevk: I don't think CSS validation/conformance has been a particularly powerful force for authors.

[14:49:03.0000] <webben>
e.g. non-validating hacks, opacity, filter, vendor-specific extensions

[14:50:00.0000] <webben>
with the last one built-in, but forbidden

[14:50:01.0000] <hsivonen>
I stopped paying attention to CSS validation when I couldn't configure the validator to grok Prince extensions

[14:50:02.0000] <annevk>
the validator should probably group extensions as a single error

[14:51:00.0000] <hsivonen>
annevk: this was the CSS validator running inside Oxygen

[14:51:01.0000] <annevk>
my impression has always been that quite a few authors do use the CSS validator, but that's anecdotical

[14:52:00.0000] <webben>
Then there's conformance for trying to change authoring practices that could lead to errors, even when they in fact aren't errors that cause problems. e.g. &name in href.

[14:52:01.0000] <hsivonen>
or spaces in URLs most of the time

[14:52:02.0000] <webben>
While I'm glad of a tool to check that, I suspect that will reduce the amount of conformance considerably.

[14:53:00.0000] <webben>
e.g. lots of relatively conforming webpages fail to encode ampersands now

[14:53:01.0000] <annevk>
&name could cause a problem going forward

[14:53:02.0000] <annevk>
though that is somewhat avoided by requiring ; for new entities

[14:53:03.0000] <webben>
annevk: yeah well I'm talking about &name= not &name;

[14:54:00.0000] <webben>
and if one didn't require ; that would break _old_ content using &name=

[14:54:01.0000] <webben>
of which there's plenty

[14:54:02.0000] <annevk>
&name is slightly suboptimal from a parser point of view

[14:54:03.0000] <annevk>
so an author might want to optimize that

[14:55:00.0000] <webben>
yeah. they _might_.

[14:55:01.0000] <annevk>
anyway, if you want &name to be conforming, you want whatwg⊙wo

[14:56:00.0000] <webben>
personally, I don't care whether it's conforming or not particularly (I just use &amp; ). I just think it's interesting as an aspect of the emerging picture of HTML5 conformance.

[14:57:00.0000] <annevk>
ok

[14:59:00.0000] <webben>
my guess (it's only a guess) is that there will be more websites failing conformance because of ampersand rules than would if HTML5 required alt (for example), since I see lots of websites with alt but that fail to encode ampersands

[14:59:01.0000] <webben>
Flickr probably won't conform whatever the conformance rules are (alt isn't a "problem" for Flickr now, it just uses alt="" on the main photo page; it doesn't encode ampersands however).

[15:00:00.0000] <webben>
given those speculations, I find much of the alt debate rather odd

[15:00:01.0000] <annevk>
hsivonen has stats on that

[15:01:00.0000] <annevk>
I find it rather odd for very different reasons :D

[15:01:01.0000] <webben>
oh there are other odd things said by most correspondents

[15:04:00.0000] <hsivonen>
annevk: what do I have stats on?

[15:05:00.0000] <webben>
hsivonen: websites with/without alt, with/without ampersands encoded properly, combinations thereof, I think?

[15:05:01.0000] <annevk>
common conformance errors on pages

[15:05:02.0000] <webben>
oh, pages is somewhat different

[15:05:03.0000] <webben>
e.g. photo sharing sites have lots of pages, few links.

[15:05:04.0000] <hsivonen>
when I ran the alexa 500 through the validator, absent alt wasn't a machine-checkable conformance error

[15:06:00.0000] <webben>
note I said "websites failing conformance"

[15:06:01.0000] <annevk>
nm then hsivonen, seems I missed his point

[15:07:00.0000] <hsivonen>
hmm. looks like my mass validation harness is broken or the disk image containing the test data has a corrupt file system...

[15:07:01.0000] <hsivonen>
or both...

[15:09:00.0000] <annevk>
http://serialseb.blogspot.com/2008/08/proposing-syntax-to-attach-behaviors-to.html#c8752145182107351356 ?

[15:09:01.0000] <annevk>
zcorpen ? ^^

[15:10:00.0000] <annevk>
http://ajaxworld.com/general/sessiondetail1008.htm?id=57

[15:11:00.0000] <annevk>
people from Kaazing Corporation will speak about HTML5...

[15:51:00.0000] <jcranmer>
I prefer i.m.o's GHOST quit message

[15:58:00.0000] <hsivonen>
what's the deal with the new delicious.com wanting me to sign in again and again and putting autocomplete=off in the login fields?

[15:58:01.0000] <Hixie>
zcorpan: yeah, it's a should on purpose. I'm sure somebody will find some case where it would be appropriate. :-)

[15:58:02.0000] <hsivonen>
do they think they are a bank now?

[16:10:00.0000] <hsivonen>
This Week in HTML5 makes it on Robin Cover's radar: http://xml.coverpages.org/newsletter/news2008-08-22.html

[16:11:00.0000] <Hixie>
sweet

[16:11:01.0000] <Hixie>
go mark

[16:12:00.0000] <hsivonen>
Google found it for me through a splog instead of the original source, though, so slogs are useful for something

[16:13:00.0000] <Hixie>
hah

[16:13:01.0000] <Hixie>
it would have found the real thing eventually

[16:13:02.0000] <Hixie>
i think we need to do something to raise awareness of the way we're doing html5

[16:13:03.0000] <Hixie>
 - the openness

[16:13:04.0000] <Philip`>
/me thinks the XML Daily Newslink would be somewhat easier to read if it used paragraphs

[16:13:05.0000] <Hixie>
 - the history

[16:14:00.0000] <Hixie>
 - the non-consensus approach

[16:14:01.0000] <Hixie>
i'm not sure how to get this awareness raised exactly

[16:15:00.0000] <Hixie>
but several people have come to me recently with quite wrong preconceptions and then changed their attitude quite dramatically when i explained to them the history and so on

[16:22:00.0000] <webben>
hsivonen: Have you tried following "Why does Delicious keep asking me to log in?" at http://delicious.com/help/faq ?

[16:23:00.0000] <webben>
hsivonen: the inclusion of autocomplete is, hmm, curious to say the least

[16:24:00.0000] <webben>
we don't have that on our main sign in boxes.

[16:24:01.0000] <jgraham>
Hixie: What preconceptions did they have?

[16:26:00.0000] <Hixie>
that it was a concensus-driven, closed working group that ignored the needs of users

[16:30:00.0000] <jgraham>
Did they mainly associate HTML5 with WHATWG or W3C?

[16:30:01.0000] <Hixie>
i don't think they really understood the difference

[16:30:02.0000] <Hixie>
which makes sense

[16:30:03.0000] <Hixie>
i'm not really worried about that per se

[16:31:00.0000] <jgraham>
Interesting. I was wondering if they thought we were just another W3C working group or if they had independtly reached the wrong conclusions about WHATWG

[16:32:00.0000] <Hixie>
good question

[16:32:01.0000] <Hixie>
i don't know

[16:34:00.0000] <jgraham>
(it should be noted that some people in the HTMLWG think we are a consensus driven group apt to ignore the needs of users...)

[16:35:00.0000] <gsnedders>
My tongue hurts. I oughtn't bite it.

[16:35:01.0000] <jgraham>
gsnedders: Literally?

[16:35:02.0000] <gsnedders>
Literally.

[16:36:00.0000] <webben>
yeah, eating yourself - not the best of ideas ;)

[16:36:01.0000] <Hixie>
i bite my tongue literally all the time, hurts a bitch

[16:37:00.0000] <Hixie>
and it inflates when you bite it, which just makes biting it more likely

[16:37:01.0000] <Hixie>
it's terrible

[16:37:02.0000] <Hixie>
jgraham: yeah, i'm not too worried about them. :-)

[16:37:03.0000] <jgraham>
Hixie: They know :)

[16:38:00.0000] <gsnedders>
Hixie: Exactly. I normally do it to the inside of my cheek, though, and not my tongue.

[16:39:00.0000] <jgraham>
gsnedders: I take it this was accidentially biting it when you were aiming for food or something

[16:39:01.0000] <gsnedders>
jgraham: Oh, probably just from thinking too hard

[16:41:00.0000] <Philip`>
You can get various bitter-tasting liquids that you can put on your fingernails to stop yourself biting them; maybe you could put the same stuff on your tongue to stop you biting that

[16:42:00.0000] <Hixie>
-_-

[16:42:01.0000] <gsnedders>
Philip`: Tongue has tastebuds though, fingernails don't

[16:43:00.0000] <Philip`>
Oh, darn

[16:43:01.0000] <gsnedders>
/me notes he can't wear any sort of toxic nail varnish, because he bites his nails too much (or at least, sort of sucks on the end of his fingers)

[16:44:00.0000] <gsnedders>
(Yes, I did just make a reference to wearing nail varnish)

[16:45:00.0000] <gsnedders>
And even if I put a bitter-tasting one on I still do it

[16:45:01.0000] <gsnedders>
/me is hopeless

[16:46:00.0000] <jgraham>
gsnedders: You can wear toxic nail varnish in just the same way that everyone else can. It's just you're more likely to die the less-than-magnanimous death by nail-varnish poisoning

[16:46:01.0000] <gsnedders>
jgraham: :)

[16:47:00.0000] <Philip`>
Better than death by tea cosy

[16:47:01.0000] <Philip`>
(Wikipedia reckons there's a 1 in 20 billion chance of that)

[16:48:00.0000] <gsnedders>
Philip`: Link?

[16:48:01.0000] <gsnedders>
http://en.wikipedia.org/wiki/List_of_causes_of_death_by_rate ?

[16:48:02.0000] <gsnedders>
No, that doesn't contain tea cosy

[16:48:03.0000] <Philip`>
http://en.wikipedia.org/wiki/Tea_cosy

[16:49:00.0000] <gsnedders>
One in twenty billion isn't bad.

[16:49:01.0000] <gsnedders>
I think the probability of dying is higher than that.

[16:50:00.0000] <csarven>
HMm.. Suicide is more likely then drowning

[16:50:01.0000] <gsnedders>
/me almost committed suicide by drowning

[16:50:02.0000] <csarven>
*common

[16:50:03.0000] <gsnedders>
Which means I could've counted towards both of those stats!

[16:52:00.0000] <BenMillard>
jgraham, the "Deliverable for Action 72 @headers" got through my junk filters

[16:53:00.0000] <jgraham>
BenMillard: What do I win?

[16:53:01.0000] <BenMillard>
jgraham, me as an audience for your messages :)

[16:54:00.0000] <jgraham>
gsnedders: I guess given some sort of afterlife it would have given you the oppertunity to have a party anecdote about things you had in common with Virgina Woolfe</macabre>

[16:54:01.0000] <jgraham>
BenMillard: A fine prize indeed, I feel

[16:55:00.0000] <BenMillard>
wewt

[16:55:01.0000] <BenMillard>
your table inspector's highlighting seems to work more reliably than I remember from last year

[16:56:00.0000] <jgraham>
BenMillard: Did you start using Firefox 3 in the interim? That might have made a difference, or I might have changed the code

[16:56:01.0000] <BenMillard>
jgraham, still Firefox 2. changes to the code sounds likely

[16:57:00.0000] <BenMillard>
jgraham, you wrote "I think the absolute simplest message that we can give authors is 'mark up your table headers as <th>'."

[16:57:01.0000] <jgraham>
gsnedders: Although I should say that the rest of us are much happier with a gsendders without that particular anecdote :)

[16:57:02.0000] <BenMillard>
jgraham I agree with that quote...think I said that at the November 2007 meeting

[16:58:00.0000] <BenMillard>
there's a common misconception I come across, where accessibility is assumed to require complexity

[16:59:00.0000] <jgraham>
BenMillard: You might well have done. It bothers me that a lot of accessibility types seem to have the idea that as long as there is some way to do something then that's good enough, even if it is really hard to use

[16:59:01.0000] <jgraham>
(the first sentence referred to "said that in Nov. 2007)


2008-08-25
[17:00:00.0000] <BenMillard>
yeah, I gathered :)

[17:00:01.0000] <BenMillard>
often the things they want to do aren't at all useful

[17:00:02.0000] <BenMillard>
like creating associations between things that are right next to each other

[17:01:00.0000] <Hixie>
careful, you might lose your accessibility expert credentials if you keep saying sensible things like that :-)

[17:01:01.0000] <BenMillard>
hixie, lol :P

[17:03:00.0000] <jgraham>
BenMillard: That's why I keep suggesting that user testing is a stronger argument than theory. 

[17:03:01.0000] <BenMillard>
jgraham, indeed

[17:04:00.0000] <BenMillard>
I wonder if I have lost more than I've gained by ignoring Public-HTML for the past several months

[17:04:01.0000] <BenMillard>
I've certainly gained a lot of time, with which I've done many fun and some productive things

[17:06:00.0000] <jgraham>
BenMillard: It was pretty quiet for a bit. Then about a week or two ago it all kicked off again

[17:07:00.0000] <jgraham>
Same topics, similar arguments

[17:08:00.0000] <BenMillard>
since it all gets publically archived, I wonder if there's any gain in re-responding

[17:08:01.0000] <Philip`>
It gets archived but there's so much that nobody's ever going to actually read the archives

[17:09:00.0000] <Philip`>
so clearly the solution is to keep reposting the better arguments, so that nobody will miss them

[17:10:00.0000] <jgraham>
BenMillard: I guess the problem is xkcd syndrome http://xkcd.com/386/

[17:11:00.0000] <BenMillard>
Philip`, alternatively, the people who decide HTML5's development probably saw and took on board the better points first time around, so repeating them doesn't change HTML5's direction very much?

[17:12:00.0000] <BenMillard>
jgraham, perfect!

[17:14:00.0000] <Hixie>
there are three main topics being discussed right now

[17:14:01.0000] <Hixie>
img alt: i think i missed a couple of key things the first time (how the {} thing affects tool developers)

[17:15:00.0000] <Hixie>
well, i guess the second time, not the first time

[17:15:01.0000] <Hixie>
so i don't know what i'll do with that, probably go back to just making alt="" be not given when you have no alt text

[17:15:02.0000] <Hixie>
table headers: i haven't yet seen a clear description of the problem, but if there is a problem, then we'll have to address it

[17:16:00.0000] <Hixie>
rdfa: i haven't managed to get ben to actually describe the problem yet

[17:17:00.0000] <BenMillard>
Hixie, could you disambiguate when you talk about "Ben"? a couple of times I've seen you say that and it's put me into a state of panic since I don't remember that happenning :)

[17:18:00.0000] <webben>
if "no alt text" means no precise equivalent, I think that would probably force me into non-conformance since I don't think that yields the best results for end-users.

[17:19:00.0000] <Hixie>
oh sorry

[17:19:01.0000] <Hixie>
i mean ben adida in this case

[17:21:00.0000] <Hixie>
webben: so what text would you suggest for the following cases? a generated image for a fractal explorer; the images in google maps in map view and satellite view; the street view images in google maps; an image uploaded as part of a batch upload with no information; a webcam

[17:21:01.0000] <webben>
Hixie: I'm happier with a text description of those than no alt text.

[17:21:02.0000] <Hixie>
in all of these cases, the software has no idea what the image is so can't describe it at all

[17:21:03.0000] <webben>
Yes, a text equivalent is better, but I think a text description is better than zilch.

[17:22:00.0000] <webben>
Hixie: alt="fractal" alt="uploaded photo"

[17:22:01.0000] <jgraham>
Hixie: FWIW my guess is that there is a strong enough use case for people accessing the images out of the main flow that something like "fractal", "map tile", "street view", "photo", "webcam" would be better than nothing

[17:23:00.0000] <Hixie>
webben: yeah see i don't agree that that is better, i think it's actually worse

[17:23:01.0000] <jgraham>
but IANAATU

[17:23:02.0000] <Hixie>
ATU?

[17:23:03.0000] <Hixie>
AT user?

[17:23:04.0000] <jgraham>
yes :)

[17:23:05.0000] <Hixie>
woot

[17:23:06.0000] <Hixie>
well that's what the {phoot} idea was for

[17:23:07.0000] <Hixie>
{photo} even

[17:24:00.0000] <webben>
well, i'd prefer being able to demark an alt as non-equivalent with an attribute, but i'd still include it.

[17:24:01.0000] <jgraham>
My feeling is that in order of worst to best we have nothing, alt={photo}, no-text-equivalent alt="photo"

[17:25:00.0000] <jgraham>
But I agree that all three suck for different reasons

[17:25:01.0000] <Hixie>
i just don't see an attribute working for this

[17:25:02.0000] <Hixie>
it has so many issues

[17:26:00.0000] <Hixie>
i think the net misuse of a new attribute would make the total level fo accessibility lower

[17:26:01.0000] <BenMillard>
I've seen gallery software use alt="Photo" as a default

[17:26:02.0000] <BenMillard>
forums have "User posted image" or similar

[17:26:03.0000] <jgraham>
I guess the fourth possibility is no attribute and no special syntax, just require the end user to work out if it is supposed to be a description or a text equivalent

[17:27:00.0000] <webben>
Hixie: does the attribute have any use-cases other than for hypothetical future software?

[17:27:01.0000] <BenMillard>
jgraham, yes, which is hardly much of a challenge in the cases I've seen

[17:27:02.0000] <webben>
Hixie: e.g. do you expect a near-future version of Lynx to treat non-equivalents differently?

[17:28:00.0000] <webben>
or was the idea that {} would force Lynx into treating them differently?

[17:28:01.0000] <jgraham>
webben: I think in theory AT could only voice "image" if there is @n-t-e or {} and just read the alt text with no delimiter otherwise

[17:29:00.0000] <webben>
/me can't see why an AT would want to do that.

[17:29:01.0000] <jgraham>
But in practice misuse of {} or an attribute might make that not work

[17:29:02.0000] <jgraham>
webben: Want to do what?

[17:29:03.0000] <webben>
'voice "image" if there is @n-t-e or {} '

[17:30:00.0000] <webben>
oh wait i see what you mean

[17:30:01.0000] <Philip`>
jgraham: Misuse of alt="photo" (regardless of what the spec says) means that in practice AT can't just read the alt text with no delimiter

[17:32:00.0000] <jgraham>
Philip`: I would expect that to be the case as well. So unless there is something I haven't thought of (likely) it seems just as well to say "sometimes @alt is a replacement, sometimes it is a description, work it out for yourselves"

[17:32:01.0000] <jgraham>
(with rules like the current ones about when it should be a replacement and when a description)

[17:33:00.0000] <Hixie>
the idea is that on an ideally written page, an <img> with alt="" can be completely substituted for its alt="" text, with no indication that an image was there

[17:33:01.0000] <jgraham>
I guess it makes Henri's image checker less useful

[17:33:02.0000] <Hixie>
but that these <img>s we're talking about would need to indicate that there is an image present

[17:33:03.0000] <Hixie>
since that's the whole point

[17:34:00.0000] <Hixie>
so we theoretically need a distinction

[17:34:01.0000] <Hixie>
lack of alt="", {}, n-t-e, something

[17:34:02.0000] <jgraham>
Hixie: The problem with designing for an ideally written page is that you're pretty much the only person who writes them :)

[17:35:00.0000] <webben>
/me isn't persuaded that would be an ideal page.

[17:35:01.0000] <jgraham>
(although when I read feeds in thunderbird with images disabled, it is immensely irritating when people describe a picture rather than replace it either with "" or an actual equivalent)

[17:36:00.0000] <jgraham>
webben: For whom? I think part of the issue might be that the text browser case is a bit different from the screenreader case

[17:36:01.0000] <webben>
People like to communicate with images. Therefore being able to retrieve and share and bookmark etc images is an important use-case.

[17:36:02.0000] <webben>
i want to be able to do that with a text browser, I think I'd want to be able to do that if I were a screen reader user too.

[17:37:00.0000] <jgraham>
Hmm.

[17:37:01.0000] <Hixie>
you don't want to bookmark most images

[17:37:02.0000] <Hixie>
you probably don't even notice most images

[17:37:03.0000] <jgraham>
/me is going to bed now even though he hasn't managed to put a stop to wrongness on the internet :)

[17:37:04.0000] <Hixie>
e.g. "RSS" is often an image

[17:37:05.0000] <webben>
/me never said most images.

[17:37:06.0000] <Hixie>
sure

[17:38:00.0000] <Hixie>
i'm just saying that it's an important use case but only for a small subset of images

[17:38:01.0000] <webben>
although being able to grab that icon _is_ useful to people authoring content

[17:39:00.0000] <BenMillard>
jgraham, it's good to break that habit

[17:39:01.0000] <BenMillard>
as in, put your personal wellbeing ahead of timely responses to internet arguments

[17:53:00.0000] <webben>
Hixie: being able to differentiate that subset (@role?) might be useful. but i think that would need to work as an authorial opt-in, since most people probably wouldn't bother and there's a huge body of existing content not making such a differentiation.

[18:06:00.0000] <Hixie>
the problem with any attribute here is it'd be massively misused

[18:07:00.0000] <Hixie>
much more so than a special syntax, based on what i've seen of web authoring practices so far

[18:07:01.0000] <Dashiva>
It seems they're all tripping over each other in making @scope seem like a plague...

[18:10:00.0000] <webben>
Hixie: not having to include alt will also be massively "misused".

[18:10:01.0000] <Hixie>
it already is

[18:10:02.0000] <Hixie>
i don't think this would make things worse

[18:10:03.0000] <webben>
What's the reasoning for that?

[18:10:04.0000] <Hixie>
if anything it might be a net benefit since we might reduce the amount of totally bogus alt text, which is arguably worse than no alt text in some cases

[18:11:00.0000] <webben>
is there evidence that totally bogus alt text is a worse overall problem for users than no alt text?

[18:12:00.0000] <webben>
and might that change if the "no alt text" group got higher?

[18:12:01.0000] <webben>
*larger

[18:13:00.0000] <Hixie>
"I then went on safari. blahblah The elephants were cool, but the tigers were the best: bdasflsjd" is not as good an experience as "I then went on safari. [IMAGE] The elephants were cool, but the tigers were the best: [IMAGE]"

[18:14:00.0000] <Hixie>
(not that either is especially great, of course)

[18:14:01.0000] <webben>
i'm not talking about the individual experience

[18:14:02.0000] <webben>
i'm talking about which actually constitutes the bigger problem for end-users on the web now.

[18:14:03.0000] <Hixie>
(I posit that the second one above is also better than "I then went on safari. elephants The elephants were cool, but the tigers were the best: tigers")

[18:15:00.0000] <webben>
I posit "I then went on safari. image elephants The elephants were cool, but the tigers were the best: image tigers " is even better.

[18:15:01.0000] <Hixie>
(I personally alsos think the [IMAGE] case, where "[IMAGE]" is something the UA shows in a different font/voice/color/whatever, is better than "I then went on safari. photo The elephants were cool, but the tigers were the best: photo")

[18:16:00.0000] <Hixie>
i don't know of any reasearch into what problems blind users have experienced on the web so far

[18:17:00.0000] <Hixie>
my own personal attempts at using screen readers have indicated that the screen readers themselves are a far bigger problem than anything the web pages do (based primarily on my experience with JAWS)

[18:17:01.0000] <webben>
well it's not just blind users who suffer from no alt and bogus alt.

[18:18:00.0000] <Hixie>
text browser users suffer from it a lot too, but there aren't many of us (i use text browsers several times a week and alt="" isn't a big deal for me, though personally I would much rather people omit alt="" than give bogus or repeating alt"")

[18:18:01.0000] <BenMillard>
search engines are also a UA which makes use of alt for the benefit of many users, for example, image searches

[18:19:00.0000] <Hixie>
i work for a search engine

[18:20:00.0000] <Hixie>
i assure you alt="" is the least of our worries

[18:20:01.0000] <Hixie>
though if we were forced to comment on it, we'd much rather have no alt="" at all than have bogus alt=""

[18:20:02.0000] <Hixie>
repetition isn't such a big deal for us though

[18:21:00.0000] <Hixie>
since we can just filter out repeated text

[18:21:01.0000] <webben>
also voice recognition users

[18:21:02.0000] <Hixie>
voice recognition users?

[18:21:03.0000] <webben>
*speech recognition, sorry

[18:21:04.0000] <Hixie>
you mean for input? or do you mean speech synthesis?

[18:21:05.0000] <webben>
input

[18:22:00.0000] <Hixie>
i don't understand how alt="" affects input, can you elaborate?

[18:22:01.0000] <webben>
images and icons that are controls mainly

[18:22:02.0000] <webben>
e.g. an image of home, you say "home", the speech recognition looks for a link or button with "home" as a detectable text equivalent

[18:23:00.0000] <webben>
I think in practice they have to make a lot of use of title too; I dunno how they try to handle missing alt and title other than maybe link numbering.

[18:24:00.0000] <webben>
Hixie: here's some actual HTML guidelines from a vendor: http://ct.scansoft.com/customerfiles/kbasefiles/3067/wp_DNS_HTML.pdf in case that's of interest

[18:25:00.0000] <Hixie>
oh well for images that are controls the spec is clear that alt="" is required

[18:25:01.0000] <webben>
I was going to ask. That's good at least.

[18:25:02.0000] <webben>
that won't change if you make alt non-conforming?

[18:26:00.0000] <webben>
what if you have a div with tabindex and onclick and an img inside it?

[18:26:01.0000] <Hixie>
the only thing in http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded0.html#alt that is likely to change is the tiny subsubsection labeled "Images whose contents are not known"

[18:27:00.0000] <webben>
yeah, problem is it's not easy for a validator to distinguish those from other images.

[18:27:01.0000] <Hixie>
/me looks at the file above

[18:27:02.0000] <Hixie>
oh well images are far from the biggest problem there

[18:28:00.0000] <Hixie>
i mean, you have to use <h1> only for headers but how is the validator supposed to know?

[18:29:00.0000] <webben>
Hixie: well, yes, it's a general problem with conformance testing. but in terms of trying to ensure control images have alt, it's potentially important

[18:29:01.0000] <webben>
depends how you define control images in such a way that a checker could distinguish the two

[18:30:00.0000] <Hixie>
i don't think we can

[18:30:01.0000] <Hixie>
but we have this problem anyway distinguishing decorative elements from other elements

[18:30:02.0000] <webben>
Hixie: can you define a subset of controls that can be checked?

[18:30:03.0000] <Hixie>
well if you read the link i gave above you'll see one of the first things it does say though is that an image in a link that has no other text must have non-empty alt=""

[18:31:00.0000] <Hixie>
which is probably all wecan do really

[18:31:01.0000] <Hixie>
machine-checkability-wise, anyway

[18:32:00.0000] <Hixie>
dragon's document seems in line with what the spec says today

[18:32:01.0000] <Hixie>
doesn't really affect the problematic images we were talking about

[18:32:02.0000] <webben>
also button and input type="image" and imagemap?

[18:33:00.0000] <Hixie>
<input type=image> and <area> are dealt with separately and have separate requirements that basically always require non-empty alt=""

[18:33:01.0000] <webben>
Hixie: I don't think it does affect the problematic images, no.

[18:33:02.0000] <webben>
okay

[18:33:03.0000] <Hixie>
(well <input> isn't in the spec yet but will be)

[18:33:04.0000] <webben>
button?

[18:33:05.0000] <Hixie>
what about it?

[18:33:06.0000] <webben>
shouldn't the spec mention button along with link at the above url?

[18:33:07.0000] <Hixie>
oooh, good call

[18:57:00.0000] <Hixie>
webben: spec updated

[22:14:00.0000] <takkaria>
urgh, 521 new mails

[01:10:00.0000] <Hixie>
what does the "@" mean in "-rw-r--r--@"?

[01:12:00.0000] <hsivonen>
Hixie: ACL I think

[01:13:00.0000] <Hixie>
any idea how i view the acl info?

[01:14:00.0000] <Hixie>
ls -e didn't work

[01:15:00.0000] <jacobolus>
Hixie: http://arstechnica.com/reviews/os/macosx-10-4.ars/8

[01:15:01.0000] <jacobolus>
really, -e didn't work?

[01:15:02.0000] <jacobolus>
hrm

[01:15:03.0000] <Hixie>
that suggests @ does not mean acl

[01:15:04.0000] <Hixie>
since no @ appears in their listings

[01:15:05.0000] <jacobolus>
indeed

[01:15:06.0000] <hsivonen>
Hixie: I can't remember and Googling gives me the article jacobolus mentioned

[01:15:07.0000] <Hixie>
maybe it means there's fork data or something

[01:16:00.0000] <Hixie>
or attributes

[01:16:01.0000] <hsivonen>
I might misremember the meaning of @

[01:17:00.0000] <jacobolus>
Hixie: If the file or directory has extended attributes, the permissions field printed by the -l option is followed by a '@' character.

[01:18:00.0000] <hsivonen>
xattr --list file

[01:18:01.0000] <jacobolus>
so actually the page you want is http://arstechnica.com/reviews/os/macosx-10-4.ars/7 :)

[01:18:02.0000] <Hixie>
yeah i was just playing with mdls which also seems to list attributes

[01:19:00.0000] <Hixie>
ok anyway

[01:19:01.0000] <Hixie>
it is an attributes thing

[01:20:00.0000] <jacobolus>
no, mdls lists info from the spotlight store

[01:20:01.0000] <jacobolus>
which is different from xattrs

[01:21:00.0000] <hsivonen>
are ACLs a special case of attributes?

[01:21:01.0000] <jacobolus>
and it is *very very* frustrating that setting custom xattrs can’t put custom metadata in spotlight :(

[01:21:02.0000] <jacobolus>
hsivonen: yes, ACLs are stored in xattrs

[01:21:03.0000] <hsivonen>
ok. so my memory isn't totally broken

[01:22:00.0000] <Hixie>
man mac is confusing :-P

[01:23:00.0000] <jacobolus>
hsivonen: I'm not sure you can get to them though through normal xattr APIs

[01:24:00.0000] <Hixie>
/me sighs loudly

[01:25:00.0000] <Hixie>
i wish i had some data to go on with this <img> thing

[01:25:01.0000] <Hixie>
i really have no idea what to do

[01:25:02.0000] <Hixie>
which is a pretty ridiculous position to be in given the rarity with which this will affect users

[01:25:03.0000] <Hixie>
(it's not like blind users surf to flickr often)

[01:26:00.0000] <jacobolus>
why do you have no idea what to do?

[01:26:01.0000] <jacobolus>
I thought it was figured out

[01:26:02.0000] <jacobolus>
with the {}'s etc.

[01:26:03.0000] <Hixie>
people have raised a valid concern with the {}

[01:27:00.0000] <Hixie>
which is that it increases the effort of people trying to write accessible content

[01:27:01.0000] <hsivonen>
Hixie: do you have data on how often people with text input difficulties execise their right to speech by publishing photos?

[01:27:02.0000] <jacobolus>
increases the effort?

[01:27:03.0000] <jacobolus>
as in, it's too hard to put {…} in?

[01:27:04.0000] <Hixie>
because they now have to work around cases where the alt="" might end up containing {} through no fault of their own, e.g. if the alt is user input

[01:28:00.0000] <tantek>
what's wrong with {} ?

[01:28:01.0000] <Hixie>
hsivonen: 100% of my sample of 1 has done it

[01:28:02.0000] <jacobolus>
that seems like a vanishingly unlikely edge case

[01:28:03.0000] <jacobolus>
what naïve user is going to make the whole alt name start w/ { and end w/ }?

[01:29:00.0000] <Hixie>
jacobolus: yeah but the people who care about accessibility (the group we want to piss off the least) will want to do the right thing in all cases anyway, so it will disproportionately hurt them

[01:30:00.0000] <hsivonen>
tantek: if you give the user a text field for writing alt text, your program needs to do something when the value entered into the field starts with { and ends with }, since just passing that through would leak user input into writing the wrong meaning into the file format

[01:30:01.0000] <jacobolus>
Hixie: so “do the right thing” here will be interpreted as what? strip out the {} if users add them?

[01:30:02.0000] <Hixie>
jacobolus: or add a space or something

[01:30:03.0000] <jacobolus>
bleh

[01:30:04.0000] <jacobolus>
Hixie: it strikes me that it's still easier to “do the right thing” under the {} system than currently

[01:30:05.0000] <tantek>
why would { } be the wrong meaning?

[01:31:00.0000] <jacobolus>
e.g. what if a user types nothing in the box?

[01:31:01.0000] <tantek>
what if the img was a picture of some curly braces?

[01:31:02.0000] <Hixie>
tantek: {...} has special meaning according to html5 right now (will likely change)

[01:31:03.0000] <Hixie>
jacobolus: like i said, i've no idea what to do

[01:31:04.0000] <tantek>
then escape it

[01:32:00.0000] <tantek>
& has special meaning hence &amp;

[01:32:01.0000] <tantek>
not sure what the big deal is

[01:32:02.0000] <hsivonen>
tantek: right, so escaping means more data munging cases to handle than putting the flag out of band into a separate attribute

[01:32:03.0000] <Hixie>
tantek: there's no defined escaping mechanism for this right now, but even if there was, the big deal is that we are making life harder for the only group who are doing the right thing, which is staggeringly bad language design

[01:32:04.0000] <jacobolus>
so what's the alternate proposal?

[01:32:05.0000] <jacobolus>
make up a new attribute?

[01:33:00.0000] <Hixie>
i have 23 alternative proposals on my list right now

[01:33:01.0000] <Hixie>
they all suck in one way or another

[01:33:02.0000] <jacobolus>
eep

[01:33:03.0000] <jacobolus>
how many with any realistic chances?

[01:33:04.0000] <Hixie>


[01:34:00.0000] <tantek>
what is the URL that documents the need for this { } feature?

[01:34:01.0000] <jacobolus>
tantek: whatwg.com/html5

[01:34:02.0000] <Hixie>
whatwg.org

[01:34:03.0000] <jacobolus>
erm, org

[01:34:04.0000] <Hixie>
but the spec doesn't document the need

[01:35:00.0000] <tantek>
the use case, the background, the group/people that asked for it etc.

[01:35:01.0000] <Hixie>
tantek: http://lists.w3.org/Archives/Public/public-html/2008Aug/0602.html documents some of it

[01:35:02.0000] <Hixie>
search for {...}

[01:35:03.0000] <Hixie>
er

[01:35:04.0000] <Hixie>
actually

[01:35:05.0000] <tantek>
similar problem exists today if someone exj�8s < or " or > into user input meant for an alt attribute right?

[01:35:06.0000] <Hixie>
search for "the tentative decision"

[01:35:07.0000] <Hixie>
no, we have a defined escaping for that and it is wildly implemented already

[01:36:00.0000] <hsivonen>
I really need to get to the bottom of 4 concurrent validations of up to 1 MB pages exhausting 1 GB of RAM

[01:36:01.0000] <hsivonen>
there has to be some silly data structure growth somewhere

[01:36:02.0000] <hsivonen>
and some crazy markup out there on the Web

[01:36:03.0000] <Hixie>
jacobolus: http://damowmow.com/temp/alt lists the proposals so far

[01:37:00.0000] <Hixie>
hsivonen: how is the 1,000,000 validations plan going?

[01:37:01.0000] <hsivonen>
Hixie: I was able to download 93% of the pages successfully. validating them now

[01:37:02.0000] <hsivonen>
Hixie: it seems that either the parser or the validator has a memory leak

[01:37:03.0000] <Hixie>
ah

[01:37:04.0000] <hsivonen>
(I'm reusing the parser and validator objects)

[01:38:00.0000] <jacobolus>
Hixie: just go for B.1

[01:38:01.0000] <hsivonen>
the dmoz validation run hung for unknown reasons

[01:38:02.0000] <Hixie>
jacobolus: disallowing flickr is not a workable solution

[01:39:00.0000] <jacobolus>
(joke)

[01:41:00.0000] <Hixie>
oh many people think B.1 is a good solution

[01:41:01.0000] <jacobolus>
many people are nutgs

[01:41:02.0000] <hsivonen>
Hixie: oh, and real Web content has made a couple of assertions to fire

[01:41:03.0000] <jacobolus>
or nuts even

[01:41:04.0000] <hsivonen>
which is interesting

[01:41:05.0000] <Hixie>
real web content is insane

[01:42:00.0000] <hsivonen>
also, Japanese Web content exposed a bug in my charset decoder driver code that I hadn't seen with UTF-8 content

[01:46:00.0000] <jacobolus>
Hixie: what would the practical effect be if a website allowed user input and the user put in “{picture of grandma on her 80th birthday}” or whatever?

[01:46:01.0000] <jacobolus>
i.e. what would a screen reader do with that content

[01:47:00.0000] <Hixie>
well that would actually be correct alternative text for that image

[01:47:01.0000] <Hixie>
it would cause the UA to attract attention to the fact that there is an image with the caption "picture of..."

[01:47:02.0000] <jacobolus>
that is, if the {} proposal goes through

[01:48:00.0000] <jacobolus>
how is {foo} treated differently from foo

[01:48:01.0000] <Hixie>
right that 's what i was assuming

[01:48:02.0000] <Hixie>
{foo} causes the UA to attract attention to the fact that there is an image

[01:48:03.0000] <Hixie>
foo causes the UA to say "foo" without mentioning the image

[01:48:04.0000] <jacobolus>
I see

[01:48:05.0000] <jacobolus>
that seems like pretty minimal harm then from the vanishingly unlikely false positives

[01:49:00.0000] <Lachy>
Hixie, that doesn't match what existing screen readers doo with the "foo" case. They do say "image: foo"

[01:49:01.0000] <virtuelv>
JohnResig: yt?

[01:49:02.0000] <Hixie>
existing screen readers fail long before they get around to rendering any html

[01:49:03.0000] <Hixie>
jacobolus: i agree

[01:49:04.0000] <Hixie>
jacobolus: won't stop the people who care from being even more careful though

[01:49:05.0000] <jacobolus>
I mean, {} aren’t even meaningful characters in english

[01:50:00.0000] <tantek>
FWIW - as someone who prefers not to see new syntax, capturing such an annotation in another attribute seems preferable to overloading *both* an existing attribute and the use of characters which could mean something themselves.

[01:50:01.0000] <Lachy>
sure, but in the "foo" case, why should they stop notifying the user that there's an image there?

[01:50:02.0000] <jacobolus>
maybe someone will accidentally put them in a description of a mathematical formula or something

[01:51:00.0000] <Hixie>
tantek: another attribute doesn't work because the abuse such an attribute would get would almost certainly cause more damage than the little benefit it would cause

[01:51:01.0000] <Hixie>
tantek: we've already seen that authors are completely incapable of reliably making use of attributes like alt="" or longdesc=""

[01:51:02.0000] <Hixie>
tantek: and this would be an even more subtle attribute

[01:52:00.0000] <Hixie>
Lachy: the whole point of alt="" is that it is replacement text that removes mention of the image

[01:52:01.0000] <jacobolus>
virtuelv: isn't it 4 AM in EDT?

[01:52:02.0000] <virtuelv>
jacobolus: probably

[01:52:03.0000] <virtuelv>
I'll have to catch up when I get to Turin this afternoon then

[01:52:04.0000] <Hixie>
jacobolus: the original complaint re alt={} was latex, fwiw, which could quite legitimately end up with {}

[01:53:00.0000] <virtuelv>
I'm a bit puzzled about one thing he's doing with Sizzle

[01:53:01.0000] <jacobolus>
how does it legitimately end w/ {} ??

[01:54:00.0000] <Hixie>
i mean {somethingsomething}

[01:54:01.0000] <jacobolus>
Hixie: you just mean that's what latex → html converters do currently?

[01:54:02.0000] <Hixie>
yes

[01:54:03.0000] <jacobolus>
probably one particular latex → html converter, which could be trivially changed

[01:56:00.0000] <tantek>
Hixie, your point that the cognitive/incentive barrier to usage of such an attribute makes it useless is a good one, and justifies not only not introducing such an effectively useless attribute, but any such equivalent mechanism, *especially* one that introduces additional syntax.

[01:57:00.0000] <Hixie>
tantek: agreed

[01:58:00.0000] <Hixie>
tantek: so that suggests making alt="" optional for this case, but then we have no way to orient speech synthesis users when they navigate image-by-image

[01:58:01.0000] <tantek>
the lack of a mechanism does not rationally justify introducing an ineffective mechanism

[01:58:02.0000] <Hixie>
tantek: agreed

[01:59:00.0000] <Hixie>
tantek: however the aforementioned case is important enough to certain authors that at least one has said they will always include some alternative text, even if it is not suitable for linear navigation (i.e. reading the page straight through)

[02:00:00.0000] <tantek>
do they do so now on existing pages?

[02:00:01.0000] <Hixie>
those do yeah

[02:00:02.0000] <Hixie>
it makes those pages harder to read if you have images disabled though, e.g. in a text browser

[02:00:03.0000] <Hixie>
which is bad

[02:00:04.0000] <Hixie>
and so not really what we want to require

[02:01:00.0000] <hsivonen>
those authors have *way* fewer photos in their photostream than tantek

[02:01:01.0000] <Hixie>
yup

[02:02:00.0000] <tantek>
below what author/content/sample size threshold do you reach the point of diminishing returns of language design improvements/features?

[02:04:00.0000] <Hixie>
tantek: oh we're way below that already

[02:05:00.0000] <Hixie>
tantek: but i still have to say _something_

[02:05:01.0000] <hsivonen>
Hixie: it seems to me, though, that part of the problem is that you want the result to make sense in Lynx instead of sacrificing Lynx and focusing solely on the AT case

[02:07:00.0000] <Hixie>
as a regular lynx user, i don't think it's unreasonable to have things make sense in lynx

[02:07:01.0000] <Hixie>
it's also a matter of it making sense in graphical browsers with images disabled

[02:08:00.0000] <hsivonen>
Hixie: that means you are letting non-accessibility issues get in the way of accessibility issues

[02:08:01.0000] <Hixie>
this isn't an accessibility issue

[02:08:02.0000] <Hixie>
it's an html issue

[02:09:00.0000] <hsivonen>
right :-)

[02:10:00.0000] <hsivonen>
however, the accessibility alt-need conditions aren't curable. the need to turn off images in a graphical browser is curable (by flat-rate 3G/EDGE mobile subscriptions)

[02:11:00.0000] <Hixie>
yeah well it might be curable in theory, but many parts of africa aren't getting 3G in the next ten years

[02:11:01.0000] <Hixie>
and hell, many parts of america don't have decent EDGE

[02:11:02.0000] <tantek>
another case for alt text being readable is the very common occurrence of web based mail interfaces (e.g. Gmail) presenting HTML email with image loading disabled for security/privacy reasons.

[02:12:00.0000] <Hixie>
hsivonen: and frankly, it's only a problem for AT users when they navigate image-by-image

[02:12:01.0000] <hsivonen>
Hixie: the experts say that's the usual mode of operation

[02:13:00.0000] <Hixie>
hsivonen: sure, so we have to handle it

[02:13:01.0000] <hsivonen>
Hixie: after all, VO, for one, doesn't even have sane UI for reading content in a continuous manner

[02:13:02.0000] <tantek>
hsivonen, the usual mode of operation of web based email interfaces is image loading disabled.

[02:13:03.0000] <Hixie>
anyway

[02:14:00.0000] <hsivonen>
(I'd use VO for non-accessibility things if it had a "read on from here" command)

[02:14:01.0000] <Hixie>
no. idead. what. to. do.

[02:14:02.0000] <Hixie>
hsivonen: just select a bunch of text and use the speech service

[02:14:03.0000] <tantek>
thus the Lynx / graphical browsers with images disabled case is far more common than those two specific use cases might imply

[02:15:00.0000] <Hixie>
doesn't really matter how common all these cases are, they're all common enough that we have to deal with them

[02:20:00.0000] <hsivonen>
it seems to me that the headers/id discussion and the RDFa discussion revolve around a similar issue (except headers/id has more UA legacy than RDFa)

[02:21:00.0000] <Hixie>
they certainly have in common that i don't understand the problems they are trying to solve

[02:21:01.0000] <hsivonen>
that is, if we consider something like "simple things should be easy and complex things should be possible", there's one faction that is willing to make simple cases less easy in order to support complex cases with the same model and another faction that mainly cares about making simple case easy and is willing to cut support for some complex cases

[02:22:00.0000] <annevk2>
hsivonen, turning images off is also common in e.g. Russia

[02:23:00.0000] <annevk>
as bandwidth is ridiculously expensive there still

[02:24:00.0000] <hsivonen>
Hixie: I *think* I understand the problems RDFa is supposed to solve. In the case of ccREL, I just think that some of the problems are non-problem and some problem won't be successfully solved by the proposed solution

[02:24:01.0000] <hsivonen>
Hixie: also, I don't think that the generic problem RDFa solves is a problem whose solutions yield good markup

[02:25:00.0000] <Hixie>
hsivonen: i'm hoping ben adida can actually explain the problem

[02:26:00.0000] <Hixie>
i can come up with problems that i think maybe they are solving

[02:26:01.0000] <hsivonen>
I think RDFa proponents aren't doing their cause any favors when their conference slides take well-known microformats and then show how it can by done "right" with RDFa

[02:26:02.0000] <hsivonen>
and the RDFa version is always more verbose

[02:26:03.0000] <hsivonen>
i.e. crufty

[02:26:04.0000] <annevk>
yeah, that's funny

[02:28:00.0000] <Hixie>
can anyone find the link to joshue's usability videos?

[02:28:01.0000] <tantek>
crufty, more verbose markup examples are nearly always promoted by those who don't actually write markup (publish content on the web) for a living, nor talk with those who do so on a daily basis.

[02:28:02.0000] <Hixie>
i searched youtube but can't figure out the right terms

[02:28:03.0000] <Hixie>
i know they're on youtube somewhere

[02:30:00.0000] <annevk>
does "HTML5 WG header/id test" help?

[02:31:00.0000] <annevk>
it should anyway, http://www.youtube.com/results?search_query=HTML5+WG+header%2Fid+test

[02:31:01.0000] <hsivonen>
also, I have non-techincal reasons why I don't like some aspects of ccREL

[02:32:00.0000] <hsivonen>
the "extensibility" afforded by "requires"/"permits" is effectively catering for license proliferation

[02:33:00.0000] <Hixie>
yeah i'm not a big fan of CC's work to be honest

[02:33:01.0000] <hsivonen>
and in the public licensing space, CC is responsible for more license proliferation than any other single organization

[02:33:02.0000] <Hixie>
they made some interesting licenses, but then they kept on doing things and i'm not sure i like it

[02:33:03.0000] <Hixie>
yeah

[02:34:00.0000] <hsivonen>
Hixie: I like CC-by, CC-by-sa and perhaps CC-by-nc-nd

[02:34:01.0000] <Hixie>
/me notes he works for a group that has done a lot of work to reduce license proliferation in software

[02:34:02.0000] <Hixie>
hsivonen: which is nd again?

[02:34:03.0000] <tantek>
I like the cc-pd and CC0 work

[02:34:04.0000] <Hixie>
annevk: yes, thanks

[02:34:05.0000] <Hixie>
annevk: perfect

[02:34:06.0000] <hsivonen>
Hixie: I seriously dislike it when they promote CC-by-nc and CC-by-nc-sa under the same brand as the Free licenses

[02:35:00.0000] <hsivonen>
Hixie: no derivatives

[02:35:01.0000] <hsivonen>
/me looks up CC0

[02:35:02.0000] <annevk>
/me used Google to figure out the right text then submitted that to Youtube for the rest

[02:36:00.0000] <hsivonen>
/me makes a mental note to refer Europeans to CC0 instead of cc-pd

[02:36:01.0000] <hsivonen>
tantek: thanks

[02:36:02.0000] <Hixie>
annevk: :-)

[02:36:03.0000] <tantek>
by headers/id do you mean the td attributes?

[02:37:00.0000] <hsivonen>
tantek: yes

[02:37:01.0000] <tantek>
there has been more and more use of those in hCalendar microformatted conference schedules

[02:37:02.0000] <tantek>
e.g.

[02:37:03.0000] <tantek>
http://adactio.com/extras/schedules/web2expo-berlin/

[02:38:00.0000] <hsivonen>
I really want CC-by and CC-by-sa to succeed. I've even donated some money to CC, but I really wish they didn't confuse things with NonCommercial and RDF

[02:39:00.0000] <tantek>
a few more Jeremy has created here: http://adactio.com/extras/schedules/

[02:39:01.0000] <Hixie>
man i wish i could find some video of blind users just browsing the web

[02:40:00.0000] <hsivonen>
tantek: it really sucks if authors need to take steps like that to make stuff work with the current state of AT

[02:42:00.0000] <tantek>
actually, the use of headers and id in those examples is not for AT (not directly anyway) but rather to better markup the data in an extractable way with minimal duplication of data

[02:42:01.0000] <tantek>
it turned out that headers/id was designed well enough for this use

[02:42:02.0000] <tantek>
and incidentally, AT benefits result

[02:42:03.0000] <hsivonen>
jgraham: the table inspector crashes with the berlin URL above

[02:43:00.0000] <hsivonen>
tantek: shouldn't data extractors and AT implement the HTML5 header cell association algorithm?

[02:43:01.0000] <tantek>
hsivonen, indeed (to your previous remarks re ccRel). note there are also incrementally advanced license microformat efforts underway: http://microformats.org/wiki/licensing

[02:45:00.0000] <hsivonen>
tantek: interesting to see that the microformats community is on the case. 

[02:46:00.0000] <hsivonen>
tantek: Ben Adida on the WHATWG list said the microformats folks turned them down

[02:46:01.0000] <tantek>
slowly and steadily, but there have been more and more folks interested in incrementally taking steps beyond rel-license, but much more practical than ccRel/RDFa

[02:46:02.0000] <hsivonen>
tantek: did you just turn down some of the more theoretical parts of ccREL?

[02:47:00.0000] <tantek>
microformats folks turned down assumptions that namespaces are necessary etc.

[02:47:01.0000] <tantek>
and asked for research to be done on content/license use in the wild on the web to document use cases

[02:47:02.0000] <tantek>
theoretical proposals are rejected as a matter of course in the microformats community

[02:48:00.0000] <Hixie>
heh

[02:48:01.0000] <Hixie>
i just requested the same thing :-)

[02:48:02.0000] <Hixie>
and henri rejected the namespaces :-)

[02:48:03.0000] <Hixie>
glad to see we're on the same ball :-)

[02:48:04.0000] <hsivonen>
tantek: that's amusing considering how Hixie and I responded on the WHATWG list

[02:49:00.0000] <tantek>
and I do not read the WHATWG list, except when it turns up in search result while doing research.

[02:49:01.0000] <tantek>
(or when given specific URLs to read)

[02:50:00.0000] <tantek>
Hixie, hsivonen, perhaps the subspace neural link is functioning at a subconcious level :)

[02:54:00.0000] <tantek>
speaking of namespaces, I added more documentation citing WHATWG logs and quoting hsivonen and othermaciej to the microformats namespaces page: http://microformats.org/wiki/namespaces

[03:03:00.0000] <tantek>
speaking of which, I noticed that the WHATWG IRC logs are on a server that apparently keeps logs of other channels as well http://krijnhoetmer.nl/irc-logs/

[03:03:01.0000] <hsivonen>
webben: thanks. I had had special cookie settings for del.icio.us.

[03:03:02.0000] <tantek>
the server/machine that was keeping #microformats logs is down and will take some time fix unfortunately. would it be possible to get #microformats added to the list of channels logged at http://krijnhoetmer.nl/irc-logs/   ?

[03:03:03.0000] <hsivonen>
krijnh: ^

[03:05:00.0000] <hsivonen>
tantek: thanks for considering my remarks quotable. 

[03:05:01.0000] <hsivonen>
tantek: this guy was in charge of MSXML and starts with "If there is any one of the W3C's family of XML specifications, that has caused me the most grief, XML Namespaces is probably it.": http://nothing-more.blogspot.com/2004/10/loving-and-hating-xml-namespaces_21.html

[03:06:00.0000] <tantek>
wow

[03:16:00.0000] <hsivonen>
the distinction between xml:lang parsing into {}xml:lang or {http://www.w3.org/XML/1998/namespace}lang depending on content type is already confusing people...

[03:17:00.0000] <hsivonen>
I already suppressed a technically correct but confusing validator warning in this area...

[03:19:00.0000] <zcorpan>
hsivonen: what was the warning?

[03:20:00.0000] <zcorpan>
hsivonen: that xml:lang isn't serializable as xml?

[03:20:01.0000] <hsivonen>
zcorpan: that one

[03:20:02.0000] <zcorpan>
ok

[03:21:00.0000] <zcorpan>
hsivonen: did you get emails about that? or why did you remove it?

[03:23:00.0000] <hsivonen>
zcorpan: I got an email

[03:36:00.0000] <hsivonen>
/me finds memory leaks in Jing

[03:47:00.0000] <hsivonen>
some day I should grab certain Java 1.0 classes from Harmony, put them in a different package, remove synchronization and make Jing use the unsynchronized versions instead

[04:31:00.0000] <Hixie>
how about this

[04:32:00.0000] <Hixie>
we just say that if an image has no alt text, then it must have at least one of the following, and the first of the following that it has must have enough information about hte image to help orient the user if he is navigating by image (and have the ATs read out the relevant information):

[04:32:01.0000] <Hixie>
 - title="" attribute on the <img> itself

[04:32:02.0000] <Hixie>
 - <legend> of the <figure> that contains the <img>

[04:32:03.0000] <Hixie>
 - heading of the section that contains the <img>

[04:46:00.0000] <Lachy>
/me notes that he suggested something like that a long time ago

[04:47:00.0000] <webben>
Hixie: Existing software isn't going to make any use of legend or heading, but requiring title where alt is absent _might_ be a viable approach (in backwards compatibility terms). One would need to check exactly what existing software does.

[04:47:01.0000] <Lachy>
except for the heading of the section one

[04:47:02.0000] <webben>
e.g. I think some screen readers will read title where there is no alt.

[04:47:03.0000] <webben>
I don't know if, when you disable images in your browser, you see title if there is no alt though.

[04:48:00.0000] <Hixie>
just saying "image" is good enough for legacy handling

[04:48:01.0000] <Hixie>
i mean it's not what we want long term

[04:48:02.0000] <Hixie>
but we shouldn't constrain ourselves to only what today's ATs do

[04:48:03.0000] <Hixie>
especially given how horrifically bad today's ATs are

[04:48:04.0000] <webben>
Is saying "image" what they all do.

[04:49:00.0000] <webben>
/me again maintains today's AT is little worse than today's other UAs, but they're both what the spec needs to work iwth.

[04:49:01.0000] <Hixie>
in the case of JAWS, there's about 212 configuration options that control what it does

[04:49:02.0000] <Lachy>
I don't like allowing alt to be omitted just when the section has a heading because there's no way to distinguish between when the image is the primary content of the section, or if it's just a minor part of a larger section

[04:49:03.0000] <Hixie>
ATs today are far, far worse usability wise than today's browsers

[04:49:04.0000] <Hixie>
i mean there's not even a comparison

[04:49:05.0000] <webben>
I know that's your opinion. I just don't think the differences are as radical as you do.

[04:49:06.0000] <Hixie>
it's like comparing the usability of CP/M with the usability of Mac OS X

[04:50:00.0000] <webben>
I also don't think it matters, just like it doesn't really matter how bad I think IE is.

[04:50:01.0000] <webben>
it still needs to be taken into account

[04:50:02.0000] <Hixie>
Lachy: if it's a minor part, then it can't have alt omitted anyway. you can only omit it for key content.

[04:50:03.0000] <Hixie>
webben: like i said, i think saying "image" is good enough

[04:51:00.0000] <Hixie>
webben: saying the title is even better

[04:51:01.0000] <Lachy>
Hixie, no, that's not what I meant.

[04:51:02.0000] <Hixie>
Lachy: do you have a page that shows an example of an img where alt="" could legitimately be omitted without the section header being useful?

[04:51:03.0000] <Hixie>
and where it would have neither title="" nor <figure>?

[04:52:00.0000] <Lachy>
Hixie, no. You seem to have misunderstood me.

[04:52:01.0000] <Hixie>
possibly :-)

[04:53:00.0000] <Lachy>
there are 2 cases that I think need to be distinguished:

[04:53:01.0000] <webben>
Hixie: Could one require describedby attributes explicitly linking the image to the heading?

[04:53:02.0000] <Lachy>
1. The flickr case where there's an image and the section heading relates directly to that image, much like a caption.

[04:54:00.0000] <Hixie>
webben: wouldn't that be redundant?

[04:54:01.0000] <webben>
Hixie: You mean if the image was the sole content of the section?

[04:54:02.0000] <Lachy>
In that case, alt could be omitted according to the rule "heading of the section that contains the <img>"

[04:54:03.0000] <webben>
Hixie: Probably not, given the variety of existing web content.

[04:54:04.0000] <Hixie>
webben: well there's only one section that could possibly be the most relevant one to the image, and that's the header of hte section the image is in

[04:54:05.0000] <webben>
Hixie: Which won't be using HTML5 header semantics.

[04:55:00.0000] <Hixie>
webben: ?

[04:55:01.0000] <webben>
Hixie: With real-world content that's dubious. But in any case being relevant to the header is not the same thing as the header titling the image.

[04:55:02.0000] <Lachy>
2. The case where the image isn't the primary content of the section, but just a small part of it. e.g. <section><h1>Foo</h1><p>some paragraphs...<p>something about this image <img ...><p>more paragraphs</section>

[04:55:03.0000] <webben>
Hixie: Existing content won't be using HTML5 header semantics.

[04:56:00.0000] <Hixie>
Lachy: when would that image ever legitimately not have alt text?

[04:56:01.0000] <Lachy>
in that case, the image should have an alt="whatever" because it's a key part of the content. But there's no way to distinguish that from case #1

[04:56:02.0000] <Hixie>
webben: why not?

[04:56:03.0000] <Hixie>
Lachy: oh well sure, we can't distinguish bogus alt="" text from correct alt="" text either

[04:56:04.0000] <Hixie>
Lachy: we're talking about what's conforming here

[04:56:05.0000] <webben>
Hixie: Because HTML5 clarifies and reinvents HTML4's ambiguous header semantics, which were close to dead-letter in the real world.

[04:57:00.0000] <Lachy>
yeah, but the conformance rules you gave above are ambiguous about it

[04:57:01.0000] <Hixie>
Lachy: do you mean doc conformance or ua confomance?

[04:57:02.0000] <Lachy>
doc conformance

[04:57:03.0000] <Hixie>
webben: so? the html5 rules are based on what pages do

[04:57:04.0000] <Hixie>
Lachy: i don't understand what's ambiguous

[04:57:05.0000] <webben>
Hixie: yes but the conformance rules need to encourage content that AT can actually use, which means helping it distinguish between different scenarios.

[04:58:00.0000] <Hixie>
webben: we can never help ATs distinguish between correct and incorrect cases

[04:58:01.0000] <webben>
of course you can.

[04:58:02.0000] <Hixie>
webben: there are plenty of images with wrong alt="" text

[04:58:03.0000] <webben>
yes, that's a particular case you'd struggle to help with

[04:58:04.0000] <webben>
in this case, one clearly could help

[04:59:00.0000] <Hixie>
webben: there are no cases where authors wouldn't mis-use whatever indicator of conformance we came up with

[04:59:01.0000] <webben>
er ... obviously.

[04:59:02.0000] <Hixie>
well then

[04:59:03.0000] <webben>
authors will make mistakes. with everything. and the spec can't help.

[04:59:04.0000] <Lachy>
Hixie, because your proposed conformance rules stated: "we just say that if an image has no alt text, then it must have [a] heading of the section that contains the <img>". Both cases #1 and #2 I gave fit that condition, even though #2 isn't supposed to be a legitimate case.

[05:00:00.0000] <webben>
guess we don't need a spec?

[05:00:01.0000] <Hixie>
webben: ??

[05:00:02.0000] <webben>
Hixie: Saying including a feature because some authors will misuse it doesn't make any sense to me.

[05:00:03.0000] <Hixie>
Lachy: the actual rule would say that the heading has to be the caption of the image or some such

[05:01:00.0000] <Lachy>
ok, that makes it less ambiguous

[05:01:01.0000] <webben>
Hixie: You'd need to make a case that it would be misused in enough high-profile cases that a UA would be forced to ignore the feature.

[05:01:02.0000] <Hixie>
webben: you are saying that we shouldn't do what i proposed because pages that aren't conforming would fail on this case, right? but isn't that always goign to be the case?

[05:01:03.0000] <webben>
Hixie: Not sure that is what I'm saying.

[05:02:00.0000] <Hixie>
webben: what are you saying then?

[05:02:01.0000] <webben>
Hixie: I'm saying that additional markup would help UAs differentiate between a (assumed) common class of non-conformance and a (my guess) rarer class of non-conformance.

[05:03:00.0000] <webben>
hmm or rather class of conformance

[05:03:01.0000] <webben>
that is distinguish between image amongst other content which is not labelled by the heading but is missing alt (common failure)

[05:04:00.0000] <webben>
and image amongst other content which is labelled by the heading and is missing alt (this would be easy for Flickr to implement, for example)

[05:04:01.0000] <webben>
of course, authors might misuse _all_ the features involved.

[05:05:00.0000] <Hixie>
seems to me that you would be be obtaining this distinction only by increasing the number of cases that were non-conforming in other ways

[05:05:01.0000] <Hixie>
e.g. <img> with alt and said to be described by somehting else, or <img> said to be described by something that isn't there, etc

[05:06:00.0000] <Hixie>
<img> said to be described by something that it isn't described by

[05:06:01.0000] <webben>
I think only the last one is a substantial problem.

[05:07:00.0000] <webben>
the middle case can be trivially caught in validation

[05:07:01.0000] <webben>
and in practice I think the last one isn't going to have a big impact

[05:07:02.0000] <webben>
again, this is markup that's trivial for Flickr to implement

[05:07:03.0000] <Hixie>
i also don't want to introduce an attribute for this one case, and i don't want to rely on aria at this point

[05:07:04.0000] <webben>
and that's the big use case here.

[05:08:00.0000] <webben>
Hixie: I think trying to solve these problems without features is ambitious.

[05:08:01.0000] <Hixie>
i'd rather not solve the problem at all than introduce a new attribute

[05:08:02.0000] <Hixie>
attributes are expensive and this is not a case important enough to warrant that level of investment

[05:09:00.0000] <webben>
Yes, that's not a position that makes sense to me.

[05:10:00.0000] <webben>
In practice, UAs/AT have to try and solve these problems whether there are features for them or not.

[05:10:01.0000] <Hixie>
anyway, even in the case of incorrectly-non-alt'ed images that have no title=, reading the local section title when the user navigates by image to that image isn't a big deal, it still orients the user

[05:10:02.0000] <webben>
and that leads to non-interoperability

[05:10:03.0000] <Hixie>
and we can say that there must only be one such img in that section, to catch errors of that nature

[05:11:00.0000] <webben>
Hixie: it only orients the user if you say something like "in section {Section title}"

[05:11:01.0000] <webben>
otherwise it misleads the user

[05:11:02.0000] <Hixie>
right

[05:12:00.0000] <webben>
it also sucks for extraction, since you can't easily map the section title into metadata about the image.

[05:12:01.0000] <webben>
whereas if you knew that something was intended as a title field, you could dump it into a title field

[05:13:00.0000] <Hixie>
that's not a use case i've previously been worrying about

[05:13:01.0000] <Hixie>
doesn't seem like something users are going to need particularly

[05:13:02.0000] <Hixie>
i mean, when i see someone save an image, they don't save any metadata with it today

[05:13:03.0000] <Hixie>
why would they start?

[05:13:04.0000] <webben>
Hixie: I give what I save titles.

[05:14:00.0000] <webben>
it would be a nice-to-have if it were prepopulated with a sane title.

[05:14:01.0000] <Lachy>
Hixie, given this markup: <section><h1>My Cat</h1><img src="cat.jpg"></section>, what should a screen reader say?

[05:14:02.0000] <webben>
whether in the filename or other metadata that could be found by beagle/spotlight etc.

[05:14:03.0000] <hsivonen>
Hixie: how would you get a reasonable section heading when the user supplies only bitmap data?

[05:15:00.0000] <hsivonen>
(Flickr Uploadr defaults to junk titles by default)

[05:16:00.0000] <hsivonen>
(one of the reasons why I don't use Flickr Uploadr)

[05:16:01.0000] <Hixie>
webben: we are far, far beyond "nice to have"s here. "nice to have" would be alternative text and non-blind users.

[05:16:02.0000] <Hixie>
Lachy: in response to what user command?

[05:16:03.0000] <webben>
Hixie: alternative text and non-blind users involves substantial human effort.

[05:16:04.0000] <Hixie>
hsivonen: if there is no useful data to give, there is no useful data to give.

[05:16:05.0000] <webben>
All I'm suggesting is Flickr add an attribute.

[05:16:06.0000] <Lachy>
Hixie, I don't know the commands. I assumed in just reading the page linearly

[05:17:00.0000] <Hixie>
webben: apparently finding a solution that makes even a small number of people happy here also requires a lot of effort.

[05:17:01.0000] <hsivonen>
hmm. unless I'm mistaken, fixing the memory leaks in Jing and in my parser made the batch validator go *a lot* faster

[05:18:00.0000] <Hixie>
Lachy: reading linearly you would get something like. "Untitled page. One heading. My cat. Image." or some such.

[05:18:01.0000] <webben>
Oh well, at least with the implementation of aria attributes, they'll be some way to explicitly associate images and titles, regardless of the conformance criteria.

[05:18:02.0000] <Lachy>
ok. So it wouldn't say the heading twice? Once when it saw the heading, and then again when it saw the image related to that heading?

[05:18:03.0000] <hsivonen>
I guess that's both the effect of less GC and the effect of a smaller pattern cache in Jing

[05:18:04.0000] <Hixie>
Lachy: the proposal i mentioned only has relevance for when the user navigates on a per-image basis.

[05:19:00.0000] <Lachy>
ok

[05:20:00.0000] <webben>
It might be worth taking someone taking something like WebAnywhere or Fire Vox or NVDA or Orca and hacking it  to see if making the spec's assumptions about how UAs should derive text alternatives actually work for real users. Might be useful for jgraham 's table markup variations too.

[05:21:00.0000] <webben>
dunno if there's an easily hackable text browser.

[05:22:00.0000] <webben>
emacs/w3 perhaps, if you like lisp.

[05:30:00.0000] <annevk>
More namespace confusion: http://lists.w3.org/Archives/Public/public-html/2008Aug/0675.html

[05:31:00.0000] <annevk>
Martin made really weird assumptions there...

[05:32:00.0000] <annevk>
Though since he assumed I didn't get namespaces maybe they did make sense...

[05:49:00.0000] <hsivonen>
so my tokenizer code had a bug with document.write('</SCRIPT\> \n');

[05:50:00.0000] <hsivonen>
yay  for crazy real-world test data

[05:50:01.0000] <Hixie>
what was the bug?

[05:51:00.0000] <hsivonen>
Hixie: it returned to data state upon \ instead of returning to whatever flattened content model flag state it came from

[05:51:01.0000] <webben>
Hixie: If an img without alt is under a heading but also has labelledby, would labelledby take priority over the heading as a label for img during non-linear navigation?

[05:52:00.0000] <Hixie>
hsivonen: fun

[05:52:01.0000] <Hixie>
hsivonen: that's a pretty funny case, i'd never thought of escaping the > as a way to make the tag not an end tag :-)

[05:52:02.0000] <Hixie>
webben: i have no idea, depends on how aria is defined

[05:53:00.0000] <Hixie>
webben: right now aria is a big mess, so it's probably not defined in enough detail to determine an answer

[05:54:00.0000] <hsivonen>
/me suspects that Jing still has some cache hashtable somewhere that just keeps filling up...

[06:09:00.0000] <Hixie>
nn

[06:14:00.0000] <webben>
night Hixie :)

[07:32:00.0000] <hsivonen>
interestingly, javac compiles the tokenizer loop into a bit tighter byte code than ecj

[08:18:00.0000] <kangax>
Does anyone know if gecko 1.9 supports shadowColor/shadowOffsetX/etc.

[08:34:00.0000] <zcorpan>
/me finds three attributes in firefox he didn't know before: HTMLImageElement.naturalHeight and naturalWidth, and HTMLElement.spellcheck

[08:34:01.0000] <zcorpan>
should videoWidth be renamed to naturalWidth?

[08:35:00.0000] <annevk>
videoWidth is more intuitive

[08:36:00.0000] <zcorpan>
safari has naturalWidth too

[08:37:00.0000] <zcorpan>
but consistency is nice

[08:37:01.0000] <zcorpan>
how long has spellcheck been there?

[08:38:00.0000] <zcorpan>
spellcheck doesn't work with contenteditable -- only input and textarea

[08:41:00.0000] <annevk>
Safari has extensions?

[08:41:01.0000] <annevk>
grmbl

[09:11:00.0000] <annevk>
Anyone has a good presentation on <video> in HTML5? :angel:

[09:16:00.0000] <kangax>
Does Gecko 1.9 support canvas shadow-related methods?

[09:20:00.0000] <Philip`>
kangax: No

[09:21:00.0000] <kangax>
strange

[09:21:01.0000] <Philip`>
kangax: (https://bugzilla.mozilla.org/show_bug.cgi?id=310682 has a patch that should get into Firefox 3.1)

[09:21:02.0000] <kangax>
Philip`: Why do they actually exist in context then?

[09:23:00.0000] <kangax>
Philip`: thanks for the link

[09:25:00.0000] <Philip`>
kangax: Because it's much easier to implement the properties than the functionality

[09:25:01.0000] <kangax>
so those are sort of placeholders for now. makes sense.

[09:57:00.0000] <hsivonen>
I wish Maven allowed developers to serve GPG-signed files over HTTP from somewhere

[09:57:01.0000] <hsivonen>
last time I looked, they wanted *their* machine to ssh into the developer's package seeding machine

[09:58:00.0000] <hsivonen>
which is a bit weird as a solution

[09:58:01.0000] <gsnedders>
My tongue is so sore…

[10:39:00.0000] <hsivonen>
I wonder how much overhead python dicts have...

[10:40:00.0000] <hsivonen>
per entry that is

[10:40:01.0000] <hsivonen>
in terms of RAM

[10:46:00.0000] <Xenos>
42

[11:06:00.0000] <jcranmer>
42 yoctobits or 42 yottabits?

[11:07:00.0000] <krijnh>
hsivonen, tantek: i'll start logging #microformats as well

[11:08:00.0000] <jcranmer>
this channel is logged?

[11:08:01.0000] <jcranmer>
/me looks at the camera and waves "Hi mom!"

[11:09:00.0000] <gsnedders>
/me yawns

[11:10:00.0000] <gsnedders>
/me is working on the spec-gen

[11:11:00.0000] <BenMillard>
jcranmer, yes, the Room Topic gives the URL for it: http://krijnhoetmer.nl/irc-logs/

[11:13:00.0000] <jcranmer>
BenMillard: I was being facetious there

[11:17:00.0000] <BenMillard>
d'oh :|

[11:18:00.0000] <BenMillard>
hsivonen, are you going to the W3C meeting in October 2008?

[11:20:00.0000] <hsivonen>
BenMillard: I want to go, but I haven't secured funding yet

[11:20:01.0000] <BenMillard>
(specifically, this: http://www.w3.org/2008/10/TPAC/Overview.html)

[11:20:02.0000] <BenMillard>
hsivonen, I'm in a similar situation at the moment

[11:21:00.0000] <BenMillard>
if Hixie were to attend it would be easier for me to justify my attendence

[11:21:01.0000] <BenMillard>
I know annevk will be there, so I'm checking for other WHATWG/HTMLWG type people

[11:21:02.0000] <BenMillard>
jgraham, would you be going there?

[11:24:00.0000] <annevk>
jgraham will be away

[11:24:01.0000] <annevk>
I happen to know

[11:25:00.0000] <BenMillard>
annevk, thanks

[11:25:01.0000] <BenMillard>
annevk, will you be doing much in relation to HTML5? my intention is to present analysis and foster discussion of the 2008 collections amongst HTMLWG type people there

[11:28:00.0000] <annevk>
it's a bit unclear yet whether the HTML WG is meeting there, not?

[11:28:01.0000] <annevk>
anyway, I'm certainly available for HTML5 discussion

[11:29:00.0000] <annevk>
(the first two days, as well as the first Sunday, I'll be in the CSS WG meeting; no real plan for the rest of the week)

[11:29:01.0000] <BenMillard>
annevk, I saw HTML WG listed on Thursday and Friday: http://www.w3.org/2008/10/TPAC/Schedule.html#Thurs

[11:30:00.0000] <annevk>
ok, good

[11:30:01.0000] <annevk>
i'll be there then :)

[11:36:00.0000] <BenMillard>
Lachy, will you be at the W3C meeting in October 2008?

[11:36:01.0000] <gsnedders>
/me needs to do something about hotels

[11:41:00.0000] <BenMillard>
gsnedders, so you plan to attend?

[11:42:00.0000] <gsnedders>
BenMillard: Well I'm in Lyon till the 17th and I have a flight back from Nice in the 25th, so I need to do something in between :)

[11:42:01.0000] <BenMillard>
gsnedders, it seems W3C have a "block of rooms" reserved at the venue for attendees: http://www.w3.org/2008/10/TPAC/Overview.html#Hotel

[11:43:00.0000] <gsnedders>
BenMillard: At great expense.

[11:43:01.0000] <BenMillard>
ah

[11:43:02.0000] <gsnedders>
BenMillard: There was talk about room sharing, but I'm waiting on primarily smedro to find out what he's doing

[11:43:03.0000] <BenMillard>
yeah, Mozilla suggested that I share a room if I was comfortable with doing so, which I probably am

[11:44:00.0000] <gsnedders>
BenMillard: When are you going to be ther?E

[11:44:01.0000] <gsnedders>
*there

[11:44:02.0000] <gsnedders>
/me is likely going to spend a day or two somewhere in Cannes first

[11:44:03.0000] <BenMillard>
Thursday and Friday would be definite, since that has specific HTMLWG meetings

[11:46:00.0000] <BenMillard>
I suggested attending Monday and Tuesday for WCAG stuff, explaining HTML5's research-based approach to accessibility features to people I meet and so forth

[11:47:00.0000] <gsnedders>
Web Applications WG — Group 2 would be interesting, but member confidential :(

[11:48:00.0000] <BenMillard>
I figure the longer I'm there, the more I can do to build bridges between HTML5 and other groups face-to-face

[11:48:01.0000] <BenMillard>
not sure if Mozilla Foundation will buy that, though :P

[11:48:02.0000] <gsnedders>
:P

[11:48:03.0000] <gsnedders>
/me has the advantage of being truly independent :P

[11:49:00.0000] <BenMillard>
would you be there on Thursday and Friday?

[11:49:01.0000] <BenMillard>
hey, one of us should take a PS2 and some Gran Turismo games :)

[11:49:02.0000] <gsnedders>
I'll certainly be there Thurs/Fri

[11:49:03.0000] <gsnedders>
My PS2 crashes disks now, sometimes :(

[11:50:00.0000] <gsnedders>
*scratches

[11:50:01.0000] <gsnedders>
GT4 is scratched and unplayable

[11:50:02.0000] <gsnedders>
Well, you can play it, you just need to restart the console after every race :D

[11:50:03.0000] <gsnedders>
And nothing gets saved

[11:50:04.0000] <BenMillard>
my PS-Two is tiny; the discs are in good running order; memory cards etc work fine and have good car selection.

[11:51:00.0000] <gsnedders>
BenMillard: NTSC/PAL?

[11:51:01.0000] <BenMillard>
PAL...

[11:51:02.0000] <gsnedders>
Worth checking :)

[11:51:03.0000] <BenMillard>
yeah, will that work in Europe?

[11:51:04.0000] <gsnedders>
Yeah

[11:51:05.0000] <BenMillard>
is that "yes, it will work"? :)

[11:52:00.0000] <gsnedders>
Yes.

[11:52:01.0000] <gsnedders>
France still in part uses SECAM, but everything in all SECAM places supports PAL as well

[11:53:00.0000] <gsnedders>
/me can bring a game or two and memories cards with saves

[11:53:01.0000] <gsnedders>
heh. This is so useful for the future of the web.

[11:53:02.0000] <BenMillard>
cool, so how about we share a room for Thursday and Friday? I'd arrive on Wedesday since meetings start so early on Thursday

[11:54:00.0000] <BenMillard>
so maybe Wednesday-Friday?

[11:54:01.0000] <gsnedders>
BenMillard: you not going to the TP itself, then?

[11:54:02.0000] <BenMillard>
gsnedders, the more I think about it the less I can justify it, given that it's someone else's money I'll be going on

[11:55:00.0000] <gsnedders>
I'll see if smedro has made his mind up about when he'll be there yet, because if he's there for the entire week I'll probably share a room with him

[11:55:01.0000] <gsnedders>
s/smedro/smedero/g

[11:59:00.0000] <BenMillard>
gsnedders, the form has a "Number of persons" field so I imagine there are 3-person rooms? http://www.w3.org/2008/03/TPAC2008-hotelform.html

[11:59:01.0000] <BenMillard>
(I'm kind of a noob at travelling so I'd find it comforting to be part of a group)

[12:05:00.0000] <hsivonen>
sigh. the transcript discussion

[12:06:00.0000] <hsivonen>
the right way to deal with users not hearing the audio track of video is captioning

[12:15:00.0000] <webben>
hsivonen: that's not useful for deafblind people.

[12:16:00.0000] <webben>
if you had a compound format producing both audio descriptions and captioning, you could extract a useable transcript from that though.

[12:16:01.0000] <hsivonen>
webben: if the captioning is text (not bitmap), theoretically you could detach it from the temporal axis and feed it to a braille renderer

[12:16:02.0000] <webben>
hsivonen: captioning doesn't tell you what's being seen.

[12:17:00.0000] <hsivonen>
webben: however, it would be dumb to make things suck for deaf people just because the same solution doesn't work for a significantly less common combination of disabilities

[12:17:01.0000] <webben>
hsivonen: is someone suggesting _not_ supplying captioning?

[12:19:00.0000] <hsivonen>
webben: suggesting that video accessibility solution would be transcript pretty much implies that the person didn't consider captioning, but I could be wrong.

[12:19:01.0000] <webben>
hsivonen: I guess the assumption there (not entirely right) is that everyone could make sense of the transcript.

[12:19:02.0000] <webben>
it's at least true that deaf people could use a transcript too.

[12:20:00.0000] <webben>
I'd strongly disagree that one shouldn't also provide captioning though. :)

[12:20:01.0000] <webben>
since that's likely to be better for that user group, as you say.

[12:21:00.0000] <hsivonen>
also, transcript as fallback sucks, because transcripts are also useful to people who see and hear just fine

[12:21:01.0000] <webben>
hsivonen: I think this is a case where being able to demark an area of the page or a URL as "equivalent" to the video is preferable to merely falling back.

[12:22:00.0000] <webben>
<transcription for=

[12:22:01.0000] <webben>
(or whatever)

[12:22:02.0000] <webben>
transcription="{uri}"

[12:23:00.0000] <hsivonen>
<a href="{uri}">Transcript</a>

[12:24:00.0000] <webben>
hsivonen: That suffers from the usual problem of having to look around the page to find a transcript, rather than being able to go straight to it via the video

[12:30:00.0000] <webben>
/me wonders why the WebKit implementation of alt hides most of the alt text.

[12:30:01.0000] <webben>
(when images are disabled)

[12:31:00.0000] <webben>
hmm maybe that's OmniWeb, Safari won't even show the alt

[12:31:01.0000] <tantek>
thanks krijnh. I presume the logs will eventually be available at http://krijnhoetmer.nl/irc-logs/microformats ?

[12:32:00.0000] <gsnedders>
BenMillard: Where are you from anyway?

[12:32:01.0000] <webben>
iCab shows an empty square ... "useful"

[12:35:00.0000] <webben>
hmm ... current WebKit shows nothing when you untick Display images when the page opens

[12:35:01.0000] <BenMillard>
gsnedders, Fleet, Hampshire, England: http://projectcerbera.com/me/

[12:36:00.0000] <webben>
happens in Safari 3.1 too

[12:37:00.0000] <BenMillard>
webben, if the link is right next to the <video> then it's fine

[12:37:01.0000] <BenMillard>
putting relevant links in relevant positions is a usability issue for authors to judge, rather than a format limitation, imho

[12:37:02.0000] <webben>
BenMillard: it's not ideal f you're using a navigation control to jump from video to video.

[12:38:00.0000] <webben>
*if

[12:38:01.0000] <webben>
at which point "hunt around the page" includes stop processing by type of thing and start linearly reading the page.

[12:39:00.0000] <BenMillard>
webben, if a user interacts with a page in a restrictive ways then a restrictive experience is what they wanted, I imagine

[12:39:01.0000] <BenMillard>
a "read from here" command should make it easy as pie to find a link immediately after a <video>

[12:39:02.0000] <BenMillard>
(or any other element, for that matter)

[12:39:03.0000] <webben>
BenMillard: it's an unnecessary extra step

[12:40:00.0000] <webben>
and doubtless doomed to disappointment in most cases

[12:40:01.0000] <BenMillard>
that doesn't tally with my experience of observing users (which is far from exhaustive, though)

[12:41:00.0000] <webben>
which? that's it's unnecessary? or that it's disappointing?

[12:41:01.0000] <BenMillard>
both

[12:41:02.0000] <BenMillard>
webben, specialist features for interacting in very specific ways tend to be what cause dissapointment since pages usually get them wrong; whereas simple proximity is usually convenient and intuitive

[12:41:03.0000] <gsnedders>
BenMillard: Speaking of about pages, I need to write a new one :)

[12:42:00.0000] <BenMillard>
(convenient and intuitive for both users and authors)

[12:43:00.0000] <webben>
BenMillard: I'm not sure what I've seen of these things agrees with that. But that could be logically true without contradicting "unnecessary" or "disappointing" when you don't find a transcript.

[12:44:00.0000] <BenMillard>
webben, if a transcript is not available to a user who would like one, their dissapointment is with the lack of facilities provided by the author rather than the lack of facilities provided by the format

[12:44:01.0000] <webben>
BenMillard: What I've seen is screen reader users, at least, explicitly requesting the ability to jump between different sorts of objects _and_ the ability to jump to content _and_ the ability to read from here.

[12:44:02.0000] <webben>
BenMillard: Yes. But the format is forcing them to search for it.

[12:45:00.0000] <webben>
i.e. making the entire process more cumbersome

[12:45:01.0000] <BenMillard>
webben, if the link is immediately after the <video> (as it would be in a usably designed page) then they would find it immediately anyway :)

[12:45:02.0000] <hsivonen>
hmm. if I try to validate a million pages and log URL and message for later processing, the file will be excessive in size :-(

[12:46:00.0000] <hsivonen>
I wonder if gzip will do the right magic for me since there will be a lot of common substrings

[12:46:01.0000] <webben>
BenMillard: if I had a catalog of multiple videos in a page, I don't really want to have to jump to each video, and then look for a transcript, I just want to jump to each video and be told if there's a transcript.

[12:46:02.0000] <webben>
it's still an extra step

[12:47:00.0000] <webben>
because there's no way a UA can do that for you

[12:47:01.0000] <webben>
heck, if there was explicit markup, one could just pull up the videos in the formats acceptable to you

[12:48:00.0000] <webben>
sod inspecting each one individually :)

[12:49:00.0000] <BenMillard>
webben, you may be overthinking the problem...equally I may be underthinking it. either way, the amount of web content I've studied shows re-using simple structures in simple ways (like an <a href> near to where you need it) is more reliably authored (and therefore more robust) than specialist facilities in the language to do nearly the same thing

[12:50:00.0000] <webben>
I'm not sure I care about it being reliably authored. I'd want to see that markup in major sources of videos.

[12:50:01.0000] <webben>
That is to say, if it makes YouTube better that's a huge win.

[12:51:00.0000] <BenMillard>
if a feature is not reliably authored, nobody wins because it won't work

[12:51:01.0000] <BenMillard>
so I put a higher value on reliability, but that might just be me

[12:51:02.0000] <webben>
it not working on Joe Blogg's site is a lot less important than it working on YouTube.

[12:52:00.0000] <webben>
I suspect Google's engineers are capable of generating such markup (indeed, vastly more complicated markup, if they wanted to).

[12:52:01.0000] <BenMillard>
in contrast, a simpler feature which works everywhere provides a more stable experience for users (even if it takes one extra key press and half a second more aural output each time)

[12:53:00.0000] <BenMillard>
webben, ultimately it's Hixie who decides where the balance is, I suppose :)

[12:54:00.0000] <webben>
I'd certainly agree that simple features are better. However, having explicit markup to associate with transcripts can be an enhancement to the simpler markup.

[12:55:00.0000] <webben>
it's not an either/or situation

[12:55:01.0000] <webben>
it's a bit like microformats being built on top of simpler structures.

[12:57:00.0000] <BenMillard>
indeed. but if the simple feature is adequate for user needs (which would take user testing to establish one way or another, I imagine) then the explicit markup is just a side-show; complicating the authoring process and adding one more way for things to go wrong

[12:57:01.0000] <BenMillard>
for example, consider automatic association of <th> and <td> versus <td headers> for every data cell and <th id> for every header cell

[12:58:00.0000] <webben>
BenMillard: I think I'd agree depending on what adequate means.

[12:58:01.0000] <webben>
I'd prefer people to have a good web experience rather than a merely adequate one.

[12:59:00.0000] <BenMillard>
I'd define "adequate" as something like "task can be completed by users in a timely manner"

[12:59:01.0000] <BenMillard>
webben, but I expect the user testing people have a better idea of it that I

[12:59:02.0000] <webben>
that sounds reasonable

[13:00:00.0000] <webben>
it's the timeliness we have differing suspicions about in this case.

[13:00:01.0000] <BenMillard>
indeed :)

[13:00:02.0000] <BenMillard>
webben, will you be at the W3C meeting in October 2008?

[13:01:00.0000] <webben>
/me works for a W3C organization and therefore has no official W3C involvement.

[13:01:01.0000] <webben>
*W3C member organization

[13:01:02.0000] <BenMillard>
is that a no, then?

[13:01:03.0000] <webben>
I don't know anything about it, but I doubt it.

[13:02:00.0000] <BenMillard>
webben, oh well, perhaps we meet another time. the page about it is here: http://www.w3.org/2008/10/TPAC/Overview.html

[13:02:01.0000] <webben>
ah, thanks :)

[13:03:00.0000] <BenMillard>
I'm hoping to be there from Wednesday to Friday, funding permitting

[13:04:00.0000] <webben>
cool

[13:04:01.0000] <webben>
well, someone from my company should be there somewhere, at any rate

[13:05:00.0000] <gsnedders>
/me notes the docs for the spec-gen are crazy

[13:05:01.0000] <webben>
it's in a casino - is anyone actually going to do any work there ;)

[13:06:00.0000] <webben>
looks like a great locale

[13:12:00.0000] <Dashiva>
Firefox 3 supports 'content: url(data:...)', doesn't it?

[13:21:00.0000] <webben>
Dashiva: Does it?

[13:22:00.0000] <webben>
Dashiva: it doesn't support content:url(.....png);

[13:23:00.0000] <Dashiva>
I found the bug. They only support the css2 version, which is really crippled.

[13:24:00.0000] <webben>
/me wishes support for content:url was better since it would mean we could have significantly less failsome image replacement techniques.

[13:25:00.0000] <Dashiva>
Spec's been WD for five years now

[13:25:01.0000] <Dashiva>
And it's all Hixie's fault :P

[13:26:00.0000] <webben>
That Hixie. Tsk tsk. It's not like he's got any other specs to edit ;)

[13:29:00.0000] <BenMillard>
webben, <img src alt> does the image replacement aspect of image replace perfectly

[13:29:01.0000] <BenMillard>
*image replacement

[13:30:00.0000] <webben>
not perfectly if you want to be able to change the skin

[13:30:01.0000] <webben>
also not perfectly if you want to use css spriting

[13:30:02.0000] <webben>
but yes, I do tend to recommend (actively agitate for?) img over CSS image replacement

[13:31:00.0000] <BenMillard>
webben, skinning systems in forums I've used (phpBB, Invision, Ikonboard and vBulletin) will change the src when the user changes their theme

[13:31:01.0000] <BenMillard>
theme selection is server-side rather than client-side in those systems to achieve this

[13:31:02.0000] <webben>
BenMillard: Yeah, you've got to complicate the skinning system to include changing HTML.

[13:32:00.0000] <webben>
I'd also just prefer to use img for content images, not style.

[13:32:01.0000] <BenMillard>
if you are doing image *replacement* then you are doing it to some content

[13:32:02.0000] <BenMillard>
as such, the image is content when doing image replacement

[13:32:03.0000] <webben>
BenMillard: Doesn't help people who want to skin things themselves though.

[13:33:00.0000] <webben>
BenMillard: Only accidentally.

[13:33:01.0000] <Dashiva>
Nor people who write userjs

[13:33:02.0000] <Dashiva>
(or usercss, rather)

[13:33:03.0000] <webben>
Dashiva: that's what I mean

[13:33:04.0000] <webben>
well, or other ways of skinning

[13:33:05.0000] <BenMillard>
Dashiva, when will a user write CSS to change the image being replaced by an author's image replacement CSS?

[13:33:06.0000] <webben>
BenMillard: That is, if CSS supported it, we'd use CSS and the img would be text, not an img.

[13:34:00.0000] <Dashiva>
BenMillard: The author isn't replacing anything, the user is

[13:34:01.0000] <webben>
BenMillard: If I found it hard to read text in images, I'd probably just block content:

[13:34:02.0000] <webben>
or at least content:url with certain mime types

[13:35:00.0000] <BenMillard>
webben, that case is handled by disabling images to show their alt text, afaict?

[13:35:01.0000] <webben>
yeah, but that's a rubbish user experience

[13:35:02.0000] <webben>
I don't want to prevent myself see photos

[13:35:03.0000] <webben>
*seeing

[13:35:04.0000] <Dashiva>
Like the very reason I started talking about this: I want to make all blank gifs on a certain page become a different gif. So I use content: url(data:...)

[13:35:05.0000] <BenMillard>
Dashiva, sorry I worded that badly. I meant: "When will a user write CSS to change the image being applied by an author's image replacement CSS?

[14:03:00.0000] <Hixie>
/me writes the first and probably only ever e-mail in which he says RDF isn't the problem

[14:03:01.0000] <Hixie>
element.spellcheck is something that i specced out which will likely end up in html5 if the experimental implementations come back positive

[14:03:02.0000] <Hixie>
i think firefox had it first

[14:19:00.0000] <BenMillard>
Hixie, have you seen https://bugzilla.mozilla.org/show_bug.cgi?id=441445 ? I suggested Aaron contact you directly about processing table headers in Firefox.

[14:20:00.0000] <BenMillard>
I don't know the Andrew Downie dude but I've met Marco Zehe and Aaron

[14:21:00.0000] <BenMillard>
(interestingly, the attached table is regular so it doesn't really need headers+id, afaict)

[14:27:00.0000] <annevk>
http://www.longnow.org/about/ so five digit years to solve the deca-millennium bug, what about the hecto-millennium bug?

[14:27:01.0000] <annevk>
it doesn't make much sense at all

[14:34:00.0000] <webben>
ambitious engineering project though

[14:34:01.0000] <webben>
getting anything working for 10,000 years

[14:35:00.0000] <webben>
/me goes off to "Mind mythic depth"

[14:37:00.0000] <annevk>
as long as the project doesn't require everyone to relabel xxxx as 0xxxx...

[14:37:01.0000] <BenMillard>
annevk is talking specifically about: "The Long Now Foundation uses five digit dates, the extra zero is to solve the deca-millennium bug which will come into effect in about 8,000 years."

[14:38:00.0000] <Xenos>
Let's just hope that we get cryogenics in time to deep-freeze some COBOL programmers, or they'll have problems fixing them timer bugs near y10k :p

[14:38:01.0000] <Dashiva>
Yeah, because the Y2K problem happened because we didn't have enough decimal digits to write 2000...

[14:38:02.0000] <BenMillard>
variable-length year numbers naturally solve the problem anyway

[14:39:00.0000] <annevk>
I think HTML5 does variable-width year at some places, and fixed width at some others...

[14:40:00.0000] <BenMillard>
1996 isn't going to mean 11996 after time has progressed 10,000 years if people write 11996 for that era

[14:40:01.0000] <annevk>
ouch, http://blog.digitalbazaar.com/2008/08/23/html5-rdfa-and-microformats/ is six pages :/

[14:41:00.0000] <BenMillard>
just like 996 didn't start meaning 1996 when we reached 1996

[14:41:01.0000] <annevk>
BenMillard, yeah, like I said, it's silly

[14:41:02.0000] <tantek>
indeed it is. I wonder how long leading 0s have been unnecessary in numerical notation.

[14:41:03.0000] <BenMillard>
annevk, I'm just illustrating how right you are about that :)

[14:41:04.0000] <annevk>
but maybe they mean it in the Dashiva way, in which case I'd argue for solving that by something better than just adding space for an extra digit

[14:42:00.0000] <Dashiva>
Like hexadecimal years :D

[14:44:00.0000] <tantek>
0x7D8 - still takes 5 characters ;)

[14:44:01.0000] <BenMillard>
annevk, each page seems very short except for page 3...even that one isn't very long

[14:44:02.0000] <Dashiva>
But the 0x removes the whole ambiguity problem

[14:45:00.0000] <BenMillard>
annevk, for me, the entry would be more readable (and less daunting) on a single page. I think pagination is a mistake in this case.

[14:46:00.0000] <BenMillard>
I do like the theme of that site, though. :)

[14:46:01.0000] <BenMillard>
(visual theme, I mean)

[14:46:02.0000] <tantek>
it's too bad people view posts on email lists as work/progress

[14:47:00.0000] <tantek>
in my experience, quantity of posts on email lists usually mean noise, inexperience, not RTFM, repeating discussions etc.

[14:47:01.0000] <BenMillard>
tantek, repeating discussions seems a common cause on Public-HTML

[14:48:00.0000] <annevk>
that guy makes a lot of assumptions too about who's part of the WHATWG and who's not

[14:48:01.0000] <annevk>
annoying

[14:48:02.0000] <annevk>
and they assume they are smarter "We started out where most in the HTML5 community are, we thought Microformats would be the solution to the semantic web…"

[14:49:00.0000] <tantek>
microformats never claimed to be a/the "solution to the semantic web"

[14:49:01.0000] <tantek>
why would you claim to be something that might be impossible?

[14:49:02.0000] <annevk>
:)

[14:50:00.0000] <tantek>
I long gave up on trying to "keep up" with people that have more time than me on email lists.

[14:51:00.0000] <tantek>
It's a futile effort, and doing so just succumbs to their (perhaps unintentional) denial of productivity attack (dopa) http://tantek.pbwiki.com/DenialOfProductivityAttack

[14:52:00.0000] <Hixie>
"it's too bad people view posts on email lists as work/progress" <-- couldn't agree more

[14:52:01.0000] <Hixie>
BenMillard: not sure what to say on that bug, but haven't heard from aaron yet

[14:52:02.0000] <Hixie>
ok meeting bbiab

[14:53:00.0000] <BenMillard>
Hixie, looked like they were thinking of implementing header association as defined in HTML5

[14:53:01.0000] <hsivonen>
BenMillard: I think it would be great if Gecko implemented the association algorithm

[14:53:02.0000] <annevk>
I'm giving up on replying to non-technical e-mails in <video> threads on public-html

[14:53:03.0000] <annevk>
(well, I will try)

[14:55:00.0000] <BenMillard>
hsivonen, yes me too!

[14:55:01.0000] <BenMillard>
hsivonen, I'd prefer if HTML5's algorithm was nearer to the Smart Headers algorithm in allowing headers to point at <td> and letting headers associate with headers...maybe a real implementation will cast more light on those aspects.

[14:56:00.0000] <annevk>
http://5090.fawm.org/songs.php?id=2303 is funny

[14:57:00.0000] <hsivonen>
BenMillard: I don't know which algorithm is better, but I'm confident that it has a better chance of working in practice if it is up to Gecko developer rather than up to AT developers

[14:59:00.0000] <BenMillard>
hsivonen, well, there are far fewer AT-compatible browsers than there are ATs

[15:00:00.0000] <BenMillard>
so doing it in the browser means fewer implementations are needed to provide the feature to the same number of users

[15:00:01.0000] <Dashiva>
hsivonen: At least the Gecko developers try to develop :)

[15:01:00.0000] <hsivonen>
annevk: the six-page blog format is weird

[15:01:01.0000] <annevk>
uhuh

[15:02:00.0000] <BenMillard>
hsivonen, I thought the same thing :)

[15:02:01.0000] <annevk>
http://www.w3.org/2008/08/20-xhtml-minutes.html#item04 "SP: trying to reflect reality, so it is ok to deliver XHTML using text/html because IT JUST WORKS"

[15:02:02.0000] <annevk>
right

[15:03:00.0000] <annevk>
meanwhile on planet earth...

[15:09:00.0000] <hsivonen>
the digitalbazaar blog post reminds me of http://webbackplane.com/standard

[15:10:00.0000] <BenMillard>
Hixie, turns out my message to Aaronlev about that bug didn't reach him, so I've resent to a different account he has

[15:15:00.0000] <webben>
aaronlev is actually in channel, if that helps

[15:16:00.0000] <BenMillard>
webben, he's in #accessibility which is where I'm talking with him, too

[15:17:00.0000] <webben>
oh, so you are :)

[15:21:00.0000] <hsivonen>
http://esw.w3.org/topic/PF/XTech/HTML5/AccesskeyRequirements

[15:21:01.0000] <krijnh>
tantek: jep, indeed (sorry, my train and wifi went away)

[15:22:00.0000] <annevk>
today we thank him, tomorrow we "curse" him for all the pr0n that replaced our logs

[15:22:01.0000] <krijnh>
:)

[15:23:00.0000] <krijnh>
tantek: I'll probably have time to fix the pages on Wednesday, perhaps tomorrow

[15:23:01.0000] <Dashiva>
SM: putting onBlur on SCRIPT doesn't mean anything

[15:28:00.0000] <hsivonen>
/me wonders how many email iterations it will take to tease out that SQL storage and RDFa are different, because at least two of {Mozilla, WebKit, Opera, IE, Gears} teams wanted to ship a JS API for sqlite

[15:29:00.0000] <BenMillard>
hsivonen, I don't know if there's an actual desire for header association to be implemented in Gecko by following HTML5...it's just what I read between the lines of comments near the end of https://bugzilla.mozilla.org/show_bug.cgi?id=441445

[15:29:01.0000] <BenMillard>
I don't want to get in trouble or raise false hopes :)

[15:30:00.0000] <gsnedders>
hsivonen: 42n, where n approaches ∞.

[15:33:00.0000] <gsnedders>
Lachy: I realized that if I do ever write RSS5, nobody is actually going to publish it as a standard, so I may as well put things like, "The editor recommends eating pasta while implementing this specification; however, Lachlan Hunt recommends eating a hamburger instead." in it

[15:34:00.0000] <Dashiva>
Can't you make Atom5 instead?

[15:35:00.0000] <gsnedders>
Dashiva: But it defines RSS, and not Atom!

[15:35:01.0000] <gsnedders>
Atom is in a far better state than RSS!

[15:36:00.0000] <Dashiva>
Exactly, so let's just let RSS wander off and die over the next twenty years :)

[15:37:00.0000] <gsnedders>
Dashiva: People still need to implement it, like HTML :P

[15:37:01.0000] <Dashiva>
We have the universal feed parser :P

[15:37:02.0000] <gsnedders>
Only in Python :P

[15:37:03.0000] <gsnedders>
But hey, SimplePie mostly works in PHP :P

[15:38:00.0000] <Dashiva>
One platform, one language, one runtime

[15:38:01.0000] <gsnedders>
Not one prize, one goal, one golden glance of what should be?

[15:39:00.0000] <Dashiva>
Glances aren't all that useful for binding things in darkness

[15:43:00.0000] <gsnedders>
/me wonders whether he could get into Stanford

[15:43:01.0000] <annevk>
W3C Validator does HTML5 too, albeit with the same code as Validator.nu

[15:43:02.0000] <annevk>
(and only the beta)

[15:43:03.0000] <gsnedders>
annevk: Since when!?

[15:44:00.0000] <hsivonen>
gsnedders: since a few minutes ago

[15:44:01.0000] <gsnedders>
heh.

[15:45:00.0000] <annevk>
#whatwg is like a very live feed reader now and then, maybe we should feed every single line to twitter :)

[15:46:00.0000] <Xenos>
twit-rooms :p

[15:46:01.0000] <gsnedders>
hah.

[15:46:02.0000] <annevk>
some IRC thingies have this option to do "blog: hahaha", maybe something to look into (<< krijnh)

[15:46:03.0000] <gsnedders>
High volume tweeting.

[15:46:04.0000] <gsnedders>
blog: gsnedders is God.

[15:46:05.0000] <hsivonen>
it seems to me that there are four big runtimes: C, JVM, CLR and The Browser 

[15:46:06.0000] <annevk>
CLR?

[15:47:00.0000] <hsivonen>
JavaScript and Java run on all four

[15:47:01.0000] <gsnedders>
annevk: .NET

[15:47:02.0000] <hsivonen>
are there others that run on all four?

[15:47:03.0000] <annevk>
Python, for some browsers

[15:48:00.0000] <hsivonen>
annevk: but there isn't anything like GWT for Python, is there?

[15:48:01.0000] <annevk>
I don't think so, though they might be planning it I suppose

[15:49:00.0000] <hsivonen>
(when I said Java runs on The Browser runtime, I meant via GWT)

[15:51:00.0000] <annevk>
/me was thinking "milennium, meganium"; as it happens Meganium has 139.000 results on Google; it's a Pokémon

[15:51:01.0000] <hsivonen>
/me wants to get the Validator.nu HTML Parser running on all the four runtimes

[15:51:02.0000] <hsivonen>
2 out of 4 covered already

[15:52:00.0000] <hsivonen>
(chances are that getting the core to run on CLR would be very easy)

[15:53:00.0000] <roc>
IKVM.NET is pretty good, it should just run your Java code

[15:54:00.0000] <hsivonen>
roc: yeah, I was thinking that IKVM.NET would probably compile the core OK

[15:54:01.0000] <annevk>
(which is based on geranium)

[15:55:00.0000] <hsivonen>
roc: a proper package would need a .Net-compatible tree builder and a port of all the charset stuff

[15:55:01.0000] <roc>
I see

[15:55:02.0000] <hsivonen>
(or for the charset stuff, compiling ICU4J for CLR while at it)

[16:02:00.0000] <annevk>
/me didn't know about http://en.wikipedia.org/wiki/Long_and_short_scales (I thought it was a US-UK thing)

[16:02:01.0000] <annevk>
Netherlands uses the long scale afaict

[16:02:02.0000] <gsnedders>
annevk: No, it isn't US v. UK at all

[16:03:00.0000] <hsivonen>
/me thought it was an English vs. other European languages thing

[16:03:01.0000] <annevk>
gsnedders, well, "not at all" is a bit of an overstatement, according to Wikipedia at least it was known as such for quite a while

[16:04:00.0000] <gsnedders>
It goes back further than that, because the US used to long scale for a long time

[16:04:01.0000] <gsnedders>
For most of its existence, in fact

[16:18:00.0000] <annevk>
/me experiments with text messaging through transactions

[16:19:00.0000] <annevk>
(ie, you give someone EUR 0,01, the minimum transaction fee, and write some message to go with it)

[16:24:00.0000] <Dashiva>
Isn't the latency high?

[16:26:00.0000] <annevk>
yeah, but it's funny

[16:26:01.0000] <annevk>
/me wonders whem spammers will fill up his back account

[16:26:02.0000] <annevk>
when, even

[16:27:00.0000] <annevk>
anyway, bedtime

[16:27:01.0000] <annevk>
/me -> bed

[16:49:00.0000] <Dashiva>
So on one hand, we're supposed to submit to PFWG, WAI, WCAG and who knows what, but on the other hand we're supposed to reinvent accessible video from scratch?

[16:49:01.0000] <gsnedders>
Yeah.

[16:50:00.0000] <gsnedders>
Why not?

[16:50:01.0000] <gsnedders>
Doesn't it make perfect sense?

[16:50:02.0000] <Dashiva>
Yeah, if your goal is to run the spec into the ground

[16:50:03.0000] <gsnedders>
Everything should be in the HTML WG's spec and written by PFWG, WAI, and WCAG WGs

[16:51:00.0000] <hsivonen>
Dashiva: the probability that someone asks the HTML WG to submit to the Timed Text WG of the W3C is close to 1

[16:52:00.0000] <Dashiva>
hsivonen: Okay, but that still doesn't explain why we can't just use an accessible format 

[16:52:01.0000] <webben>
Dashiva: What one's that?

[16:53:00.0000] <Dashiva>
A wrapper format that contains separate streams or whatnot for audio, video and fifteen types of text

[16:53:01.0000] <webben>
Dashiva: no, I mean which format?

[16:54:00.0000] <webben>
or formats plural, if there's lots of em

[16:54:01.0000] <Dashiva>
I bet SMIL can do it, the monster that it is

[16:54:02.0000] <Philip`>
You could have an ISO rip of a DVD, with subtitles and everything

[16:54:03.0000] <Dashiva>
mkv can do it

[16:55:00.0000] <webben>
are DVD captions scalable?

[16:55:01.0000] <webben>
(visually)

[16:55:02.0000] <Philip`>
<video src="file:///d:/"> would be a neat way to implement a DVD player

[16:55:03.0000] <Dashiva>
Philip`: Nice match for the 'free cup holder' :)

[16:55:04.0000] <webben>
Does <video> support SMIL?

[16:56:00.0000] <Philip`>
webben: They're stored as text and not graphics, so the player can scale them however it wants, as far as I'm aware, if that's what you're referring to

[16:56:01.0000] <webben>
Philip`: ah okay.

[16:56:02.0000] <Dashiva>
webben: The video element supports whatever the browser (or one of its plugin backends) supports

[16:56:03.0000] <Philip`>
(which is necessary so tha you can easily switch between different subtitle tracks)

[16:57:00.0000] <webben>
Philip`: that could just be overlays (like Photoshop layers).

[16:57:01.0000] <gsnedders>
Philip`: You don't need an empty authority at all

[16:58:00.0000] <webben>
/me wonders how much or little of SMIL Real, Windows Media, and QuickTime implement these days.

[16:59:00.0000] <gsnedders>
webben: <em>little</em>

[16:59:01.0000] <webben>
I guess browsers could just incorporate ambulant.


2008-08-26
[17:00:00.0000] <webben>
"just"

[17:04:00.0000] <Dashiva>
Well, the anime I watch in mkv manages to handle multiple audio tracks and multiple subtitle tracks in a single file

[17:07:00.0000] <greghouston>
With VobSub there are several options for the subtitles, position on the screen, opacity, fadein/out ... It also looks like you can probably add more subtitle tracks to the file, though there is only one in this file that I am looking at.

[18:24:00.0000] <webben>
am I right in a vague recollection that we have text/html html5 parsers that are faster than xml parsers?

[18:36:00.0000] <Philip`>
webben: I don't think that's currently true

[18:38:00.0000] <Philip`>
(unless I'm misremembering what hsivonen has said in the past)

[18:38:01.0000] <Hixie>
i don't think we have any html5 parsers that aren't paying some sort of VM tax yet do we?

[18:39:00.0000] <Philip`>
takkaria has been writing one in plain C

[18:39:01.0000] <webben>
just saw the XML parsing is faster claim on an internal mailing list and seemed to remember that actually not necessarily being the case from a claim here

[18:40:00.0000] <webben>
(of course, it's difficult to tell anything in theory from actual implementations, but still)

[18:40:01.0000] <Philip`>
It seems the only real difference is that XML parsers are more mature and more heavily optimised than HTML5 parsers

[18:41:00.0000] <Hixie>
how is takkaria's, performance-wise?

[18:43:00.0000] <Philip`>
/me has no idea

[19:15:00.0000] <Hixie>
hsivonen: did you get any response to your feedback on WCAG2?

[19:30:00.0000] <Hixie>
i wonder where MikeSmith is

[19:30:01.0000] <Hixie>
haven't seen him in a while

[19:45:00.0000] <takkaria>
Hixie: about half as fast as libxml2's html parser

[19:45:01.0000] <takkaria>
Hixie: IIRC it parses the HTML5 spec in about 1.3s

[19:46:00.0000] <Hixie>
not bad, not bad

[19:46:01.0000] <Hixie>
what platforms does it run on?

[19:47:00.0000] <takkaria>
anything that runs C99

[19:47:01.0000] <takkaria>
though there are some unresolved crasher bugs on RISC OS at the moment

[19:48:00.0000] <takkaria>
runs on Linux fine, though, and I assume it'll cross-compile to Windows fairly easily

[19:48:01.0000] <Hixie>
could it be retrofitted into libxml2 to expose the same API but with the html5 parser?

[19:48:02.0000] <takkaria>
yeah, the browser it's being used in (NetSurf) does exactly that

[19:50:00.0000] <Hixie>
cool

[19:50:01.0000] <Hixie>
you should get gsnedders to use it

[19:50:02.0000] <Hixie>
:-)

[19:50:03.0000] <Hixie>
in spec gen

[19:50:04.0000] <Hixie>
is there any optimisation work planned?

[19:51:00.0000] <takkaria>
well, the tokeniser is pretty well optimised already but there are a couple of places it could be improved

[19:51:01.0000] <takkaria>
the treebuilder is pretty much unoptimised in any way whatsoever

[19:51:02.0000] <takkaria>
and there are some nice low-hanging fruit there

[19:53:00.0000] <takkaria>
problem with people using it at the moment is that it's a build-from-svn thing.  when we have out a release with versioning and built as a shared library and things using it in things like spec-gen becomes plausible

[19:53:01.0000] <Hixie>
ah ok

[19:53:02.0000] <Hixie>
it's so cool that you're doing this stuff

[19:54:00.0000] <takkaria>
well, I got summer of code funding, so I've been having a pretty good time writing it :)

[19:54:01.0000] <Hixie>
:-D

[19:55:00.0000] <takkaria>
my laptop broke the other day though so atm I'm somewhat offline and I've forgotten all my todolist items and things, and the profiling data I have

[19:55:01.0000] <takkaria>
but I think about 25% of the time in the treebuilder can be shaved off without much work

[19:56:00.0000] <Hixie>
nice

[19:56:01.0000] <Hixie>
you should get jgraham, Philip`, hsivonen to help you out, their experience with optimising parsers is bound to suggest some things you could do

[19:57:00.0000] <Hixie>
henri in particular has done some pretty fancy things with buffers that might have unexpected benefits

[19:58:00.0000] <takkaria>
amusingly we found the same things with buffers

[19:58:01.0000] <Hixie>
yeah? nice

[19:58:02.0000] <takkaria>
Hubbub used to try and buffer only the things that needed buffering and use offsets the rest of the time but it made the code simpler and faster to just always use buffers

[19:59:00.0000] <Hixie>
heh

[19:59:01.0000] <takkaria>
we discovered this roughly in parallel I think

[19:59:02.0000] <Hixie>
cool

[20:00:00.0000] <takkaria>
I've been keeping an eye on the v.nu parser for optimisations to steal, too

[20:00:01.0000] <takkaria>
but I've not touched the code for a week and a half and it's getting a bit hazy again :)

[20:04:00.0000] <Hixie>
:-)

[20:15:00.0000] <takkaria>
the alt thread makes me cry, I really did hope it would stop at some point in the last few days

[20:18:00.0000] <takkaria>
mainly because it becomes more and more obvious that the accessibility people haven't been taking anything other people have been saying in

[20:23:00.0000] <takkaria>
so much rhetoric, so little argument

[21:00:00.0000] <Hixie>
man watching usability studies is depressing

[21:11:00.0000] <Hixie>
so... depressing...

[21:13:00.0000] <othermaciej>
is it the quality of the software that's depressing or the quality of the users?

[21:14:00.0000] <Hixie>
hard to say

[21:15:00.0000] <Hixie>
i mean it's always the software's fault

[21:16:00.0000] <Hixie>
but what's depressing is the utter failure of many interactions

[21:37:00.0000] <Hixie>
i think blind users might just be desensitised to the sheer amount of repetition they hear

[21:43:00.0000] <Hixie>
good lord

[00:08:00.0000] <Dashiva>
Is it bad that my attention just drifts away whenever someone mentions search engine indexing as a benefit of RDF?

[00:11:00.0000] <roc>
what I don't understand is how search engines and browser are supposed to do magic things with this data while there is no central authority

[00:12:00.0000] <roc>
is each of N user agents supposed to independently discover all M namespaces and implement their semantics?

[00:12:01.0000] <Dashiva>
I just want to yell "It's the SEO, stupid"

[00:13:00.0000] <Dashiva>
roc: You just download the internet and walk the graph, it's simple enough :)

[00:22:00.0000] <Hixie>
roc: i'm always baffled as to how these people think search engines are going to use this data, given that if they did use this data, it would immediately be utterly unusable due to spammers

[00:22:01.0000] <Hixie>
roc: i'm especially amused though when people tell me something is needed for search engines

[00:22:02.0000] <Hixie>
hello

[00:22:03.0000] <Hixie>
i work for google

[00:23:00.0000] <Hixie>
how about _i_ tell you what we need, and you tell me what _you_ need

[00:24:00.0000] <roc>
heh

[00:25:00.0000] <Hixie>
Lachy: so i've found a huge problem with requiem

[01:41:00.0000] <Hixie>
if anyone wants to see the new draft alt="" regime before i commit it, it's at http://www.whatwg.org/specs/web-apps/current-work/#alt

[01:44:00.0000] <hsivonen>
Philip`: I thought DVD subtitles were bitmaps

[01:46:00.0000] <hsivonen>
Hixie: I got a reply for my ATAG feedback. I didn't get feedback on my WCAG feedback (from the WG that is).

[01:48:00.0000] <Hixie>
what was the ATAG feedback and what did they reply?

[01:49:00.0000] <hsivonen>
Hixie: well, if people think they know what Google needs, people think they know what CC needs :-)

[01:50:00.0000] <hsivonen>
/me looks up the ATAG mail urls

[01:50:01.0000] <hsivonen>
http://lists.w3.org/Archives/Public/public-atag2-comments/2008Aug/0000.html

[01:51:00.0000] <hsivonen>
http://lists.w3.org/Archives/Public/w3c-wai-au/2008JulSep/0072.html

[01:51:01.0000] <hsivonen>
http://lists.w3.org/Archives/Public/w3c-wai-au/2008JulSep/0073.html

[01:51:02.0000] <hsivonen>
http://lists.w3.org/Archives/Public/w3c-wai-au/2008JulSep/0074.html

[01:51:03.0000] <hsivonen>
http://lists.w3.org/Archives/Public/w3c-wai-au/2008JulSep/0075.html

[01:51:04.0000] <hsivonen>
those are the URLs for the ATAG thread

[01:52:00.0000] <Hixie>
man, these guideline docs are almost as vague as scripture

[01:52:01.0000] <hsivonen>
like I said in one of the emails, I was unable to verify an interpetation of the draft by reading the draft

[01:56:00.0000] <hsivonen>
sometimes I suspect that the guidelines are vague because the experts couldn't agree, so what got written down is something that each WG participant can see their own opinion in

[01:57:00.0000] <Hixie>
again like scripture then...

[01:57:01.0000] <hsivonen>
or at least I've gotten mixed interpretation from former editor and current editor

[02:17:00.0000] <Hixie>
ok.

[02:17:01.0000] <Hixie>
alt="" debate reply sent.

[02:17:02.0000] <Hixie>
for me to consider further new feedback, it has to say something that hasn't been said before.

[02:22:00.0000] <Philip`>
hsivonen: Oh, okay, you're right

[02:38:00.0000] <Dashiva>
Hixie: Implement robot9000 for the list :)

[02:42:00.0000] <Hixie>
hah

[02:42:01.0000] <Hixie>
steven's recent complaints are even more insane

[02:42:02.0000] <Hixie>
now he's complaining about my decisions before i even make them

[02:45:00.0000] <Dashiva>
Why doesn't he just stop pussyfooting around and talk to the chairs?

[02:46:00.0000] <annevk>
heh

[02:51:00.0000] <Dashiva>
Is that museum artifact example for real?

[02:51:01.0000] <hsivonen>
/me tries the million URL exercise again. this time with the output going to a dedicated hard disk and gzipped on the fly

[02:51:02.0000] <Dashiva>
Because more than anything else it just demonstrates the madness of dereferencing namespace URIs

[02:51:03.0000] <annevk>
stop hating distributed extensibility so much

[02:52:00.0000] <hsivonen>
Dashiva: museum people use RDF

[02:52:01.0000] <hsivonen>
Dashiva: a friend of mine was on a research group that created an ontology for museums

[02:53:00.0000] <Dashiva>
hsivonen: Sure, but did they go around downloading random pages?

[02:53:01.0000] <hsivonen>
I don't know.

[02:54:00.0000] <Dashiva>
hsivonen: More specifically, Charlie deciding to dereference is the unreal part, not people putting tuples in their pages

[02:54:01.0000] <hsivonen>
what I've heard about the ontology creation process corroborated my own guess about the process from my time at the National Archives

[02:54:02.0000] <virtuelv>
/me recalls the awesome presentation about rijksmuseum.nl from Dublin earlier this year

[02:55:00.0000] <virtuelv>
(awesome, as in, someone actually did something cool with RDF)

[03:00:00.0000] <hsivonen>
Hixie: I wish someone answered my question on how well the CC URI upgrade worked with tools that existed at the time

[03:02:00.0000] <Philip`>
"A somewhat strained analogy would be bringing

[03:02:01.0000] <Philip`>
in representatives from all of the cultures of the world and having them

[03:02:02.0000] <Philip`>
agree on a universal vocabulary."

[03:02:03.0000] <Philip`>
That sounds kind of similar to agreeing on a universal character encoding, rather than having everyone invent their own independent ones

[03:03:00.0000] <hsivonen>
Philip`: but that's not decentralized

[03:05:00.0000] <hsivonen>
though as far as universal vocabuliaries between cultures go, English is doing pretty well with decentralized extensibility

[03:06:00.0000] <Philip`>
hsivonen: Why not? People can decentralisedly invent and use their own character encodings, and they just need to convince their favourite tool vendors to support it, which seems easy enough for governments to do

[03:07:00.0000] <hsivonen>
Philip`: even the Chineses government can't make software developers change the way text is represented in RAM

[03:07:01.0000] <hsivonen>
Philip`: but the Unicode folks were able to

[03:14:00.0000] <Hixie>
hsivonen: got any early numbers regarding the validity study?

[03:26:00.0000] <hsivonen>
Hixie: no

[03:26:01.0000] <jgraham>
Good lord it's getting bad when Rob Burns is producing more useful comments on accessibility than the "accessiility experts"

[03:27:00.0000] <jgraham>
hsivonen: Thanks for the bug report yesterday. Score namespaces 1 james 0

[03:29:00.0000] <Hixie>
only 1? :-)

[03:30:00.0000] <Dashiva>
namespaces 1.<xsd:integer>

[03:31:00.0000] <hsivonen>
/me expects the gzipped dump of validation worker output to go to somewhere around 60 GB

[03:32:00.0000] <hsivonen>
even though I suppress duplicate messages per URL before writing out

[03:32:01.0000] <annevk>
/me beat the namespaces spec a few times by getting it revised

[03:37:00.0000] <hsivonen>
Hixie: at this rate, it will take 34 hours to get even the preliminary data dump

[03:37:01.0000] <Hixie>
what's your rate?

[03:38:00.0000] <Philip`>
hsivonen: You could try looking at fewer pages, which would make it go faster and probably wouldn't make things significantly less statistically significant :-)

[03:38:01.0000] <hsivonen>
465 urls per minute

[03:38:02.0000] <Hixie>
validator.w3.org apparently deals with millions of requests per day, so you need to get to the point of being able to do the whole sample from the network in less than 24 hours. :-)

[03:39:00.0000] <hsivonen>
now that I think about it, I started the process on the wrong HotSpot version...

[03:39:01.0000] <Hixie>
oh that's not too bad, i thought you meant a small sample when you said "preliminary data dump"

[03:39:02.0000] <hsivonen>
I could throw away 47 minutes now and restart it with a better JIT

[03:39:03.0000] <Hixie>
seems wise

[03:40:00.0000] <Hixie>
also if you're doing any console I/O, hide it, that ends up being a massive cost when i'm doing debugging runs

[03:40:01.0000] <Hixie>
dunno if it applies to you

[03:40:02.0000] <hsivonen>
it applies to me

[03:42:00.0000] <hsivonen>
I print progress every 100 urls though, to avoid spending too much time printing

[03:42:01.0000] <Hixie>
that helps

[03:46:00.0000] <zcorpan>
Hixie: the current spec text on alt makes sense to me

[03:46:01.0000] <Hixie>
woot!

[03:47:00.0000] <Hixie>
i have 100% support from all those who have commented!

[03:47:01.0000] <Hixie>
/me does a little dance

[03:47:02.0000] <zcorpan>
:)

[03:47:03.0000] <Hixie>
sadly it can only go down from here :-(

[03:47:04.0000] <zcorpan>
it could stay at 100%

[03:47:05.0000] <Hixie>
that seems unlikely

[03:47:06.0000] <hsivonen>
/me is considering how productive it would be to rain on the parade at this point

[03:47:07.0000] <Hixie>
:-)

[03:47:08.0000] <zcorpan>
yes, at least if you get further comments

[03:47:09.0000] <Hixie>
well, if i missed something important, do say

[03:50:00.0000] <hsivonen>
aargh. NPE in class loading code on the other JVM

[03:51:00.0000] <Philip`>
/me read the end of the email first, to see what the decision was, but that just said it was "option F", so he had to read the top of the email, which only went to option E, and so he had to end up reading the entire email to find the actual chosen solution :-(

[03:51:01.0000] <Hixie>
Philip`: heh - you could have just read the spec diff :-P

[03:51:02.0000] <Hixie>
/me admits that that was partially intentional

[03:51:03.0000] <zcorpan>
Hixie: i wonder if the captcha case could have a caption in the form of <fieldset><legend> or <th> instead

[03:52:00.0000] <Hixie>
ooh, that's not a bad idea

[03:52:01.0000] <Philip`>
Hixie: The spec diff looks longer than the email

[03:52:02.0000] <Hixie>
Philip`: heh maybe

[03:52:03.0000] <Hixie>
zcorpan: we'll see what other feedback i get, but that seems reasonable on the face of it

[03:53:00.0000] <Hixie>
though we'd have to decide on a priority order

[03:53:01.0000] <Hixie>
i guess it could be nearest-enclosing

[03:54:00.0000] <hsivonen>
new this is *weird* the code that works on Sun Java 1.6 and OpenJDK 6 has broken classloader behavior on Apple's port of JDK 6

[03:54:01.0000] <hsivonen>
aargh

[03:54:02.0000] <zcorpan>
Hixie: yeah... and i'm sure there are cases where it wouldn't make sense

[03:54:03.0000] <hsivonen>
Class loader issues suck

[03:54:04.0000] <hsivonen>
NPE sucks

[03:54:05.0000] <hsivonen>
now I have NPE in class loading

[03:54:06.0000] <hsivonen>
/me goes out to lunch

[03:54:07.0000] <Hixie>
NPE?

[03:54:08.0000] <hsivonen>
NullPointerException

[03:56:00.0000] <Hixie>
oh

[03:56:01.0000] <zcorpan>
Hixie: editorial: in the example listing images, i'd say "   &lt;tr> &lt;th> Image &lt;th> Description" (in singular)

[03:57:00.0000] <Hixie>
/me sends himself an e-mail saying that

[03:59:00.0000] <MikeSmith>
hsivonen: found what seems to be another discrepancy in HTML5 schema vs. the spec

[03:59:01.0000] <MikeSmith>
spec says that col element can have a span attribute

[04:00:00.0000] <MikeSmith>
but schema does not allow span on col

[04:00:01.0000] <zcorpan>
hey i tested <col span> in v.nu the other day

[04:00:02.0000] <cyclist>
pls help.  In the HTML5 spec it sais that block level elements inside stricly inline level elements are "not allowed".

[04:00:03.0000] <zcorpan>
wonder why i didn't catch that it didn't validate

[04:01:00.0000] <cyclist>
If somebody does that nontheless, what is the browser supposed to do?

[04:01:01.0000] <cyclist>
Break the inlines around the block perhaps?

[04:01:02.0000] <zcorpan>
cyclist: rendering-wise it's defined in css

[04:02:00.0000] <zcorpan>
cyclist: you can even get that with conforming markup: <span>x<span style=display:block>y</span>z</span>

[04:02:01.0000] <annevk>
cyclist, you're reading an old version of HTML5

[04:02:02.0000] <cyclist>
yes indeed.

[04:02:03.0000] <annevk>
cyclist, try http://whatwg.org/html5 or http://html5.org/spec

[04:02:04.0000] <cyclist>
the newer version defines action to be taken on the "not allowed"?

[04:03:00.0000] <Philip`>
The newer version allows it

[04:03:01.0000] <cyclist>
ah. ok

[04:03:02.0000] <Philip`>
in some cases, I think

[04:03:03.0000] <annevk>
cyclist, the newer version doesn't have block or inline level elements

[04:03:04.0000] <Philip`>
unless I'm misremembering

[04:03:05.0000] <Philip`>
Oh, okay

[04:04:00.0000] <MikeSmith>
hsivonen: I see why now.. col.attrs.span is defined but not referenced -- because the schema is missing "col.attrs &= col.attrs.span"

[04:04:01.0000] <zcorpan>
Philip`: it doesn't allow non-phrasing flow content in phrasing content

[04:04:02.0000] <Hixie>
cyclist: what's the exact case you are trying to do?

[04:05:00.0000] <zcorpan>
wow the spec's terminology won't be understandable to J. Author

[04:05:01.0000] <cyclist>
write a web browser :)

[04:05:02.0000] <cyclist>
or at least render pages

[04:05:03.0000] <Hixie>
oh you're looking for UA behaviour, ok

[04:05:04.0000] <Hixie>
well the parsing is defined in the parsing section of HTML5

[04:05:05.0000] <Hixie>
and the rendering is defined by CSS

[04:05:06.0000] <zcorpan>
"the spec doesn't allow block content in inline content" would be more understandable

[04:06:00.0000] <Hixie>
annevk: any idea if opera has any feedback on the workers stuff? (in particular the latest proposal)

[04:06:01.0000] <Hixie>
(not what the latest spec says)

[04:07:00.0000] <Philip`>
zcorpan: I was thinking of the <a><div>... thing which I think changed, but I have no idea what terminology the spec uses for that kind of thing

[04:07:01.0000] <cyclist>
OK. I had a really old version of HTML5. The new version passes the ball to css it seems. thanks

[04:07:02.0000] <zcorpan>
Philip`: oh right

[04:08:00.0000] <Hixie>
cyclist: np

[04:09:00.0000] <Hixie>
/me watches his cat pull down the kitten-sized cat furniture

[04:10:00.0000] <Hixie>
she really needs to stop leaping on it

[04:10:01.0000] <annevk>
Hixie, no, other than the feedback you already got I think we don't have anything new (nor have we debated this much)

[04:12:00.0000] <zcorpan>
Hixie: could the spec use the terminology "inline content" and "flow content" (and "block content" which would be flow excluding inline) instead of "phrasing"? it'd be closer to what authors use and could still be distinguished from css ("content" vs "-level")

[04:12:01.0000] <Philip`>
You could put metal spikes on top of the furniture to keep the cat off

[04:21:00.0000] <Hixie>
zcorpan: i don't want to use the word "inline"

[04:21:01.0000] <Hixie>
Philip`: it's cat furniture! she's welcoem to use it :-P

[04:21:02.0000] <Hixie>
just wish she'd not pull the whole thing over each time she leaps at it :-P

[04:24:00.0000] <Hixie>
ok julian is clearly just trolling me

[04:24:01.0000] <annevk>
and missing the point

[04:27:00.0000] <hsivonen>
Hixie: regarding the throughput competitiveness of Validator.nu as a back end to the W3C Validator:

[04:27:01.0000] <hsivonen>
the online version should have better throughput than this offline version

[04:27:02.0000] <hsivonen>
the online version doesn't build a tree

[04:27:03.0000] <hsivonen>
this offline worker does

[04:30:00.0000] <Hixie>
ah ok

[04:30:01.0000] <Hixie>
why the difference?

[04:31:00.0000] <hsivonen>
what the online version does yield better throughput and is acceptable for cases where an author is using it as a tool and even tries to keep the markup sane

[04:32:00.0000] <Hixie>
(btw if anyone read my response to manu, did it seem right? or am i insane?)

[04:32:01.0000] <hsivonen>
however, to discover all the craziness that's going on out there, I don't want to stop at the first non-streamable error in the offline research tool

[04:32:02.0000] <Hixie>
hsivonen: oooh, cunning

[04:32:03.0000] <Hixie>
can't wait to see this data

[04:32:04.0000] <Hixie>
i actually just want to see the % of pages that have zero errors :-)

[04:43:00.0000] <hsivonen>
Hixie: want to see what pages have zero parser-level errors and are in stardards mode and then I want to see what we could do to make the validation layer errors go away on those pages

[04:44:00.0000] <hsivonen>
(zero parser level errors with legacy doctype errors ignored)

[04:45:00.0000] <annevk>
maybe legacy doctypes should become conforming HTML5 doctypes

[04:46:00.0000] <annevk>
though this could be controversial

[04:47:00.0000] <hsivonen>
fwiw, I'm running the parser in the survey mode, which suppresses tree builder -level doctype errors

[04:48:00.0000] <hsivonen>
I consolidate the kind of messages that cite attribute values

[04:48:01.0000] <hsivonen>
then I consolidate duplicate messages per URL

[04:49:00.0000] <hsivonen>
then I log tuples that have URL, quirkiness, layer and message

[04:49:01.0000] <hsivonen>
where quirkiness is HTML5, other Standards, almost standards or quirks

[04:49:02.0000] <hsivonen>
and layer is parser layer or validation layer

[04:50:00.0000] <zcorpan>
Hixie: why not?

[04:50:01.0000] <hsivonen>
also, when an URL didn't have errors on a given layer, I log a special NO ERRORS message

[04:52:00.0000] <annevk>
hmm, I'm going to "North Korean Flying Circus" tonight, it seems to be so new that there's no wikipedia page yet...

[04:54:00.0000] <Lachy>
annevk, it could mean that it's just not notable enough, regardless of how new it is.

[04:54:01.0000] <hsivonen>
/me tries to resist a rant about the pointlessness of non-notability deletionism

[04:54:02.0000] <Lachy>
or that it just doesn't have a page on the english wikipedia, since it's korean

[04:55:00.0000] <annevk>
all kinds of small Dutch villages have English Wikipedia entries

[04:55:01.0000] <zcorpan>
hsivonen: what v.nu calls "legacy doctype" can trigger standards mode; are you counting those?

[04:55:02.0000] <hsivonen>
zcorpan: I'm counting them as non-HTML5 Standards mode

[04:56:00.0000] <hsivonen>
zcorpan: I'm counting quirkiness from the DocumentModeHandler--not from messages

[04:56:01.0000] <zcorpan>
hsivonen: ok

[04:56:02.0000] <Lachy>
/me agrees the notability rules for wikipedia are a bit strict and arbitrarily applied sometimes

[04:57:00.0000] <annevk>
I think it's just that it's quite new and so far only shown in the Netherlands

[04:57:01.0000] <annevk>
apparently the World Premier was August 1

[04:59:00.0000] <Hixie>
zcorpan: i want to completely avoid confusion with css

[05:00:00.0000] <Hixie>
holy crap, 2/2 on feedback, and one of them was an accessibility person

[05:02:00.0000] <hsivonen>
Hixie: a person creates a new document in Nvu, drags a JPEG from Finder to the blank document and saves. What does Nvu do to make the output pass a machine-administered syntax check?

[05:03:00.0000] <hsivonen>
assuming an HTML5 version of Nvu under the new spec text

[05:03:01.0000] <annevk>
was Hixie's "Great!" just now a disguised "+1"? :p

[05:04:00.0000] <zcorpan>
Hixie: i understand that, but i think that authors in the wild simply aren't going to use the current spec's terminology

[05:04:01.0000] <zcorpan>
Hixie: so it's not helping, it's just adding to the confusion

[05:05:00.0000] <Hixie>
hsivonen: i do not believe the current spec allows the editor in such a scenario to output conforming markup without further input from the user.

[05:05:01.0000] <zcorpan>
Hixie: also, try looking up "phrasing" in wikipedia or a dictionary

[05:05:02.0000] <Hixie>
annevk: no, i wasn't agreeing with him, i was expressing happiness. :-)

[05:06:00.0000] <Hixie>
zcorpan: i think given time, authors will begin to use the new terminology. there's no rush here. i'm open to other terms of "phrasing" isn't any good though.

[05:06:01.0000] <hsivonen>
Hixie: do you expect glazou to go along with either putting up a modal prompt or writing non-conforming output?

[05:06:02.0000] <zcorpan>
Hixie: how about "text-level content"?

[05:07:00.0000] <hsivonen>
class loading is such a PITA

[05:07:01.0000] <Hixie>
hsivonen: depends. If this happens for a lot of images on the page, then Nvu could probably assume empty alt="" and just put that in, with the argument that the UI allows the user to specify the alternative text but defaults to assuming decorative images, and that it's the user's fault if the user doesn't give the right data.

[05:08:00.0000] <Hixie>
hsivonen: if it's just a single image, but the user gives alt="" for the other images explicitly, then i'd say that for that image, maybe nvu should see if there is a header or legend that follows what the spec says now, and if not, could include an apologetic title="" attribute

[05:09:00.0000] <Hixie>
hsivonen: or the editor could warn on output that the content isn't valid

[05:09:01.0000] <zcorpan>
Hixie: "why doesn't this validate?" "it's not allowed to have block children of inline elements" is a common conversation. how would you say that using the spec's terminology?

[05:09:02.0000] <Hixie>
hsivonen: (either way, accessibility is screwed if the author, as in this case, is actively avoiding helping even when they really should and could)

[05:10:00.0000] <Hixie>
zcorpan: "You can't put a <div> in a <span>."

[05:10:01.0000] <Hixie>
zcorpan: (or whatever)

[05:10:02.0000] <hsivonen>
Hixie: an editor that hides the HTML file format from the user should always be able to write output that passes a machine-admistered syntax check

[05:10:03.0000] <hsivonen>
adiministered

[05:10:04.0000] <hsivonen>
doh

[05:10:05.0000] <zcorpan>
Hixie: but that would only give him a fish, not teach him to fish

[05:10:06.0000] <hsivonen>
/me can't spell today

[05:10:07.0000] <Hixie>
zcorpan: or "phrasing content can only contain other phrasing content, and you're putting elements in your phrasing content that aren't phrasing content" or some such

[05:11:00.0000] <hsivonen>
Hixie: if you want writing output to proceed despite a check failing, that check belongs in a different class of checks

[05:12:00.0000] <Hixie>
hsivonen: i think it's quite reasonable for an editor to assume images are decorative unless told otherwise, assuming the ui makes that clear

[05:12:01.0000] <hsivonen>
so we're back to emitting alt="" silently

[05:13:00.0000] <Hixie>
i haven't spent any time trying to cater for the case of editors where the user is actively hostile to making accessible content

[05:14:00.0000] <Hixie>
for certain images you can do things like OCR, there's been some research in that same (q.v. e.g. webinsight)

[05:14:01.0000] <Hixie>
s/same/area/

[05:14:02.0000] <hsivonen>
that OCR really should be on the client side

[05:15:00.0000] <Hixie>
certainly would be nice for reader-side uas to have ocr, yes, and that's what webinsight does actually

[05:16:00.0000] <Hixie>
for images that are sole content of links, alt="" text can be synthesised by grabbing the title of the target page, too

[05:16:01.0000] <Hixie>
there are various things that editors can do to silently handle hostile authors

[05:16:02.0000] <Hixie>
but fundamentally i don't think there's a great solution. assumin decorative images will handle most images, frankly.

[05:16:03.0000] <Philip`>
<a href="logout.cgi"><img src="logout.png"></a> - grabbing the target page's title won't work so well there

[05:17:00.0000] <Hixie>
it doesn't work in all cases, no

[05:17:01.0000] <Hixie>
webinsight's data suggests you can fill about 40% of images and get about a 90% success rate on those images

[05:17:02.0000] <Hixie>
which is better than nothing

[05:17:03.0000] <Philip`>
It doesn't just not work in all cases - it fails badly with negative consequences in some cases, so it's not safe to do it at all

[05:17:04.0000] <hsivonen>
Hixie: my point is that you have an abstaction leak if an agent can let user input or lack thereof break the syntax of the output file format

[05:17:05.0000] <Hixie>
this is just like these editors assuming that when you hit bold you mean bold, as opposed to important, and that when you hit font-size+1 you mean to change the style sheet, not change to a <h1>, or whatever

[05:18:00.0000] <Hixie>
Philip`: going from no alt text to crap alt text 0.4*5% of the time and good alt text 0.4*95% of the time is better than staying with no alt text always.

[05:19:00.0000] <Hixie>
hsivonen: i'm saying you don't have to break the syntax. you can make assumptions, just like editors do for all their other features.

[05:19:01.0000] <Philip`>
Hixie: It's not just affecting the usefulness of the alt text, if it's silently making unexpected extra HTTP requests and accidentally logs you out of the site you're on or deletes pages

[05:19:02.0000] <hsivonen>
Hixie: this is not just like that. If the author makes a semantic error, a machine-administered check doesn't catch it

[05:20:00.0000] <hsivonen>
Hixie: I'm saying that editors programmed by people who object to abstraction leaks will plug the machine-detectable leaks so that they can't be machine-detected

[05:20:01.0000] <Hixie>
Philip`: i'm not suggesting sending cookies or http auth credentials with these requests

[05:21:00.0000] <hsivonen>
and that plugging will occur in the program code in case the user is hostile, so the code will run even when the user isn't hostile

[05:21:01.0000] <Hixie>
hsivonen: i'm not sure which case you're referring to

[05:22:00.0000] <Hixie>
hsivonen: why is assuming the user meant alt="" not the same as assuming the user meant font-size: 2em?

[05:22:01.0000] <Hixie>
(as opposed to alt="Logout" or <h1>)

[05:22:02.0000] <Hixie>
(which we presume would both be better)

[05:22:03.0000] <hsivonen>
Hixie: I mean that if the user uses <h1> to make text bold, that's not the kind of problem programmers will plug

[05:23:00.0000] <hsivonen>
Hixie: if the validator whines about a piece of syntax being absent, (some) programmers will make it present

[05:23:01.0000] <Hixie>
what syntax would be absent?

[05:23:02.0000] <hsivonen>
Hixie: alt, or legand, or <hn>

[05:24:00.0000] <Hixie>
why would alt="" be absent? I'm saying to include it always.

[05:24:01.0000] <Hixie>
i'm saying it seems legitimate to have the UA assume that images are decorative unless told otherwise, and so to give them empty alt="".

[05:24:02.0000] <hsivonen>
Well, then you move the autogenerated junk problem to <h1>

[05:24:03.0000] <Hixie>
(with different behaviour for images in links, e.g. fetching target page titles.)

[05:25:00.0000] <Hixie>
wysiwyg editors with hostile authors will output pages with poor accessibility, yes

[05:25:01.0000] <Hixie>
why is this surprising?

[05:25:02.0000] <hsivonen>
Hixie: if you'd specified that the chain goes from <h1> to <title>, this would reduce to empty <title></title> being there

[05:26:00.0000] <hsivonen>
Hixie: it's not

[05:26:01.0000] <Philip`>
Hixie: That wouldn't solve the problem with <a href="logout.cgi?s=1910764e1a0f1976d"><img src="logout.png"></a>

[05:26:02.0000] <Hixie>
hsivonen: i really don't understand

[05:26:03.0000] <Philip`>
(...where the ?s is the session identifier)

[05:26:04.0000] <Philip`>
(for sites that don't want to rely on cookies)

[05:26:05.0000] <Hixie>
Philip`: why would the page author include a session identifier in a link he is putting in a wysiwyg editor??

[05:27:00.0000] <Hixie>
oh you're talking about the client-side thing

[05:27:01.0000] <hsivonen>
Hixie: what's surprising is that you seem to join accessibility advocates in asking lack of accessibility to be flagged by introducing an error on another layer

[05:27:02.0000] <Philip`>
Hixie: They wouldn't, but a UA can't tell whether the link came from a WYSIWYG editor, so it's a problem if the UA tries to request the page in situations which the author didn't expect

[05:28:00.0000] <Hixie>
Philip`: well, tell the webinsight people. i'm only suggesting it here for the editor side, the spec doesn't even suggest it for the browser side.

[05:28:01.0000] <Philip`>
Hixie: Oh, if it's just in the editor then I guess that's safer

[05:29:00.0000] <Hixie>
hsivonen: this isn't new

[05:29:01.0000] <Hixie>
hsivonen: the spec has always required alt="" for all images that could posibly have it.

[05:30:00.0000] <Hixie>
hsivonen: just like it requires using <h1> for headers, or not having a headers="" attribute point to something outside the table, or whatever

[05:30:01.0000] <hsivonen>
making Apple's Java 6 work is too much trouble. I'l go back to Apple's Java 5

[05:30:02.0000] <hsivonen>
now having lost more than 47 minutes of processing time

[05:31:00.0000] <hsivonen>
Hixie: you know that those are different from the point of view of programming a WYSIWYG editor

[05:31:01.0000] <Philip`>
hsivonen: You should make your processor incremental, so it can carry on from where it left off if it's interrupted or crashes :-)

[05:32:00.0000] <hsivonen>
Philip`: it didn't seem worth the trouble at firts

[05:32:01.0000] <Hixie>
hsivonen: i honestly can't see the difference between an editor assuming that something should be a style="" attribute instead of an <h1> attribute and an editor assuming something should say alt="" instead of alt="W3C".

[05:32:02.0000] <hsivonen>
first

[05:32:03.0000] <Hixie>
er, <h1> element

[05:32:04.0000] <Hixie>
in both cases it is syntactically ok either way, and in both cases it is bad for accessibility.

[05:33:00.0000] <hsivonen>
Hixie: when you select text and choose "Heading 1", it becomes a <h1> in an operation that's atomic in a GUI

[05:33:01.0000] <hsivonen>
Hixie: inserting an image and entering alt isn't atomic without modal dialogs

[05:34:00.0000] <annevk>
you could have several insert image operations

[05:34:01.0000] <annevk>
insert decorative image, insert content image, etc.

[05:34:02.0000] <Hixie>
hsivonen: so? inserting a link is also not "atomic without modal dialogs".

[05:35:00.0000] <Hixie>
hsivonen: i'm not saying that you need to ask for alt="" or href="" when you insert the image or link

[05:35:01.0000] <Hixie>
hsivonen: i'm just saying that it's fine for the UA to make assumptions as part of its UI

[05:35:02.0000] <Hixie>
hsivonen: e.g. that images are decorative unless told otherwise

[05:35:03.0000] <Hixie>
hsivonen: or taht font size changes mean style="" attributes and not <h1> or <h2> or <small>, unless told otherwise

[05:35:04.0000] <Philip`>
You could have a modeless dialog box that requires you to select an image and type in some alt text, before clicking or dragging or something to insert it into the page, and there's no need to modally block all other interaction while you're doing that

[05:35:05.0000] <hsivonen>
Hixie: when a user inserts a link, the user would fail the primary goal of the operation if the URL wasn't supplied 

[05:36:00.0000] <Hixie>
hsivonen: let's back up. what is your fundamental objection?

[05:37:00.0000] <Hixie>
because right now i'm agreeing with everything you say, so we're clearly off the controversy.

[05:37:01.0000] <hsivonen>
My fundamental objection is that given the current state of HTML generating programs, there's a case where a user can put a program in a state that can't be serialized without either violating the file format syntax or without stuffing junk

[05:39:00.0000] <hsivonen>
unless, of course, we assume that the UI should block users from inserting images without there being a heading

[05:39:01.0000] <Hixie>
so you believe it is inappropriate for an editor to assume that images are decorative unless told otherwise?

[05:39:02.0000] <hsivonen>
or from removing a heading if there are images

[05:39:03.0000] <hsivonen>
Hixie: yes

[05:39:04.0000] <Hixie>
hsivonen: why?

[05:40:00.0000] <Hixie>
(note: statistically, most images are decorative)

[05:40:01.0000] <Hixie>
(according to some studies i was looking at earlier, anyway.)

[05:40:02.0000] <hsivonen>
because then users who don't see images aren't alerted about relevant presence of non-decorative images authored with such UI

[05:42:00.0000] <Hixie>
do you therefore also think it is inappropriate for an editor to assume that when the user says "font size bigger", he wants a style="font-size:2em" attribute instead of an <h1> element, even if the user is really changing the font size of his headers?

[05:42:01.0000] <hsivonen>
otoh, if the conclusion is that users don't need to be alerted about that, and the distinction in current UAs between empty alt and no alt is useless, then this has been a rather veiled conclusion

[05:42:02.0000] <Hixie>
i do not believe that the distinction in current UAs between empty alt and no alt is useless

[05:42:03.0000] <hsivonen>
I think it is inappropriate to map UI that says "bigger" in words or icons to <h1>

[05:43:00.0000] <Hixie>
but surely if we don't mark up these headers as headers, then users who don't see css aren't alerted about relevant presence of headers authored with such UI

[05:44:00.0000] <Hixie>
and based on the usability studies i was looking at earlier, users of ATs navigate using headers far more than they look for images.

[05:44:01.0000] <hsivonen>
yes

[05:44:02.0000] <Hixie>
i don't understand the difference.

[05:45:00.0000] <hsivonen>
the presence of the text doesn't get hidden when marked up as big

[05:45:01.0000] <hsivonen>
the presence of the image gets hidden when marked up as alt=""

[05:46:00.0000] <Hixie>
for all intents and purposes, the presence of headers gets hidden when marked up as big. AT users don't just read the page, they navigate by headers and then spot read around them.

[05:46:01.0000] <hsivonen>
as far as I can tell, we've now gotten back to where HTML 4.01 was

[05:46:02.0000] <Hixie>
they only very rarely read the whole page through.

[05:47:00.0000] <Hixie>
we are where HTML4 was, except that it is possible for flickr and your image report tool to be conforming, yes.

[05:47:01.0000] <zcorpan>
" Line 1, Column 24: RCDATA element nu.validator.htmlparser.impl.ElementName@1c1ace8 contained the string </, but this did not close the element.."

[05:47:02.0000] <Hixie>
(and except for the spec having examples of good alt text. a lot of examples.)

[05:47:03.0000] <hsivonen>
and with HTML 4.01, we don't have ATAG 2.0 (comprehensibly) telling generator agent developers to emit alt=""

[05:48:00.0000] <Philip`>
Authoring tools don't know whether the author intended text to be a header or intended it to just be bigger, so they can't indicate the difference in the markup; but they do know whether the author intended an image to be decorative or just didn't think about it at all, so they could indicate that in the markup, except they can't because HTML5 doesn't allow that distinction, so information is lost

[05:49:00.0000] <Philip`>
which seems to be a difference between those cases

[05:49:01.0000] <Hixie>
they only know that the user didn't bother because they give the option to the user of saying decorative vs not decorative, but they also give the user the option of saying header vs paragraph, so i don't think the distinction is valid

[05:50:00.0000] <hsivonen>
zcorpan: thanks

[05:51:00.0000] <hsivonen>
Hixie: anyway, I'm too tired to keep insisting that the distinction be made

[05:51:01.0000] <Hixie>
fair enough

[05:52:00.0000] <hsivonen>
Hixie: I would like to see comprehensible and honest guidance for generator agent authors, though, saying that they should rather do alt="" than junk like alt="file/path.jgp"

[05:52:01.0000] <Hixie>
imho it is quite possible to make non-modal ui that encourages alternative text to be given, e.g. by having text inputs hover over images where alt text is not yet given, with a little button to be used if hte image is really decorative

[05:52:02.0000] <Hixie>
yeah adding suggestions for this would be useful

[05:54:00.0000] <hsivonen>
if the ATAG people as a WG agreed with this conclusion, though, one would think they'd written it down comprehensibly

[05:55:00.0000] <hsivonen>
Hixie: that is, "if the markup generator does not have a text alternative available, it should generate ..."

[05:56:00.0000] <hsivonen>
where I assume "..." ends up being "alt=''"

[05:57:00.0000] <Hixie>
noted

[05:57:01.0000] <Hixie>
me sleep now

[05:57:02.0000] <Hixie>
nn

[06:15:00.0000] <zcorpan>
am i the only one getting tired of garret smith?

[06:18:00.0000] <annevk>
no

[06:23:00.0000] <hsivonen>
Hixie: your reply to Manu Sporny made sense from the Google point of view

[06:24:00.0000] <hsivonen>
Hixie: however, if state of the art was the same on handheld devices (i.e. without a lot of processing against a large repository of previous data), surely we could do away with <h1>?

[06:43:00.0000] <zcorpan>
hsivonen: you could use aria-describedby instead of <label> for your select box (that replaces the label "Document") to avoid the focus problem

[06:45:00.0000] <zcorpan>
/me notes that aria keywords are in the opposite direction of accessibility api keywords

[06:45:01.0000] <zcorpan>
e.g. aria-labelledby vs RELATION_LABEL_FOR

[06:46:00.0000] <zcorpan>
hsivonen: oops, i meant aria-labelledby, not aria-describedby

[06:48:00.0000] <hsivonen>
zcorpan: ok.

[06:48:01.0000] <hsivonen>
I don't have a way to test though. I can't get Orca to work with my keyboard

[07:12:00.0000] <BenMillard>
damn, I just noticed the purpose of <address> and the difference between <ul> and <ol> has been discussed here in recent days but I missed it all :(

[07:14:00.0000] <zcorpan>
hsivonen: hmm. multiline <script> doesn't work in livedom in opera. wonder why

[07:15:00.0000] <hsivonen>
zcorpan: could it be that Opera puts CRLF in the DOM but Gecko and WebKit put an LF there?

[07:16:00.0000] <annevk>
could be

[07:16:01.0000] <hsivonen>
if so, it could be my bug

[07:16:02.0000] <annevk>
well, browser incompat is a bug too

[07:17:00.0000] <zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cscript%3Efoo%0D%0Abar%3C%2Fscript%3E%3Cscript%3Ew(document.getElementsByTagName('script')%5B0%5D.textContent.replace('%5Cr'%2C'%5C%5Cr').replace('%5Cn'%2C'%5C%5Cn'))%3C%2Fscript%3E

[07:17:01.0000] <zcorpan>
that's it

[07:18:00.0000] <zcorpan>
/me files a bug on opera

[08:03:00.0000] <Philip`>
/me sees that http://blogs.msdn.com/ie/archive/2008/08/25/privacy-beyond-blocking-cookies-bringing-awareness-to-third-party-content.aspx links to the #whatwg IRC logs

[08:10:00.0000] <annevk>
MS is watching you :D

[08:13:00.0000] <Xenos>
Hm :p

[08:30:00.0000] <zcorpan>
i wonder when we get a bug saying that &nbsp; doesn't work with an "-//W3C//DTD XHTML+RDFa 1.0//EN" doctype

[08:31:00.0000] <hsivonen>
I'm seeing mixed messages about the expected future browser-sensitivity of RDFa by RDFa proponents

[08:35:00.0000] <hsivonen>
I've learned various things from trying to validate a million documents

[08:35:01.0000] <hsivonen>
1) Jing leaks memory

[08:35:02.0000] <hsivonen>
2) parser bugs get exposed

[08:36:00.0000] <hsivonen>
3) the results take more space than expected

[08:36:01.0000] <hsivonen>
4) HFS+ sucks

[08:36:02.0000] <hsivonen>
5) Time Machine is brittle

[08:37:00.0000] <hsivonen>
6) Apple's Java 6 has a weird classloader bug

[08:46:00.0000] <Dashiva>
"In this scenario Charlie is making a mistake, and treating strings that look like URIs as if they had other properties of URIs."

[08:46:01.0000] <Dashiva>
How is this any different for namespace URIs?

[08:57:00.0000] <gsnedders>
Hixie: That'd be easier if I had any knowledge of C :)

[08:59:00.0000] <Philip`>
gsnedders: C is just like JavaScript - it's got curly braces and everything

[08:59:01.0000] <gsnedders>
Philip`: :)

[08:59:02.0000] <gsnedders>
OK, I was exaggerating.

[08:59:03.0000] <gsnedders>
I have some knowledge of C.

[08:59:04.0000] <gsnedders>
Just almost none.

[09:01:00.0000] <Dashiva>
Just use something that compiles to C

[09:02:00.0000] <gsnedders>
/me notes he already has an almost release-ready version in Python, and doesn't care to rewrite it

[09:02:01.0000] <Philip`>
Use JavaScript with a JIT and then extract the compiled code from memory and disassemble it and decompile it to C

[09:03:00.0000] <Dashiva>
Use parrot

[09:03:01.0000] <Philip`>
/me made a C-to-Parrot compiler once

[09:03:02.0000] <Philip`>
(called Carrot)

[09:03:03.0000] <Philip`>
(but never released since I got bored before it worked quite properly)

[09:04:00.0000] <Philip`>
(but it was able to compile somebody's Coke-to-Parrot compiler which was written in C, and run that in Parrot)

[09:19:00.0000] <Lachy>
does anyone here have experience with ruby?

[09:19:01.0000] <Lachy>
I need to install libxml-parser-ruby1.8 on Mac, but I can't figure out how?

[09:56:00.0000] <codedread_>
Hixie: quick question - is it better for me to raise a bug with the HTML spec by using the W3C  Bug / Issue Tracking Service, or is an email ok ?

[09:58:00.0000] <gsnedders>
codedread_: If anything, he'd prefer email

[09:58:01.0000] <codedread_>
ok, thanks

[09:58:02.0000] <zcorpan>
codedread_: i believe he's said that he'll guarantee response if sent to bugzilla or the whatwg mailing list, and that he prefers email

[09:59:00.0000] <codedread_>
zcorpan - thanks

[09:59:01.0000] <gsnedders>
(He didn't say how soon that response would be, though)

[10:01:00.0000] <zcorpan>
gsnedders: statistically of all issues i've raised, the issues i've raised in bugzilla have been addressed a lot quicker than those i've raised in email :)

[10:02:00.0000] <gsnedders>
Lies, damned lies and statistics!

[10:07:00.0000] <Philip`>
zcorpan: If your issues have been addressed quicker, that means hundreds of other people's issues have been delayed, so it is very selfish of you, and you should instead work out how to raise issues so that it will take two or three years before they're addressed

[10:10:00.0000] <zcorpan>
Philip`: i'm a selfish person

[10:27:00.0000] <Philip`>
http://developer.yahoo.net/blog/archives/2008/08/rdf_xslt_and_the_monkey_makes_3.html - "<dc:identifier>http://www.nytimes.com/2008/05/26/business/media/26link.html?_r=1&oref=slogin</dc:identifier>" - alas for ampersands

[10:28:00.0000] <zcorpan>
Philip`: markup is hard

[10:29:00.0000] <Philip`>
Hmm, and the second half of that blog post is in fixed-width font, presumably because they forgot a </code> somewhere :-(

[10:30:00.0000] <zcorpan>
/me rests his case

[11:06:00.0000] <Philip`>
krijnh: Do you know where all the search.live.com referrers to the IRC logs are coming from? They claim to have URLs like http://search.live.com/results.aspx?q=joins which is weird since your page isn't high in a search for that term and since it's less crufty than real search URLs would be

[11:11:00.0000] <gsnedders>
If you want weirdness, look at what my site gets.

[11:11:01.0000] <gsnedders>
:)

[11:14:00.0000] <Philip`>
gsnedders: That's just weird people, which is different to weird URLs :-)

[11:14:01.0000] <gsnedders>
Philip`: "32" — that found my site.

[11:16:00.0000] <Philip`>
/me sees a page that uses <!--googleoff: snippet-->...<!--googleon: snippet-->, and wonders if that's something that should be addressed by HTML properly

[11:17:00.0000] <Philip`>
Hmm, looks like it's only used by the Google Search Appliance, so it's not really a web thing

[11:58:00.0000] <takkaria>
hsivonen: did you use my dmoz.org fetcher thing at all or did you roll your own?

[12:58:00.0000] <jruderman>
aboodman!

[12:58:01.0000] <jruderman>
<3 greasemonkey

[13:41:00.0000] <billyjack>
hsivonen: you around?

[14:08:00.0000] <BenMillard>
gsnedders, did you talk to smedero?

[14:08:01.0000] <gsnedders>
/me looks innocent

[14:08:02.0000] <gsnedders>
No.

[14:08:03.0000] <smedero>
/me waves

[14:08:04.0000] <gsnedders>
smedero: Decide when you're going to be at the TPAC!

[14:08:05.0000] <smedero>
I'll be in France for the entire week.

[14:08:06.0000] <BenMillard>
smedero, I take it you and gsnedders will be sharing a room?

[14:09:00.0000] <smedero>
hrm, that seems to be the plan so far. What is the occupancy size of the room?

[14:09:01.0000] <smedero>
I'm all for cutting costs.

[14:09:02.0000] <gsnedders>
smedero: Yay! You've decided!

[14:10:00.0000] <jmb>
decisiveness is overrated

[14:10:01.0000] <BenMillard>
smedero, if it's OK I'd like to tag along with you two from Wedesnday to Friday

[14:10:02.0000] <BenMillard>
the form has a "Number of persons" field so I imagine there are 3-person rooms? http://www.w3.org/2008/03/TPAC2008-hotelform.html

[14:10:03.0000] <smedero>
of course.

[14:10:04.0000] <gsnedders>
BenMillard: I'm sure we can give you the floor.

[14:11:00.0000] <gsnedders>
Molly said there were three person rooms, FWIW

[14:11:01.0000] <BenMillard>
cool

[14:11:02.0000] <BenMillard>
I can bring a PS2 (PAL) with Gran Turismo games to sweeten the deal :)

[14:11:03.0000] <gsnedders>
I can bring some games too

[14:11:04.0000] <smedero>
That was my somewhat shoddy memory of the rooms situation as well...

[14:11:05.0000] <gsnedders>
/me has no memory of any TPAC :D

[14:11:06.0000] <BenMillard>
hotels usually have a TV and, since it's europe, I expect they'll have that 3-cable connection system

[14:11:07.0000] <BenMillard>
s/hotels/hotel rooms/

[14:12:00.0000] <gsnedders>
or SCART

[14:12:01.0000] <gsnedders>
who knows?

[14:12:02.0000] <smedero>
I stayed with friends at the last TPAC.... yay free.

[14:12:03.0000] <BenMillard>
I stayed in a hotel for the November 2007 one...paid for by Google :P

[14:12:04.0000] <gsnedders>
I couldn't get anywhere to stay last year :(

[14:12:05.0000] <gsnedders>
(Due to being under 18)

[14:12:06.0000] <gsnedders>
And everyone already having arrangements

[14:13:00.0000] <smedero>
I think we only met briefly Ben.... I saw your presentation on the table data and we we're probably in a few hallway chats... but I doubt you'd remember me.

[14:13:01.0000] <gsnedders>
If you want a room with three beds, which box do you tick?

[14:13:02.0000] <gsnedders>
smedero: What day are you arriving?

[14:14:00.0000] <BenMillard>
smedero, have you a photo online? that might job my memory...

[14:14:01.0000] <smedero>
Ideally Sunday... still working out Flights.

[14:15:00.0000] <smedero>
/me looks for the usual scruffy photo....

[14:15:01.0000] <gsnedders>
smedero: I'll just stay in Cannes for a day or two before I arrive

[14:16:00.0000] <BenMillard>
smedero & gsnedders, this is somewhat out of date but I've not changed much: http://projectcerbera.com/me/me.jpg

[14:16:01.0000] <smedero>
well, there's this I guess: http://www.gravatar.com/avatar/d527b871fc097b317f7993bdac0d349e?s=128

[14:16:02.0000] <gsnedders>
http://flickr.com/photos/tags/gsnedders/

[14:16:03.0000] <gsnedders>
Take your pick.

[14:16:04.0000] <BenMillard>
semedero, you look like Zane Lowe!

[14:16:05.0000] <zcorpan>
i've found a number of bugs in the validator.w3.org front-end of validator.nu

[14:17:00.0000] <gsnedders>
jgraham's is probably the best recent one

[14:17:01.0000] <smedero>
BenMillard: hahahaha

[14:17:02.0000] <gsnedders>
Expect my hair to be even crazier by October though.

[14:17:03.0000] <BenMillard>
gsnedders, ah the curly one

[14:18:00.0000] <smedero>
Ooers, to get the w3c discount I've got send that form soon.

[14:18:01.0000] <BenMillard>
I've not sent mine yet

[14:19:00.0000] <gsnedders>
BenMillard: your what?

[14:19:01.0000] <BenMillard>
the form smedero is talking about (I think): http://www.w3.org/2008/03/TPAC2008-hotelform.html

[14:19:02.0000] <gsnedders>
Ah

[14:19:03.0000] <smedero>
indeed

[14:19:04.0000] <gsnedders>
How do we deal with that?

[14:20:00.0000] <gsnedders>
Do we just send one, or…?

[14:20:01.0000] <BenMillard>
the homepage for the meeting is here: http://www.w3.org/2008/10/TPAC/Overview

[14:21:00.0000] <BenMillard>
smedero, I agree we should get this done sooner rather than later...I just saw "The discounted room rates are available until 8 September 2008."

[14:23:00.0000] <gsnedders>
hmm¬

[14:23:01.0000] <gsnedders>
*hmm…

[14:23:02.0000] <gsnedders>
"Rooms at this hotel can accommodate up to 2 person"

[14:23:03.0000] <BenMillard>
oh, where does it say that?

[14:23:04.0000] <gsnedders>
Trying to book a room directly at <http://www.pullmanhotels.com/pullman/fichehotel/gb/pul/1168/fiche_hotel.shtml>

[14:24:00.0000] <BenMillard>
Find in Page doesn't pick that text up for me..

[14:25:00.0000] <gsnedders>
Start to try and make a booking

[14:25:01.0000] <gsnedders>
Look at the first page

[14:25:02.0000] <BenMillard>
aha

[14:25:03.0000] <BenMillard>
http://www.pullmanhotels.com/gb/reservation/rooms-dates.jshtml

[14:26:00.0000] <gsnedders>
Hmm…

[14:26:01.0000] <BenMillard>
so much for that idea, then :(

[14:26:02.0000] <gsnedders>
"Unfortunately, no room with twin beds is available at this date. We suggest you consult the rates for a room with double bed."

[14:27:00.0000] <gsnedders>
Oddly, cheapest average room rate per night for six nights starting on the 19th, 149EUR, less than the W3C rate :\

[14:28:00.0000] <smedero>
heh

[14:28:01.0000] <gsnedders>
But that doesn't include breakfast

[14:28:02.0000] <Xenos>
And W3C breakfasts must be experienced to be believed ;-)

[14:28:03.0000] <gsnedders>
Inc. breakfast we're up to 189EUR

[14:29:00.0000] <gsnedders>
Still less than the W3C rate.

[14:29:01.0000] <Xenos>
They're totally worth 80 euros

[14:29:02.0000] <Xenos>
(The chef doesn't spit in those)

[14:30:00.0000] <gsnedders>
Xenos: :P

[14:30:01.0000] <BenMillard>
€362.33 per night (€1,089.40 total) is the best I can find for Wedesday 23rd to Friday 25th :(

[14:30:02.0000] <BenMillard>
directly at the hotel, that is

[14:31:00.0000] <gsnedders>
eeek.

[14:32:00.0000] <Xenos>
Man. If politics were as easy as getting "Insightful" on Slashdot I'd be world dictator in a week.

[14:32:01.0000] <zcorpan>
aaronlev: i tried to send an email to aria-ua-impl but it bounced

[14:33:00.0000] <Xenos>
... But then, so would everyone else, and we'd have to fight it out with tanks and nukes and stuff. So it might be just as well

[14:37:00.0000] <BenMillard>
gsnedders, good news is I have an adaptor for their plug type

[14:37:01.0000] <BenMillard>
train links seem adequate, so that's probably who I'll get there

[14:37:02.0000] <BenMillard>
*how

[14:37:03.0000] <BenMillard>
my mobile phone was brilliant in the USA, so I expect that will Just Work

[14:38:00.0000] <BenMillard>
and in Germany, too

[14:41:00.0000] <gsnedders>
BenMillard: It's not Britain, so of course train links are sane.

[14:45:00.0000] <Hixie>
hsivonen: i think it makes sense to have a middle ground with widely understood vocabularies where lying and abusing the structure can't cause much damage

[14:53:00.0000] <Hixie>
the responses in http://groups.google.com/group/microformats/browse_thread/thread/bd08f236b5dd1d13/08ea9b2b9f6dd911?hl=en&q=html5#08ea9b2b9f6dd911 are accurate

[14:58:00.0000] <gsnedders>
Does @id on input serve any use?

[14:59:00.0000] <BenMillard>
gsnedders, <label for>?

[15:00:00.0000] <gsnedders>
Ah.

[15:00:01.0000] <gsnedders>
I thought I was forgetting something obvious :)

[15:00:02.0000] <BenMillard>
sometimes as a styling hook in the absence of certain selectors in IE

[15:01:00.0000] <BenMillard>
or a scripting hook, equally

[15:03:00.0000] <gsnedders>
<textarea name="&#99;&#111;&#110;&#116;&#101;&#110;&#116;" id="&#99;&#111;&#110;&#116;&#101;&#110;&#116;">

[15:03:01.0000] <gsnedders>
weeee…

[15:03:02.0000] <gsnedders>
Taking advantage of the dumbness of spam-bots :)

[15:03:03.0000] <webben>
That may be over-optimistic.

[15:03:04.0000] <Xenos>
And making it real easy for people using screen readers :p

[15:03:05.0000] <BenMillard>
character reference munging is effective since spambots really *are* that dumb

[15:04:00.0000] <gsnedders>
webben: It isn't.

[15:04:01.0000] <BenMillard>
Nikita the Spider had a study on it, IIRC, trying to find it

[15:05:00.0000] <BenMillard>
aha: http://nikitathespider.com/articles/IngenReklamTack.html

[15:05:01.0000] <BenMillard>
"Obfuscation using Numeric Character References"

[15:05:02.0000] <gsnedders>
I don't know why I'm even bothering. form@action should be enough.

[15:06:00.0000] <webben>
BenMillard: interesting. last time I looked into obfuscation, I could swear I was reading about it failing.

[15:06:01.0000] <webben>
better than nothing either way though

[15:06:02.0000] <rubys>
Tiny nit, section 8.1.1; the doctype is case-insensitive in html5, but not in xhtml5.

[15:07:00.0000] <gsnedders>
rubys: 8.1 is specific to text/html

[15:08:00.0000] <Hixie>
i thought we said that explicitly somewhere, but i see we don't. i'll make a note to add a note saying that the section is for text/html only and xml rules are defined in xml.

[15:08:01.0000] <gsnedders>
smedero: When are you leaving?

[15:09:00.0000] <smedero>
Saturday, 10/25

[15:10:00.0000] <gsnedders>
smedero: ditto.

[15:10:01.0000] <BenMillard>
I anticipate sending my W3C reservation form tomorrow

[15:10:02.0000] <BenMillard>
(by plain text e-mail...ftw)

[15:12:00.0000] <gsnedders>
I take it it is one form per room, not per person?

[15:17:00.0000] <BenMillard>
gsnedders, there's a "sharing with" field and a "number of persons field" so that impression seems correct


2008-08-27
[00:27:00.0000] <krijnh>
Philip`: I have no idea (re the search.live.com referrers)

[02:35:00.0000] <hsivonen>
I got a huge email, so Mail.app has been busy with one message all morning

[02:36:00.0000] <hsivonen>
so I can't be distracted by the RDFa thread for a while still...

[02:36:01.0000] <Dashiva>
Well, unless you're a glutton for punishment... :)

[02:36:02.0000] <zcorpan>
hsivonen: great opportunity to get acctual work done :)

[02:37:00.0000] <Dashiva>
"We can't use classes because people might try to dereference them. Instead we use namespace URLs, which magically always can be dereferenced."

[02:37:01.0000] <annevk>
these "Digital Bazaar" people sure write a lot

[02:38:00.0000] <Hixie>
some of the needs of these rdfa people are things we should probably address

[02:38:01.0000] <Hixie>
i'm just not sure any of them actually require spec changes

[02:38:02.0000] <Dashiva>
"Only URIs are ever unique. A classname with all the same characters as a URI will magically cause collisions."

[02:38:03.0000] <Hixie>
but i'll get back to rdfa once i've dealt with the workers and parsing feedback

[02:38:04.0000] <hsivonen>
Dashiva: I have a solution for that one

[02:39:00.0000] <hsivonen>
Dashiva: defining that class names that contain a colon MUST be IRIs

[02:39:01.0000] <Dashiva>
"We know squat about why namespaces are bad."

[02:40:00.0000] <Dashiva>
hsivonen: That doesn't stop collisions, though. Anyone's free to type whatever they want

[02:40:01.0000] <jgraham>
Dashiva: I was going to ask how they plan to keep URIs both unique and dereferencable since it sems to me that atm having one property means that you cannot ensure the other, even assuming good faith

[02:40:02.0000] <Dashiva>
There is no URI police that will come to your house and stop you

[02:40:03.0000] <hsivonen>
Dashiva: sure

[02:40:04.0000] <Dashiva>
jgraham: Oh?

[02:40:05.0000] <hsivonen>
Dashiva: the same holds true for RDFa property attribute

[02:41:00.0000] <Dashiva>
hsivonen: Indeed

[02:41:01.0000] <jgraham>
Dereferenceable => URLs, right? and URLs imply DNS which has finite-time registrations.

[02:42:00.0000] <jgraham>
(in practice there are shorter term problems)

[02:42:01.0000] <Dashiva>
They work around that with purl et.al.

[02:42:02.0000] <annevk>
the long term problems are somewhat annoying too; if I die, will I need to set up a fund to keep annevankesteren.nl alive?

[02:42:03.0000] <Dashiva>
Yes

[02:43:00.0000] <jgraham>
Dashiva: So they basically build a second centralised registry 

[02:43:01.0000] <hsivonen>
who is the purl overlord?

[02:43:02.0000] <Dashiva>
jgraham: Ssh, it's impolite to point out elephants

[02:43:03.0000] <jgraham>
In which case why bother putting http://purl.org at the front of everything

[02:43:04.0000] <jgraham>
Why not just have unique human-readable prefixes

[02:44:00.0000] <hsivonen>
jgraham: well, purl.org NS uris depend on 3 centralized registries that are independent, so it's decentralized in that sense

[02:45:00.0000] <hsivonen>
billyjack: I'm around now

[02:45:01.0000] <jgraham>
hsivonen: what are the 3 independent registries? Presumably there is only one body handling out purl.org uris, so purl.org, not dns becomes the bottleneck in that case

[02:45:02.0000] <jgraham>
?

[02:46:00.0000] <hsivonen>
jgraham: purl.org itself (for its subspace), domain registry and the URI scheme registry

[02:47:00.0000] <billyjack>
hsivonen: can't remember what I wanted to ask you about now.. will ping you if/when I remember

[02:47:01.0000] <hsivonen>
billyjack: ok

[02:47:02.0000] <jgraham>
hsivonen: I think that's just three things that could go wrong, not three layers of redundancy

[02:47:03.0000] <hsivonen>
jgraham: yes. but the three things are distributed :-)

[02:48:00.0000] <jgraham>
"distributed modes of failiure" doesn't sound that confidence inspiring :)

[02:52:00.0000] <Hixie>
jgraham: i don't think hsivonen is defending RDFa :-)

[02:53:00.0000] <annevk>
RDF sounds very nice in theory though, still

[02:53:01.0000] <Dashiva>
annevk's question was interesting, though. Would they actually endorse a situation where an entire vocabulary becomes unusable because someone takes their webpage down?

[02:54:00.0000] <hsivonen>
I'd love to have a unified parsing model for microformats but without namespaces, CURIEs or overlong identifiers

[02:54:01.0000] <hsivonen>
my guess is that two to four letter prefixes disambiguate enough

[02:54:02.0000] <hsivonen>
IIRC, in the wikimedia commons SVG analysis, the namespace prefix collisions were:

[02:55:00.0000] <hsivonen>
1) autogenerated ns1

[02:55:01.0000] <hsivonen>
2) cc changing its URI

[02:55:02.0000] <hsivonen>
3) Adobe and Microsoft claiming the same one-letter prefix

[02:56:00.0000] <annevk>
so unified parsing means you can easily create a graph out of some nodes?

[02:56:01.0000] <Hixie>
i love that the response to 2) is "but because you can dereference it, it's ok! we can map them to each other!"

[02:56:02.0000] <annevk>
because after that you will still have to perform format specific operations, which are probably more complicated than the parsing

[02:57:00.0000] <Hixie>
the syntax introduces a problem that is solved by the syntax, i love it

[02:57:01.0000] <hsivonen>
annevk: I'd settle for a tree

[02:57:02.0000] <Hixie>
annevk: if you end up with a triple-store, you can use standard tools to do inference

[02:58:00.0000] <Hixie>
annevk: which supports things like mapping vocabularies to each other, or saying that anything that is a car is also a four-wheeled object

[02:58:01.0000] <Hixie>
annevk: so even if your tool originally only knew about wheels, it could work out that each car implied four weels

[02:58:02.0000] <annevk>
yeah, but you still need to create the "export to Vcard" function

[02:58:03.0000] <Hixie>
yeah well, i'm not sure i understand what the ui would really be

[02:59:00.0000] <annevk>
I don't think it solves that part (although some people suggest it does)

[02:59:01.0000] <hsivonen>
Hixie: one semweb enthusiast I used to have lunches with occasionally at least used to advocate a UI like NakedObjects

[02:59:02.0000] <hsivonen>
http://www.nakedobjects.org/home/index.shtml

[03:00:00.0000] <hsivonen>
(I don't think that makes a good UI, though.)

[03:01:00.0000] <hsivonen>
/me fails to find Havoc Pennington's essay that identifies "I can generate the UI" as a passing phase

[03:02:00.0000] <hsivonen>
here it is: http://www106.pair.com/rhp/free-software-ui.html

[03:04:00.0000] <jgraham>
hsivonen: Yeah, every time I hear about RDFa I think of that...

[03:07:00.0000] <Philip`>
A library that does everything your applications need except for generating the UI still sounds like a quite useful thing to have

[03:08:00.0000] <jgraham>
The other question I didn't ask about RDFa is why it will work when rel="home", "next" etc. have effectively failed

[03:09:00.0000] <Philip`>
What does "failed" mean?

[03:10:00.0000] <Philip`>
/me has never noticed them causing harm, and has found them to be useful when he uses them on his own pages

[03:10:01.0000] <jgraham>
Philip`: They aren't used that much and IIRC aren't supported by default in 99% of browsers

[03:11:00.0000] <jgraham>
In fact you even have to turn on support in Opera I think

[03:11:01.0000] <Philip`>
They're on by default in Opera

[03:11:02.0000] <Philip`>
or at least the 'next' one is

[03:13:00.0000] <jgraham>
I think mpt was right that the problem is that because they are only used for 1 in n pages people don't learn to look in the chrome for functions related to intra-site navigation

[03:13:01.0000] <Philip`>
(e.g. if I go to http://philip.html5.org/tests/canvas/suite/tests/fallback.multiple.html and use a 'forward' shortcut (ctrl+right, right-click-and-drag-rightwards, etc) then it goes to the next page)

[03:14:00.0000] <hsivonen>
feed autodiscovery got UI

[03:14:01.0000] <hsivonen>
rel=next is different from the rest

[03:15:00.0000] <mpt>
That never occurred to me before -- how did feed autodiscovery succeed while rel= mostly failed?

[03:15:01.0000] <jgraham>
Philip`: You're right of course.

[03:16:00.0000] <mpt>
or I should say, <link rel= failed -- <a rel= is useful, because you're not relying on browsers to do anything special

[03:16:01.0000] <mpt>
it's an optional extra, like microformat detection

[03:17:00.0000] <annevk>
http://ajaxian.com/archives/ubiquity-quicksilver-of-the-firefox-browser could benefit from RDFa annotated data

[03:17:01.0000] <wilhelm>
Philip`: You can also just press space on that page. It scrolls down, and when it can't scroll anymore, it goes to the next page.

[03:17:02.0000] <Philip`>
(I think I've seen Opera's next-page thing working far more often when it automatically determines an ordered list from Apache directory indexes and lets you jump from one file straight to the next in the directory, rather than from explicit rel=nexts)

[03:22:00.0000] <hsivonen>
wilhelm: that feature is awesome. (which is why I think rel=next is not like the rest)

[03:23:00.0000] <webben>
I guess feed discovery progressed for on-page links to not relying on-page links.

[03:23:01.0000] <webben>
*from on-page

[03:23:02.0000] <webben>
maybe if <a rel="next" and associated UI became commonplace

[03:23:03.0000] <annevk>
rel=next also causes a page to be prefetched in Firefox

[03:23:04.0000] <webben>
then <link rel="next" would work too

[03:23:05.0000] <wilhelm>
<link rel='next'> is a pain, though. <a rel='next'> would be much nicer.

[03:23:06.0000] <jgraham>
mpt: I would hazard a guess that it has something to do with the fact that you can predict which sites have feeds and the fact that they used the same icon that people had got used to on-page offpage

[03:23:07.0000] <annevk>
(similar to rel=prefetch)

[03:24:00.0000] <webben>
wilhelm: are feed icons nicer than feed autodiscovery?

[03:30:00.0000] <annevk>
Does anyone know how Firefox JIT compares with WebKit Squirrelfish?

[03:32:00.0000] <annevk>
seems to be faster

[03:33:00.0000] <Philip`>
Apparently it wins on SunSpider

[03:33:01.0000] <annevk>
right

[03:34:00.0000] <Philip`>
Meanwhile, the IE team has a blog post saying that JScript performance isn't that important really

[03:36:00.0000] <hsivonen>
Philip`: even though IE seems to need the most JS libraries between it and the app programmer :-)

[03:36:01.0000] <Philip`>
/me likes how http://home.kairo.at/blog/2008-08/how_fast_is_tracemonkey_in_real_world measures "stuff a real application might use" by calculating Mandelbrot fractals

[03:37:00.0000] <hsivonen>
/me wishes Apple shipped GNU versions of basic Unix tools

[04:10:00.0000] <hsivonen>
aargh. no wonder there were too many unique messages. part of the consolidation had failed

[04:10:01.0000] <hsivonen>
sigh

[04:15:00.0000] <Hixie>
man, justin's comments on workers were disturbing

[04:16:00.0000] <Hixie>
he had one example where he launched one worker PER PIXEL of a bitmap he wanted to process

[04:16:01.0000] <Hixie>
i'm scared of what authors will do now

[04:25:00.0000] <hsivonen>
hmm. I think Mail.app has just gone crazy on its own

[04:25:01.0000] <Lachy>
Hixie, in the note at the end of the alt text guidelines, "One way to think of alternative text is to think about what how you would read the page containing the image " - s/what how/how/

[04:26:00.0000] <Hixie>
please send mail, i'm about to go to bed

[04:26:01.0000] <Hixie>
just a one liner to ian⊙hc is fine

[04:26:02.0000] <Hixie>
btw thanks again for that program

[04:26:03.0000] <Hixie>
whose name escapes me

[04:27:00.0000] <Lachy>
Requiem

[04:27:01.0000] <Hixie>
i have however found one major flaw with it

[04:27:02.0000] <Hixie>
right

[04:27:03.0000] <Lachy>
what?

[04:27:04.0000] <Hixie>
requiem

[04:27:05.0000] <Hixie>
after i unencrypted everything

[04:27:06.0000] <Lachy>
what flaw?

[04:27:07.0000] <Hixie>
i mentioned it to my girlfriend 

[04:27:08.0000] <Hixie>
and her response was

[04:27:09.0000] <Hixie>
"oh! that means i can download DRMed songs again right?"

[04:28:00.0000] <Lachy>
heh

[04:28:01.0000] <Hixie>
i didn't really have a good reply!

[04:28:02.0000] <Hixie>
i mean, i don't want to encourage iTunes to continue with DRM

[04:29:00.0000] <Lachy>
just say no, because spending money on music with DRM only funds further development of DRM

[04:29:01.0000] <Hixie>
but it's hard to argue that the DRM is a problem if it can just be removed

[04:29:02.0000] <Lachy>
sure it can be removed, but it's illegal thanks to the DMCA

[04:29:03.0000] <Hixie>
i have a little cron job now that scans my library and decrypts anything that's newly added with drm and moves the old drm'ed file out of the way and renames the new file, all automatically

[04:29:04.0000] <Hixie>
it's pretty sweet

[04:30:00.0000] <Hixie>
the "breaking encryption is illegal" aspect of the DMCA is inane

[04:31:00.0000] <webben>
OT: is there a canonical straightforward explanation of the limitations of DTDs for describing the syntax of XML vocabularies (versus schemas of XSD or RNG varieties, or anything else for that matter)?

[04:31:01.0000] <Hixie>
webben: DTDs can't describe anything but trivial attribute value formats

[04:32:00.0000] <hsivonen>
webben: not canonical, but there's http://hsivonen.iki.fi/thesis/html5-conformance-checker#dtd

[04:33:00.0000] <Hixie>
anyway

[04:33:01.0000] <Hixie>
nn

[04:33:02.0000] <hsivonen>
nn

[04:37:00.0000] <hsivonen>
/me restores Mail and AddressBook-related ~/Library files from backup

[04:37:01.0000] <hsivonen>
/me wishes Mail.app were more robust

[04:41:00.0000] <webben>
hsivonen: ta :)

[05:00:00.0000] <hsivonen>
zcorpan: multiline scripts now work in the HTML5 live DOM

[05:22:00.0000] <zcorpan>
hsivonen: cool

[05:22:01.0000] <zcorpan>
hsivonen: now the question is whether we should change to match webkit and firefox or if they should change to match us... consider that %0D%0A is what is sent to the server

[05:23:00.0000] <hsivonen>
what does IE do?

[05:23:01.0000] <zcorpan>
IE uses \r (but sends %0D%0A to the server)

[05:24:00.0000] <hsivonen>
CRLF sucks, and LF is more canonical than CR, so I'd prefer the Gecko/WebKit way

[05:24:01.0000] <hsivonen>
without proper compat arguments

[05:24:02.0000] <zcorpan>
we haven't had any compat bugs on this until livedom

[05:24:03.0000] <zcorpan>
so either way should work fine

[05:25:00.0000] <annevk>
path of least resistance seems for us to match Gecko/WebKit

[05:25:01.0000] <zcorpan>
yeah

[05:28:00.0000] <hsivonen>
hmm. myspace localizers have localized id=bodytype (or something like that) to id=Vartalon tyyppi without quotes

[05:28:01.0000] <hsivonen>
wow. MySpace has totally clueless id attribute use

[05:29:00.0000] <hsivonen>
it's a wonder their markup "works" at all

[06:06:00.0000] <zcorpan>
Hixie: would it be possible to include an Archived-At or X-Archived-At header in whatwg list emails?

[06:06:01.0000] <annevk>
he's just running some package from dreamhost

[06:11:00.0000] <zcorpan>
does that mean "no"?

[06:12:00.0000] <annevk>
likely

[06:12:01.0000] <annevk>
(it has been asked before, too)

[06:13:00.0000] <zcorpan>
ok

[06:14:00.0000] <Philip`>
/me sees some discussion of it in http://wiki.list.org/display/DEV/Stable+URLs

[06:16:00.0000] <annevk>
/me was just about to past that

[06:17:00.0000] <annevk>
seems a non-trivial problem

[06:20:00.0000] <annevk>
s/past/paste/

[06:38:00.0000] <Philip`>
Hmm, my experiments indicate that Googlebot is surprisingly limited in where it looks for URLs - I only see <a href>, <embed src> and <script src>

[06:38:01.0000] <Philip`>
Yahoo looks at <a href>, <a src> (?!), <embed src>, <link href> and <object data>

[06:38:02.0000] <Philip`>
msnbot only looks at <a href>; msnbot-media also does <embed src> and <img src>

[06:38:03.0000] <Philip`>
Heritrix looks at <* cite>, <* data>, <* href>, <* lowsrc>, <* profile>, <* src> for any *

[06:40:00.0000] <Philip`>
(Of course they might be doing sufficiently clever things to break my experiments)

[06:56:00.0000] <annevk>
hmm, #html-wg is starting to look like public-html :)

[07:04:00.0000] <hsivonen>
Hixie: I'm seeing a number of pages that have not parse errors. So far, I'm not seeing pages that don't have validation errors

[07:08:00.0000] <Philip`>
Will you be able to easily determine how many don't have validation errors except for some given set of errors?

[07:08:01.0000] <hsivonen>
Philip`: "easily" will depend of the ratio of RAM and data after the validation run

[07:09:00.0000] <hsivonen>
Philip`: but grep -v should work

[07:09:01.0000] <hsivonen>
I did a couple of things that will drastically reduce the data

[07:10:00.0000] <hsivonen>
I made the validation worker consolidate "bad value 'foo'" messages properly. (it had a bug last night)

[07:10:01.0000] <hsivonen>
and I picked at most one URL per hostname

[07:21:00.0000] <hsivonen>
is there an HTML 4 "community" now?

[07:21:01.0000] <hsivonen>
"community" sounds even more hollow these days than "economy"

[07:32:00.0000] <zcorpan>
aaronlev: hey

[07:33:00.0000] <zcorpan>
bah he always disconnects when i ping him :(

[08:12:00.0000] <zcorpan>
aaronlev: ping

[08:13:00.0000] <Philip`>
A lot of people seem to be complaining about how IE8's InPrivate Blocking mode (which blocks third-party items that are included on more than ten different sites you visit) is going to break analytics and advertising and will cause the web industry to collapse

[08:14:00.0000] <Philip`>
I guess they're quite significantly distinct from the people who complain about <a ping>

[08:15:00.0000] <aaronlev>
hi zcorpan 

[08:15:01.0000] <zcorpan>
aaronlev: i tried to email aria-ua-impl last night

[08:15:02.0000] <zcorpan>
aaronlev: but it bounced

[08:16:00.0000] <aaronlev>
did you join?

[08:17:00.0000] <zcorpan>
yeah

[08:18:00.0000] <zcorpan>
aaronlev: are the logs public?

[08:21:00.0000] <aaronlev>
zcorpan: i don't know

[08:21:01.0000] <aaronlev>
i can make you an administrator

[08:21:02.0000] <aaronlev>
give me a minute

[08:42:00.0000] <takkaria>
hsivonen: oh, what's your more interesting URL set?

[08:54:00.0000] <gsnedders>
benoitc: My navigating skillz are teh 1337.

[08:59:00.0000] <gsnedders>
benoitc: sorry, wrong person.

[09:00:00.0000] <benoitc>
np

[09:19:00.0000] <annevk>
we really need better <input type=file> + XMLHttpRequest support (or async <input type=file max=100>)

[09:20:00.0000] <annevk>
having to use the commandline to upload Flickr photos easily under Ubuntu/Mac is just silly

[09:46:00.0000] <takkaria>
does no-one else thing that role="photo" is a mad suggestion?

[09:47:00.0000] <takkaria>
role="logo" or role="drawing"

[09:47:01.0000] <takkaria>
since they're not roles, blatantly, they're descriptions

[09:48:00.0000] <takkaria>
a photo could be role="decorative" but role="photo" seems a really jarring misuse of the word "role"

[09:48:01.0000] <Dashiva>
It seems more like a subtle attempt at sabotaging role

[09:48:02.0000] <takkaria>
mind you, all this role business for img elements is astronauting anyway

[09:59:00.0000] <gsnedders>
gsnedders's advice for today: Don't fall on to solid wood floors.

[10:13:00.0000] <Xenos>
Concrete floors arn't good either, I hear

[10:14:00.0000] <Philip`>
Best not to fall on any floor at all, in my experience

[10:16:00.0000] <gsnedders>
normalize-space() is expensive.

[10:31:00.0000] <annevk>
whoa, flood of RDF e-mail

[10:35:00.0000] <annevk>
though a lot of it is due to the dubious cross-list posting (www-archive and whatwg)

[10:40:00.0000] <annevk>
wow, my own e-mail to an internal mailing list got marked as spam

[12:07:00.0000] <annevk>
For those with W3C Member access: http://lists.w3.org/Archives/Member/tag/2008Aug/thread.html#msg67

[12:12:00.0000] <malde_>
Hi,

[12:13:00.0000] <malde_>
is my reading of the spec correct that within a sigle window only one transaction can execute simultaneously?

[12:13:01.0000] <Dashiva>
"The concept of namespaces is something that most regular folks can grasp if you explain it to them in the correct way."

[12:13:02.0000] <Dashiva>
How many understand namespaces, and how many just understand magical prefixes?

[12:15:00.0000] <annevk>
malde_, given that two windows can interact with the same database I don't think that's true

[12:15:01.0000] <malde_>
annevk: I was asking about a single window

[12:16:00.0000] <annevk>
I realize that and I don't think it's something that goes per window given that multiple windows can interact with the same database

[12:16:01.0000] <annevk>
(but I haven't checked that part recently)

[12:16:02.0000] <malde_>
like in db.transaction(...);db.transaction(); the second transaction wont be executed until all callbacks of the first transaction fired

[12:16:03.0000] <malde_>
It seems that webkit implements it that way

[12:17:00.0000] <annevk>
that sounds correct, yes

[12:17:01.0000] <Hixie>
malde_: it's per database, not per window

[12:18:00.0000] <annevk>
morning

[12:18:01.0000] <malde_>
OK, so it should be save to store the transaction handle inside a global variable?

[12:19:00.0000] <Hixie>
malde_: i'm not sure i'd recommend that, but it depends on what you're doing really

[12:19:01.0000] <Hixie>
zcorpan: in case you read this, i'd love to be able have X-Archived-At (i suggestd it for w3c originally) but i'm just using old stock mailman that dreamhost runs

[12:20:00.0000] <malde_>
I wrote an OR mapper that work with both html5 db and gears and I wont to get the tx out of the ORM's API

[12:20:01.0000] <annevk>
Hixie, it's standardized as Archived-At now, fwiw

[12:20:02.0000] <annevk>
(and W3C switched to using that)

[12:22:00.0000] <jgraham>
Dashiva: I thought "We realized that it would be much easier for people to understand"

[12:22:01.0000] <jgraham>
"namespaces if we used URIs as the method of namespace expression"

[12:22:02.0000] <Dashiva>
So "Namespaces are much easier to understand if we ignore that pesky indirection thing"

[12:23:00.0000] <jgraham>
Dashiva: I think the point was "Namespaces are much easier to understand if we use URIs to denote the namespace rather than human-readable strings"

[12:24:00.0000] <Dashiva>
And then bring CURIEs into it

[12:25:00.0000] <Philip`>
Dashiva: Regular folks can use the magical prefixes that they copy-and-paste from the instructions given to them by the namespace experts who understand how to use the power and flexibility to provide all the desired features and avoid collisions and everything

[12:26:00.0000] <Philip`>
But I don't think that reasoning works, because there are far too few experts, and so regular folks make it up as they go along and get it all wrong

[12:28:00.0000] <jgraham>
Hmm. the RDFa emails seem to se using URI and URL interchangably in a situation where the difference is important

[12:32:00.0000] <Dashiva>
Philip`: And we all know the non-experts would never forget to copy the enclosing element defining the CURIE prefixes :)

[12:33:00.0000] <Philip`>
Dashiva: The same non-experts who write XHTML pages with xmlns="http://www.w3.org/1999/xphp"? Surely not!

[12:45:00.0000] <hsivonen>
takkaria: the more interesting set is a set of URL from Google that shouldn't have the kind of front page bias dmoz likely has

[12:45:01.0000] <hsivonen>
s/URL/URLs/

[12:45:02.0000] <takkaria>
ah, I see :)

[12:46:00.0000] <hsivonen>
/me notes that the RDFa examples omit ns declarations

[12:48:00.0000] <annevk>
and even when they omit those they still look far more complex than the microformats they try to emulate

[12:51:00.0000] <Philip`>
That's a problem with designing scalable solutions: you want examples to be as small as possible, and much smaller than real-world uses are likely to be, so the scalability is not apparent to people who only look at the examples

[13:33:00.0000] <gsnedders>
IE8b2 is available, FWIW

[13:37:00.0000] <Philip`>
Ooh, a bug-hunting opportunity

[13:43:00.0000] <Philip`>
Seems unusual for them to talk about improving the performance of Gmail, rather than of any Microsoft site

[13:56:00.0000] <gsnedders>
I have only one item on my to-do list left for spec-gen 1.0RC1

[13:56:01.0000] <gsnedders>
checking all the docs are right.

[13:57:00.0000] <hsivonen>
hmm. IE8b2 EULA prohibits publication of benchmark results

[13:57:01.0000] <hsivonen>
that sucks

[13:58:00.0000] <webben>
seems a bit harsh

[13:58:01.0000] <webben>
(also seems highly unlikely to be an effective deterrent)

[13:58:02.0000] <gsnedders>
/me thinks hsivonen complained about that when IE8b1 was released (with the slight change of the IE8b1 EULA being the EULA in question)

[13:58:03.0000] <hsivonen>
webben: harsh to prohibit it or harsh to say it sucks?

[13:58:04.0000] <webben>
I mean the prohibition is harsh.

[13:59:00.0000] <webben>
Seems fair to say it sucks. :)

[14:00:00.0000] <Philip`>
VMware from version 4 (I think) prohibits distribution of benchmark results too, but then the VMware people complained that it was unfair when other virtualisation people distributed benchmarks comparing against VMware 3

[14:01:00.0000] <Philip`>
which is kind of their own fault

[14:02:00.0000] <hsivonen>
/me wonders if one presses cancel to the IE8b2 install when a cancel button is offered after the installer has said it removed previous IE

[14:02:01.0000] <hsivonen>
/me wonders what happens if..

[14:05:00.0000] <Philip`>
/me had to manually uninstall IE8b1 and reboot and then see Windows fail to start up and offer to restart in safe mode but then it worked the next time

[14:06:00.0000] <hsivonen>
to me, it seems like a bad idea to offer a cancel button in the second phase of a rebooting installer

[14:07:00.0000] <hsivonen>
whee. a second reboot

[14:08:00.0000] <Philip`>
It's rather odd that in these modern times it can take twenty minutes to install a web browser

[14:08:01.0000] <Dashiva>
Maybe it's more than just a browser :)

[14:09:00.0000] <annevk>
if only hardware improved installing a browser sure would be fast

[14:10:00.0000] <hsivonen>
What Gordon giveth, Bill taketh.

[14:11:00.0000] <hsivonen>
eww. Validator.nu is ugly in IE8b2

[14:11:01.0000] <gsnedders>
hsivonen: can you see if http://gsnedders.com is still badly screwed up?

[14:12:00.0000] <hsivonen>
gsnedders: it isn't

[14:12:01.0000] <hsivonen>
except Windows font rendering is hideous

[14:12:02.0000] <hsivonen>
(and some people wanted Safari on Windows to use it)

[14:13:00.0000] <Hixie>
so does ie8 work in the live dom viewer yet

[14:13:01.0000] <hsivonen>
Hixie: it at least doesn't render the link to it on damowmow portal

[14:14:00.0000] <hsivonen>
Hixie: it works!

[14:16:00.0000] <Hixie>
woot

[14:16:01.0000] <Philip`>
"Do you want to discover websites you might like based on websites you've visited?" - hmm

[14:17:00.0000] <Philip`>
Things like "<b style>" still don't show the attribute in the DOM view, though

[14:18:00.0000] <Hixie>
is that because the attribute isn't there?

[14:18:01.0000] <Hixie>
i just want it to reflect the dom

[14:18:02.0000] <Philip`>
Hmm, it works if you do <b style=color:black>

[14:19:00.0000] <Philip`>
where "works" means it shows B style="COLOR: black" in the DOM view

[14:19:01.0000] <Hixie>
ok i need to eat before i die of starvation

[14:19:02.0000] <Hixie>
and then i guess i'll respond to this svg proposal for parsing

[14:19:03.0000] <Philip`>
but <b style=color:blac> makes the attribute disappear

[14:19:04.0000] <Hixie>
Philip`: that's probably an accurate representation of the DOM

[14:21:00.0000] <annevk>
IE has some weird stuff going on

[14:22:00.0000] <annevk>
(though it probably made sense at the time (I hope))

[14:26:00.0000] <Philip`>
Hmm, I can't find any situations in which authoritative=true does anything at all

[14:31:00.0000] <Philip`>
(nor does mshtml.dll contain the string "authoritative", so maybe they just haven't implemented it?)

[14:32:00.0000] <Philip`>
(unless it's implemented in some other component of Windows)

[14:32:01.0000] <Philip`>
(but it still doesn't seem to do anything at all in practice)

[14:36:00.0000] <gsnedders>
/me fails to find any further issues blocking release of the spec-gen

[14:38:00.0000] <Philip`>
gsnedders: It needs a clever name before it's released

[14:39:00.0000] <Philip`>
"spec-gen" is a bit generic :-p

[14:39:01.0000] <gsnedders>
:P

[14:39:02.0000] <annevk>
also, drop the hyphen

[14:40:00.0000] <annevk>
specgen is easier

[14:41:00.0000] <annevk>
(you could call it "chameleon" referring to how you mimicked the original)

[14:45:00.0000] <gsnedders>
annevk: chameleon is taken :P

[14:45:01.0000] <annevk>
not for spec generators

[14:46:00.0000] <annevk>
I should ask, what do you mean with "taken"?

[14:46:01.0000] <gsnedders>
annevk: Used by other software

[14:47:00.0000] <annevk>
"Chameleon Specgen" isn't

[14:47:01.0000] <annevk>
http://www.google.com/search?q=%22chameleon%20specgen%22

[14:47:02.0000] <annevk>
though that's not really a good name :)

[14:47:03.0000] <gsnedders>
Anole?

[14:48:00.0000] <gsnedders>
Used too, actually

[14:48:01.0000] <annevk>
call it Anolis then or something like that

[14:49:00.0000] <gsnedders>
Inkfish?

[14:51:00.0000] <BenMillard>
how about Geoffrey's Specification Generator (GGen/ggen)?

[14:51:01.0000] <annevk>
I don't really see the relation

[14:52:00.0000] <annevk>
"ggen is a 2-D geometry generator expressed as a c++ class library."

[14:52:01.0000] <Philip`>
Use the Hebrew alphabet for the name

[14:53:00.0000] <gsnedders>
BenMillard: Re: you're ability to navigate — I can fine. We should make it there and back fine :)

[14:53:01.0000] <Philip`>
(Er, not the whole alphabet, just some letters from it)

[14:53:02.0000] <BenMillard>
gsnedders, coolies :)

[14:53:03.0000] <BenMillard>
gsnedders, I take it we'll stay at the hotel annevk did 2 years ago?

[14:54:00.0000] <gsnedders>
BenMillard: What hotel did annevk stay at two years ago? :)

[14:54:01.0000] <BenMillard>
gsnedders, oh, forwarded to you

[14:55:00.0000] <Lachy>
Philip`, do you mean something like this? א

[14:55:01.0000] <Lachy>
no idea what that means, but it's a hebrew character aparently

[14:55:02.0000] <annevk>
oops, forgot to cc gsnedders

[14:56:00.0000] <gsnedders>
/me feels too lazy to bother to click the "English" link

[14:56:01.0000] <gsnedders>
BenMillard: We have our TV for the PS2! :)

[14:57:00.0000] <BenMillard>
gsnedders, good, the trip will be worthwhile after all! :D

[14:57:01.0000] <gsnedders>
umm¬

[14:57:02.0000] <gsnedders>
*umm…

[14:57:03.0000] <gsnedders>
"3eme personne dans la chambre, (sans lit) "

[14:57:04.0000] <gsnedders>
So, what? You sleep on the floor?

[14:57:05.0000] <gsnedders>
(that means three people in the bedroom, without bed)

[14:58:00.0000] <gsnedders>
/me wonders if we could really book online when we only want the extra bed for three nights

[14:59:00.0000] <gsnedders>
heh. They're English is in parts on the website a literal translation from the French and odd :)

[14:59:01.0000] <gsnedders>
*Their

[14:59:02.0000] <BenMillard>
looking here, I see what you mean: http://www.villaparisiana.com/price.htm

[15:00:00.0000] <BenMillard>
"3rd person in room, (without bed) : 6 euros"

[15:00:01.0000] <gsnedders>
But also, "Lit supplémentaire"

[15:00:02.0000] <gsnedders>
(extra bed)

[15:00:03.0000] <gsnedders>
(however they translate that)

[15:00:04.0000] <BenMillard>
"Extra bed : 15 euros"

[15:00:05.0000] <BenMillard>
win!

[15:00:06.0000] <gsnedders>
totally.

[15:00:07.0000] <annevk>
just means you pay EUR 21 for the third person

[15:00:08.0000] <gsnedders>
BenMillard: Can you justify the cost of coming for a whole week now? :P

[15:01:00.0000] <annevk>
I suggest taking the breakfast as well

[15:01:01.0000] <gsnedders>
/me will undoubtedly get up too late for it :)

[15:02:00.0000] <gsnedders>
Last hotel I was at, I made breakfast zero out of three times.

[15:02:01.0000] <BenMillard>
I absolutely *must* eat a breakfast; ideally hot chocolate and cereal with a banana

[15:02:02.0000] <gsnedders>
BenMillard: Now, do we try and get three people in a small room or not?

[15:03:00.0000] <annevk>
I usually miss a few minutes of the relevant meeting :)

[15:03:01.0000] <gsnedders>
/me had no breakfast today :P

[15:03:02.0000] <BenMillard>
gsnedders, I don't think we need to slum it too badly

[15:03:03.0000] <gsnedders>
/me got into school late anyway

[15:03:04.0000] <gsnedders>
Nor do I :)

[15:04:00.0000] <gsnedders>
I mean, even the big rooms are cheaper than half what you'd pay sharing a room at the TPAC hotel

[15:04:01.0000] <BenMillard>
gsnedders, rooms don't look amazingly generous: http://www.villaparisiana.com/rooms.htm

[15:04:02.0000] <BenMillard>
so their idea of "small" might be "cupboard"

[15:04:03.0000] <gsnedders>
BenMillard: That was my thought :)

[15:04:04.0000] <BenMillard>
medium or large sounds good to me

[15:04:05.0000] <gsnedders>
We're middle/low season

[15:04:06.0000] <gsnedders>
(as we're there over the change)

[15:05:00.0000] <BenMillard>
gsnedders, yeah

[15:05:01.0000] <gsnedders>
/me still hasn't bothered clicking that English link

[15:05:02.0000] <BenMillard>
as well as 3 people (I'm about 6ft tall) we also must consider 3 people's bags of stuff

[15:06:00.0000] <BenMillard>
you two are there for like a week, so that volume may be non-trivial

[15:06:01.0000] <gsnedders>
And I'm away from home for two weeks

[15:06:02.0000] <BenMillard>
I'm thinking Large Room to be safe (and their idea of "Large" may not be that big anyway)

[15:06:03.0000] <BenMillard>
plus we're adding an extra bed

[15:06:04.0000] <gsnedders>
(Though that won't be much issue — I'll be staying with my uncle, so washing and the like can be done)

[15:06:05.0000] <gsnedders>
(so I only really need a week's stuff)

[15:07:00.0000] <BenMillard>
gsnedders, Large Room might get a better TV and more power outlets, too

[15:09:00.0000] <gsnedders>
528EUR in total, by my calculation

[15:09:01.0000] <gsnedders>
(big room, inc. breakfast)

[15:09:02.0000] <gsnedders>
(exc. the 0.40 EUR tax per day per person)

[15:09:03.0000] <gsnedders>
EUR 176 if we split it equally, which is still less than one night in the TPAC hotel.

[15:09:04.0000] <gsnedders>
Heh.

[15:10:00.0000] <gsnedders>
BenMillard: Power outlets may be useful with three geeks :)

[15:10:01.0000] <BenMillard>
gsnedders, I have a mobile phone which eats battery life but no laptop

[15:10:02.0000] <gsnedders>
BenMillard: I have a mobile phone which doesn't eat battery life and a laptop

[15:10:03.0000] <gsnedders>
(which does)

[15:11:00.0000] <BenMillard>
gsnedders, splitting the price evenly seems fair enough to me, as I'll be making the room more cramped for the 2 nights I'm there, you'll be doing the meatspace navigation for me and I'm not footing the bill personally

[15:11:01.0000] <gsnedders>
BenMillard: :)

[15:11:02.0000] <BenMillard>
plus I'm bound to whip you at GT and make you cry

[15:12:00.0000] <BenMillard>
:P

[15:13:00.0000] <gsnedders>
BenMillard: If I get back in practice at playing such games on the PS2… :)

[15:13:01.0000] <gsnedders>
/me has been playing Forza 2 on 360 a lot recently

[15:13:02.0000] <gsnedders>
it takes me a week or two to get fully up to speed on the PS2 controller

[15:13:03.0000] <gsnedders>
So, I guess I could start practising… :)

[15:14:00.0000] <gsnedders>
BenMillard: On Forza 2, in the time trial rankings, I'm #8xx, so rather high up :)

[15:16:00.0000] <BenMillard>
gsnedders, we can use different tyres and classes of car to balance things out, handicap starts, etc

[15:18:00.0000] <Xenos>
After the fifth day of PS2, smoking pot and eating junk food, someone goes "Wait, wasn't there a meeting we were supposed to go to?"

[15:19:00.0000] <BenMillard>
gsnedders, looking at the booking form, it seems 1 person books for everyone and we balance the money out afterwards? http://www.villaparisiana.com/contact_e.htm

[15:19:01.0000] <BenMillard>
("Number of persons: 3" and "Number of rooms: 1")

[16:15:00.0000] <gsnedders>
BenMillard: different classes of cars has its limits. In GT3, I once beat a friend, driving an F1 car, with a Mini.

[16:15:01.0000] <gsnedders>
BenMillard: Yeah, so it seems

[16:16:00.0000] <BenMillard>
gsnedders, we can arrange how to square up the costs in e-mail

[16:17:00.0000] <gsnedders>
BenMillard: Yeah, seems sane

[16:17:01.0000] <gsnedders>
/me emails smedero with link to the logs

[16:17:02.0000] <BenMillard>
gsnedders, as I won't be there the whole time I guess one of you should do the booking

[16:17:03.0000] <Hixie>
i can drive trucks much more easily than sports cars in gta4

[16:18:00.0000] <BenMillard>
Hixie, we're talking about Gran Turismo, but yes I have the same experience of past GTAs

[16:18:01.0000] <Hixie>
i usually prefer to do the races (against sports cars) in a bus, because the bus can take more damange. :-)

[16:18:02.0000] <gsnedders>
Hixie: Because of them skidding around more, due to their lack of weight?

[16:18:03.0000] <Hixie>
BenMillard: yeah, same kind of thing with GT

[16:18:04.0000] <gsnedders>
/me much prefers the sports cars, but he can cope with a madly light car

[16:19:00.0000] <BenMillard>
the original CD case for Grand Theft Auto describes the handling as "Hollywood" and "outrageous" which, to a greater or lesser extent, has been true in all editions of it

[16:19:01.0000] <Hixie>
gsnedders: partly, yes, but also with the bus i don't have to be as precise, and i can just plow through the competition and watch them splash into walls and buildings, while i don't even slow down

[16:19:02.0000] <gsnedders>
There again, I don't manage to be within the top 400 around the Nürburgring in Forza 2 by being a bad driver :)

[16:20:00.0000] <gsnedders>
Hixie: Peh. Being precise ain't hard. Though driving buses is fun.

[16:21:00.0000] <BenMillard>
Hixie, do you have and of the 3D GTA's PC? if so, I did a "handling overhaul" for each of them as a mod: http://projectcerbera.com/gta/3/handling-overhaul http://projectcerbera.com/gta/vc/handling-overhaul http://projectcerbera.com/gta/sa/handling-overhaul

[16:21:01.0000] <BenMillard>
s/and of/any of/

[16:21:02.0000] <Hixie>
nice

[16:21:03.0000] <Hixie>
i just have gta:vc and sa for ps2 and gta4 for ps3

[16:22:00.0000] <Hixie>
haven't played my ps3 in months, it needs to be set up in the new room

[16:22:01.0000] <BenMillard>
Hixie, you're welcome to join the PS2 Gran Turismo part with gsnedders and I :)

[16:22:02.0000] <Hixie>
:-)

[16:22:03.0000] <BenMillard>
I'll be bringing my console as it's in near-perfect running order

[16:22:04.0000] <BenMillard>
feel free to bring memory cards

[16:22:05.0000] <gsnedders>
BenMillard: You have two controllers, I assume?

[16:22:06.0000] <Hixie>
be aware of PAL vs NTSC issues

[16:22:07.0000] <BenMillard>
gsnedders, yes

[16:22:08.0000] <gsnedders>
Hixie: We're both in PAL Europe :)

[16:22:09.0000] <Hixie>
oh right, meeting is in europe

[16:23:00.0000] <Hixie>
cool!

[16:24:00.0000] <gsnedders>
/me ought to bring NFS:HP2 (IMO, the best NFS game)

[16:24:01.0000] <gsnedders>
Crazy Taxi is scratched :(

[16:25:00.0000] <Hixie>
well if y'all are going to be bringing consoles, that's certainly going to make it more interesting

[16:27:00.0000] <gsnedders>
Bringing a 360 by air is a bit too complicated, so I won't be.

[16:27:01.0000] <gsnedders>
But I'll probably bring one or two PS2 games :)

[16:27:02.0000] <BenMillard>
gsnedders, on closer inspection I have a single PS2 controller and two PS controlls

[16:27:03.0000] <gsnedders>
BenMillard: k, if you remind me, I can bring a PS2 controller

[16:27:04.0000] <Dashiva>
I have a ps2 controller that sends random signals without me pressing buttons

[16:27:05.0000] <gsnedders>
That could be fun.

[16:27:06.0000] <Dashiva>
I'm considering enrolling it in a fight game tournament

[16:28:00.0000] <Hixie>
i have a slim pal ps2 i can bring if necessary

[16:28:01.0000] <Hixie>
and a ps2 controller and some games (like worms 3d, which is great for parties)

[16:29:00.0000] <Hixie>
but the ps3 is bigger than my backpack, which is all i'm planning on bringing

[16:29:01.0000] <BenMillard>
Hixie, I have the slim one

[16:29:02.0000] <gsnedders>
Hixie: I think the simple solution is we let BenMillard bring the console :)

[16:30:00.0000] <Hixie>
i'm just sayin' :-)

[16:30:01.0000] <BenMillard>
unless you know how to connect 2 together, which could be awesome

[16:30:02.0000] <BenMillard>
although one 1 TV...hmm

[16:30:03.0000] <BenMillard>
*only 1 TV

[16:30:04.0000] <gsnedders>
For GT3 at least you can use the Firewire port…

[16:30:05.0000] <gsnedders>
But I don't think that's on the slim console

[16:32:00.0000] <gsnedders>
GT4 supports LAN with ethernet

[16:45:00.0000] <BenMillard>
LAN requires 1 TV per player: http://www.us.playstation.com/Content/OGS/SCUS-97328/Site/gt4manual.pdf

[16:46:00.0000] <BenMillard>
(page "twenty")

[16:53:00.0000] <gsnedders>
peh.

[16:55:00.0000] <BenMillard>
Hixie, don't bother bringing your PS2. we'll only have 1 TV

[16:56:00.0000] <Philip`>
Bring a TV

[16:56:01.0000] <BenMillard>
lol

[16:56:02.0000] <Philip`>
as long as it's flatscreen

[16:56:03.0000] <gsnedders>
Actually, I have a video capture thing for my computer!

[16:57:00.0000] <Philip`>
Do any laptops have video-in that's compatible with the PS2's outputs?

[16:57:01.0000] <gsnedders>
Like, one that has next to no lag and can be used with a PS2

[16:57:02.0000] <Philip`>
Ah

[16:57:03.0000] <gsnedders>
And it's what I normally use with my PS2

[16:57:04.0000] <gsnedders>
Hixie: Do bother bringing your PS2. We have two TVs.

[16:58:00.0000] <gsnedders>
We also need two copies of GT4 to do that though

[16:58:01.0000] <gsnedders>
And mine is scratched, but mostly works

[16:58:02.0000] <gsnedders>
Just need to reset the console after every race :P


2008-08-28
[17:06:00.0000] <Hixie>
hey my interview is up

[17:06:01.0000] <Hixie>
http://blogs.techrepublic.com.com/programming-and-development/?p=718

[17:07:00.0000] <BenMillard>
gsnedders, that's good news :)

[17:08:00.0000] <BenMillard>
do you want to bring the necessary network cable and learn how it all works so I don't have to? :P

[17:08:01.0000] <BenMillard>
gsnedders, I've read this and it seems pretty exhaustive: http://mastergt.250free.com/GT4/LAN_GT4/GT4_LANs.html

[17:09:00.0000] <BenMillard>
Hixie, "While Ian and I have had our disagreements over a few details of the specification, I have come to respect him immensely for the work that he puts into the spec on a daily basis." I think anyone who gives you a chance has that experience.

[17:11:00.0000] <Hixie>
aww

[17:17:00.0000] <Dashiva>
What a lie. We all know Hixie works most during the night ;)

[17:18:00.0000] <Hixie>
i even mention that in the interview :-)

[17:19:00.0000] <BenMillard>
Hixie, you wrote "I don’t think that realistically speaking HTML5 is going to make people write better markup." Well, simplifying the markup for accessible tables will realistically improve new tables markup written by the sites which try to be accessible, once we have implementations showing how well it works.

[17:19:01.0000] <BenMillard>
although that's likely to be a minority of the web for some time yet, I must admit

[17:21:00.0000] <Hixie>
yeah i think i go on to mention that we've done some things to make things a little better (like the /> thing)

[17:21:01.0000] <Hixie>
i hope we have some effect, i'm just not holding my breath :-)

[17:21:02.0000] <Hixie>
hsivonen said that in his study so far he has yet to find a single page with zero errors

[17:22:00.0000] <BenMillard>
gsnedders, just saw this: "Can you justify the cost of coming for a whole week now?" hmm, that's an idea

[17:23:00.0000] <BenMillard>
since I'm paying an equal share anyway, I might as well be there for it :)

[17:23:01.0000] <Dashiva>
It feels like beekeeping pops up inordinately often in examples

[17:25:00.0000] <Hixie>
Dashiva: it's not accidental

[17:25:01.0000] <Hixie>
Dashiva: let me know if you can work out why :-)

[17:25:02.0000] <Hixie>
(nothing to do with me or whatwg or html5)

[17:25:03.0000] <Hixie>
though if you look at html5 it'll give you a hint

[17:26:00.0000] <roc>
ooh ooh I know this one

[17:27:00.0000] <Hixie>
it's not very deep or anything

[17:31:00.0000] <Hixie>
man, every time i look at this page i find more errors http://puysl.com/view.htm

[17:31:01.0000] <Hixie>
and note that it was updated this month!

[17:31:02.0000] <Dashiva>
Is it like herding cats?

[17:31:03.0000] <Hixie>
it's a regularly updated page!

[17:32:00.0000] <Hixie>
Dashiva: it's a reference to something that was mentioned in passing in the late 90s

[17:34:00.0000] <BenMillard>
Hixie, wow...that wins the prize for stupidest way to mark up table headers :)

[17:37:00.0000] <BenMillard>
Hixie, you said "A big part of my job ends up being reinterpreting what people are asking for into what they really want." I think that's what causes much of the initial confusion and feelings of being dismissed; it's a necessary but unnatural response.

[17:38:00.0000] <BenMillard>
(I mean, your response to those requests is necessary but unnatural, since they'd normally be taken on face value)

[17:40:00.0000] <BenMillard>
Hixie, I couldn't have put this better myself: "If there’s one thing I’ve learnt through the extensive research we’ve done for HTML5, it’s that authors, on the aggregate, act in very surprising ways that don’t really match my intuition."

[17:52:00.0000] <Hixie>
http://junkyard.damowmow.com/338

[17:53:00.0000] <Hixie>
BenMillard: yeah, i think you're right (re: why people sometimes feel dismissed)

[17:53:01.0000] <Hixie>
and sometimes i just do dismiss ideas because i don't understand them, probably :-)

[17:53:02.0000] <Dashiva>
Hixie: It's obviously just an example of distributed extensibility

[17:57:00.0000] <Hixie>
http://junkyard.damowmow.com/339 is a saner size

[17:59:00.0000] <Hixie>
http://www.reddit.com/r/WTF/comments/6ycxe/html_youre_doing_it_wrong_based_on/

[18:04:00.0000] <BenMillard>
Hixie, can you remember (or better yet, find) where in the logs we talked about meeting at TPAC 2008? I've spent an hour trying but failed completely.

[18:04:01.0000] <BenMillard>
oh, I guess it's possible the logging was down during the time we were talking about it...

[18:04:02.0000] <Hixie>
i don't recall talking much about it :-)

[18:04:03.0000] <BenMillard>
you only said an entry or two

[18:05:00.0000] <Hixie>
oh that was yesterday iirc

[18:06:00.0000] <BenMillard>
aha, then it may have been in the small hours when logging wasn't working: http://krijnhoetmer.nl/irc-logs/whatwg/20080827

[18:06:01.0000] <BenMillard>
oh well

[18:22:00.0000] <guillebravo6>
I'm new to this group

[18:22:01.0000] <guillebravo6>
how's it going everybody?

[18:25:00.0000] <BenMillard>
hello there, guillebravo6

[18:25:01.0000] <Hixie>
welcome, guillebravo6 

[18:26:00.0000] <guillebravo6>
thanks guys

[18:26:01.0000] <BenMillard>
it's currently the middle of the night in Europe and quite late in USA, so not many people active AFAICT

[18:26:02.0000] <guillebravo6>
yeah I understand

[18:27:00.0000] <guillebravo6>
this is a reddit group right?

[18:27:01.0000] <guillebravo6>
are people on Digg, StumbleUpon, Delicious, etc.......?

[18:32:00.0000] <franksalim>
guillebravo6, what do you mean by a reddit group? i use reddit, but not in connection to #whatwg

[18:32:01.0000] <guillebravo6>
ok

[18:32:02.0000] <guillebravo6>
I wasn't sure

[18:32:03.0000] <Hixie>
this is an irc channel for discussing html5 development as part of the whatwg.org project

[18:32:04.0000] <guillebravo6>
I was just browsing around and saw Reddit on this channgel

[18:32:05.0000] <guillebravo6>
*channel

[18:32:06.0000] <guillebravo6>
no biggie

[19:29:00.0000] <Hixie>
takkaria: yt? did you ever implement the svgwg proposal?

[19:49:00.0000] <takkaria>
Hixie: no, I didn't.  I intended to but then Henri sent his rebuttal and I figured it was definitely not worth it

[19:49:01.0000] <Hixie>
k

[20:05:00.0000] <Hixie>
right, feedback on the svgwg proposal sent

[21:25:00.0000] <Hixie>
is there anything i should be working on?

[21:25:01.0000] <Hixie>
i'm out of prioritised stuff, so if there's anything important for me to work on, let me know

[21:25:02.0000] <Hixie>
otherwise i'll just go back to dealing with the oldest feedback

[21:32:00.0000] <jcranmer>
Hixie: specifying what should happen if authors don't comply?

[21:33:00.0000] <jcranmer>
e.g., where should they send their first-borns?

[00:19:00.0000] <hsivonen>
Hixie: well, obviously, there are no pages with no errors when you move the goalposts from HTML 4.01 Transitional 

[00:36:00.0000] <Hixie>
hsivonen: oh, right, you're assuming html5

[00:37:00.0000] <Hixie>
hsivonen: do you have the ability to do html4 transitional validation?

[00:42:00.0000] <hsivonen>
Hixie: I have to the precision validator.nu can now do it

[01:17:00.0000] <Hixie>
hsivonen: that... wasn't english. but i assume it's a yes? :-)

[01:18:00.0000] <hsivonen>
Hixie: it was a yes. (s/have to/have, to/)

[01:19:00.0000] <Hixie>
ohh, i see what you're saying

[01:38:00.0000] <annevk>
/me could bring a Wii

[01:42:00.0000] <annevk>
haha, IE8 has a "Compatibility View" button

[02:22:00.0000] <annevk>
http://blogs.techrepublic.com.com/programming-and-development/?p=718 is good Hixie

[02:22:01.0000] <Hixie>
cool, glad you like it

[02:49:00.0000] <Lachy>
From the interview: "Another topic that’s received a lot of feedback is that we dropped acronym in favor of just using abbr, and a lot of people complained, though few could agree on what exactly the two elements would mean if we kept both."

[02:49:01.0000] <Lachy>
the only viable solution if both were kept, is to define both as synonyms of each other

[03:19:00.0000] <zcorpan>
interesting, instead of setting status to "fixed", microsoft set it to "not reproducable" when a bug i've reported has been fixed in ie8b2

[03:20:00.0000] <zcorpan>
though perhaps that's a good way to get the fix verified by the reporter

[03:23:00.0000] <Philip`>
I've had two marked as Fixed

[03:23:01.0000] <Philip`>
(and they are actually fixed, I think)

[03:24:00.0000] <Philip`>
but one Not Reproducible, and I don't think it's actually fixed but maybe my test case was unreliable

[03:27:00.0000] <Hixie>
hsivonen: interesting typo in "how often the local property is within the namespace"

[03:27:01.0000] <Hixie>
(property is -> properties)

[03:30:00.0000] <Lachy>
OMG, Microsoft's download site for IE8 popped up an ad for me to install silverlight, even though I still have silverlight installed from when I needed it to watch videos on nbcolympics.com

[03:31:00.0000] <Hixie>
lordy, now hsivonen is suggesting that whatwg.org use a CDN. :-P

[03:31:01.0000] <zcorpan>
hsivonen: fwiw, i think you'd save time by learning those 3 namespace URIs

[03:32:00.0000] <zcorpan>
hsivonen: instead of looking them up

[03:32:01.0000] <Hixie>
i never learnt the HTML4 DOCTYPE

[03:32:02.0000] <Hixie>
that's the main reason for the HTML5 synax. :-D

[03:32:03.0000] <zcorpan>
http://www.w3.org/1999/xhtml http://www.w3.org/2000/svg http://www.w3.org/1999/xlink

[03:32:04.0000] <zcorpan>
<!doctype html public "-//W3C//DTD HTML 4.01//EN">

[03:33:00.0000] <zcorpan>
gotta to

[03:33:01.0000] <zcorpan>
go

[03:35:00.0000] <hsivonen>
Hixie: not a typo. artifact of a different text input method

[03:35:01.0000] <Lachy>
I learned all the HTML 4.01 and XHTML 1.0/1.1 DOCTYPEs. It's not that hard, since the follow a common pattern

[03:36:00.0000] <hsivonen>
Hixie: no, I suggested that only n.whatwg.org be on a CDN

[03:36:01.0000] <hsivonen>
Hixie: with xmlns dereferencing going on, it seems like a bad idea that W3C isn't separating its ns URIs from the rest on the DNS level

[03:36:02.0000] <Lachy>
hsivonen, wouldn't that be a costly thing to maintain?

[03:36:03.0000] <annevk>
/me read http://esw.w3.org/topic/HTML/history found http://www.hixie.ch/specs/html/apps/web-apps-1 and realized almost nothing survived

[03:37:00.0000] <Philip`>
Lachy: Pay for it by embedding adverts

[03:37:01.0000] <hsivonen>
Lachy: yes. the point of what I wrote is to shield people who don't want the Power of RDF from the RDF Tax. However, someone needs to pay for the CDN

[03:38:00.0000] <hsivonen>
Lachy: so to make the cost entirely borne by the RDF community, you'd need a way to make the RDF community to pay the CDN bill

[03:39:00.0000] <hsivonen>
I haven't learden the HTML 4 doctypes either. I always copy them from my doctype page

[03:39:01.0000] <Hixie>
hsivonen: yeah that was intentional when i minted n.whatwg.org

[03:40:00.0000] <Hixie>
hsivonen: speech recognition eh? how is it going?

[03:40:01.0000] <hsivonen>
Hixie: needs fixing

[03:41:00.0000] <Lachy>
Philip`, adverts on a server intended purely for delivering machine processable data, rather than human consumable data, seems rather ineffective.

[03:43:00.0000] <Philip`>
Lachy: Not at all - you just have to exploit how the machines process that data, so that the adverts get shown to those machines' users

[03:43:01.0000] <hsivonen>
Hixie: speech recognition is more accurate than Ink, though

[03:43:02.0000] <Hixie>
yeah well

[03:43:03.0000] <Hixie>
ink sucks :-P

[03:43:04.0000] <Lachy>
although I don't particularly like the whole idea, one way to pay for it would be to require software that intends to use it, has to pay licence fees. That way it's paid for by those who use it.

[03:43:05.0000] <Hixie>
that would work

[03:44:00.0000] <Hixie>
we could optimise away actually setting anything up if we did that

[03:44:01.0000] <hsivonen>
lol

[03:44:02.0000] <Philip`>
Use Google Cache as a free CDN

[03:44:03.0000] <Lachy>
LOL

[03:45:00.0000] <hsivonen>
speaking of free CDNs, what pays for Coral?

[03:45:01.0000] <Lachy>
or Freenet

[03:45:02.0000] <hsivonen>
is NYU bankrolling it out of the goodness of their hearts?

[03:47:00.0000] <annevk>
CDN?

[03:47:01.0000] <hsivonen>
Content Distribution Network

[03:47:02.0000] <annevk>
ta

[03:48:00.0000] <Philip`>
hsivonen: I think it's running on PlanetLab, which I think is mostly run by universities, presumably from various sources of research funding

[03:49:00.0000] <Lachy>
oh, cool. it works http://lachy.id.au.nyud.net/log/

[03:53:00.0000] <Philip`>
"As of January 2006, it receives about 25 million requests per day from more than 1 million unique clients." - that sounds like it's probably only tens of megabytes a second, which isn't particularly extreme

[04:23:00.0000] <hsivonen>
hmm. I could put accessibility APIs to use and define Typinator macros for namespace URIs

[04:25:00.0000] <Hixie>
hsivonen: on a.v.n/htmlparser/, "The parser core compiles on Google Web Toolkit and in being made more portable to other programming languages.", s/in/is/

[04:26:00.0000] <hsivonen>
Hixie: fixed. thanks

[04:27:00.0000] <annevk>
Lachy, why the double domain?

[04:28:00.0000] <Hixie>
hsivonen: your TreeBuilder interface is quite an interesting view into what i specced

[04:28:01.0000] <Hixie>
hsivonen: what is startTag() used for?

[04:31:00.0000] <Lachy>
annevk, .nydu.net is the Coral CDN. http://www.coralcdn.org/

[04:31:01.0000] <hsivonen>
Hixie: the tokenizer calls startTag() when it reports a start tag token

[04:31:02.0000] <Hixie>
sure but why?

[04:31:03.0000] <Hixie>
as in, what was your use case?

[04:32:00.0000] <hsivonen>
the code for processing start tag tokens needs to be somewhere

[04:32:01.0000] <hsivonen>
the code structure is inspired by SAX

[04:32:02.0000] <Hixie>
oh you mean that's the method where you have to implement the whole tree construction?

[04:33:00.0000] <annevk>
Lachy, interesting

[04:33:01.0000] <hsivonen>
havinge a SAX-inspired structure has certain benefits:

[04:33:02.0000] <hsivonen>
1) I can make a streaming SAX mode with reasonable ease

[04:34:00.0000] <hsivonen>
2) I avoid heap-allocating objects for tokens

[04:34:01.0000] <Hixie>
oh i agree that sax makes sense, and i understand the concept of calling a method for each type of token

[04:34:02.0000] <hsivonen>
3) the tree builder rules are easier to implement token-major mode-minor than the way the spec puts them

[04:34:03.0000] <Hixie>
i was just looking to see what one would have to implement to get document.write()

[04:35:00.0000] <Hixie>
since you mention having to implement TreeBuilder to get that

[04:35:01.0000] <hsivonen>
Hixie: TreeBuilder and Tokenizer don't implement document.write().

[04:36:00.0000] <hsivonen>
Hixie: they just have capability to get out of the way

[04:36:01.0000] <hsivonen>
Hixie: see gwt-src

[04:36:02.0000] <Hixie>
is that online anywhere?

[04:36:03.0000] <hsivonen>
Hixie: the subclass of TreeBuilder there requests suspensions when it wants a script to run

[04:37:00.0000] <hsivonen>
Hixie: no, not as HTML. only in the .zip

[04:37:01.0000] <Hixie>
ah ok

[04:37:02.0000] <hsivonen>
and in svn

[04:37:03.0000] <Hixie>
oh well no worries, i was just poking around

[04:37:04.0000] <Hixie>
putting off replying about irrelevant=""

[04:37:05.0000] <hsivonen>
when suspension is requested, control returns from the Tokenizer to the driver that pushes data to tokenizer

[04:38:00.0000] <hsivonen>
so that class is responsible for holding onto the remainder of the buffer that was being tokenized

[04:38:01.0000] <hsivonen>
and pushing the document.write stuff meanwhile

[04:38:02.0000] <Hixie>
ah

[04:38:03.0000] <hsivonen>
the code breaks with nested document.write in Gecko and with nested SVG scripts, ATM

[04:39:00.0000] <Hixie>
those are a pain to get right

[04:39:01.0000] <hsivonen>
my next item is figuring out how nested scripts run in XML

[04:39:02.0000] <Hixie>
even in english those were a pain to get right

[04:39:03.0000] <Hixie>
nested scripts fail in html5 right now

[04:39:04.0000] <Hixie>
as in, the spec is wrong

[04:39:05.0000] <Hixie>
in fact <script> in general fails in html5 right now in xml

[04:39:06.0000] <hsivonen>
I think the code exposes a difference in Gecko and WebKit script executino details

[04:39:07.0000] <Hixie>
assuming the parser inserts the element before the element's children

[04:40:00.0000] <Hixie>
(not sure how to fix it, since there's no xml parser spec)

[04:40:01.0000] <hsivonen>
Hixie: SAX to DOM tree building is well understood

[04:40:02.0000] <hsivonen>
Hixie: it's the exact byte of WF error that is ill-defined

[04:41:00.0000] <Hixie>
yeah i'll probably just say "for html documents: when it is inserted... for xml documents: when you parse the end tag..."

[04:43:00.0000] <hsivonen>
Hixie: conceptually, it would be nice if HTML side inserted early, too, but started execution at end tag

[04:44:00.0000] <Hixie>
we could do that i guess

[04:44:01.0000] <Hixie>
though then we have to handle DOM-append separately from parser-append

[04:44:02.0000] <hsivonen>
yeah.

[04:44:03.0000] <Hixie>
which will make things exciting for you over in GWT land

[04:45:00.0000] <Hixie>
and we have to handle parser-append followed by DOM-append of the same node followed by parser-end-tag

[04:45:01.0000] <Lachy>
woo hoo! A new "interoperable" DRM scheme that invades my privacy by requiring me to register my devices. Just what I wanted! http://arstechnica.com/news.ars/post/20080827-open-market-video-drm-aims-to-let-1000-retailers-bloom.html

[04:45:02.0000] <hsivonen>
the gwt land code fakes "start execution on end tag" by inserting a placeholder and replacing it on end tag

[04:45:03.0000] <Hixie>
hsivonen: nice

[04:45:04.0000] <Hixie>
hsivonen: hope nobody takes a reference to the fake :-D

[04:46:00.0000] <hsivonen>
Hixie: so I naively thought that having one level of placeholders was OK

[04:46:01.0000] <hsivonen>
Hixie: until zcorpan tried to nest SVG scripts

[04:46:02.0000] <hsivonen>
zcorpan always finds holes in my code

[04:46:03.0000] <Hixie>
heh

[05:05:00.0000] <annevk>
Hixie, any chance the icons for the Web Apps spec can be hosted on whatwg.org? all the extra requests are a bit of a pain

[05:05:01.0000] <Lachy>
OMG, IE8b2 still hasn't fixed that annoying bug/design flaw with Ctrl+L. It should focus the address bar like a normal browser, instead of popping up the Open URL dialog :-(

[05:05:02.0000] <Hixie>
the reason they aren't is to avoid any suggestion of copyright violation. (why are they a pain?)

[05:05:03.0000] <Hixie>
Lachy: use alt-d, most browsers support that too

[05:06:00.0000] <annevk>
fetching them just seems painfully slow

[05:06:01.0000] <Lachy>
oh, I didn't know that. But I'd prefer if Ctrl+L worked, since that's what I'm used to, and I use Cmd+L for Mac browsers

[05:13:00.0000] <Hixie>
Lachy: (or f6, in ie)

[05:13:01.0000] <Hixie>
annevk: odd

[05:15:00.0000] <Lachy>
woah, IE's comment handling is buggered up. http://html5.lachy.id.au/output?data=%3C%21DOCTYPE+html%3E%0D%0A%3Ctitle%3EBogus+IE8+Conditional+Comments%3C%2Ftitle%3E%0D%0A%3Cp%3EYou+should+not+see+anything+below+this+line.%3C%2Fp%3E%0D%0A%3C%21--%5Bif+IE+lt+8%5D%3EFAIL%3C%21%5Bendif%5D--%3E&type=text%2Fhtml%3B+charset%3DUTF-8

[05:15:01.0000] <Lachy>
that's an incorrectly written conditional comment, but still, ouputting it like that seems wrong

[05:15:02.0000] <Hixie>
what does it do?

[05:16:00.0000] <Lachy>
It outputs "<!--[if IE lt 8]>FAIL<![endif]-->" literally on the page

[05:16:01.0000] <Hixie>
woah

[05:16:02.0000] <Hixie>
weird

[05:16:03.0000] <Lachy>
note that the conditional comment should actually be [if lt IE 8] for it to work as intended, but I always get it backwards

[05:17:00.0000] <Lachy>
where do I file bugs for IE?

[05:18:00.0000] <Hixie>
i believe prayer is the most effective method

[05:19:00.0000] <Dashiva>
I seem to recall there being some internal thing that you had to pass through five NDAs to reach, and then needed invitation

[05:20:00.0000] <Lachy>
yeah, there is somewhere, but I can't remember where it is

[05:20:01.0000] <Lachy>
Dear Microsoft Gods, please fix the bugs I have found with your browser. Amen.

[05:21:00.0000] <Dashiva>
Why limit yourself to only the bugs you found? :)

[05:22:00.0000] <Lachy>
it's not as if they're miracle workers. I at least need to give them some info about what is wrong.

[05:23:00.0000] <Hixie>
the gods aren't miracle workers?

[05:23:01.0000] <Lachy>
Hixie, just the microsoft gods.

[05:23:02.0000] <Hixie>
ah ok

[05:26:00.0000] <hsivonen>
w00t. making typinator macros for ns uris was a good idea

[05:27:00.0000] <gsnedders>
/me sighs

[05:27:01.0000] <gsnedders>
the TPAC is going to be crazy fun. Two PS2, three geeks, are far too much geekery

[05:28:00.0000] <Lachy>
grr. Apparently, it's supposed to be connect.microsoft.com where you can submit bug reports. But after signing in, it asks me to update my profile. I check that everything is ok, click continue and it takes my right back to the beginning.

[05:28:01.0000] <Lachy>
Now, where do i file a bug report about the bug reporting system?

[05:28:02.0000] <Hixie>
q.v. earlier comments

[05:29:00.0000] <Lachy>
maybe I should just complain directly to Chris Wilson

[05:31:00.0000] <annevk>
/me learns about "conundrum"

[05:31:01.0000] <Lachy>
ah, I see. The website only works in IE

[05:33:00.0000] <gsnedders>
/me wonders what you do if an IE8 beta has a bug stopping you from using that site.

[05:37:00.0000] <Lachy>
I don't know. but I can't actually find the Report a damn bug link, or equivalent anywhere.

[05:38:00.0000] <Lachy>
this is getting annoying

[05:38:01.0000] <annevk>
Hixie, dom-command-ro-hidden is removed from IDL (where it was dom-command-hidden) but not from prose

[05:38:02.0000] <Lachy>
although I can see one bug listed that I had filed in 2006 about another product. Still not fixed :-)

[05:38:03.0000] <annevk>
+  title="command-element">command</code>, <code

[05:38:04.0000] <annevk>
+  title="bb-command">command</code>.</p>

[05:38:05.0000] <annevk>
also seems wrong

[05:39:00.0000] <Hixie>
dom-command-ro-hidden and dom-command-hidden are unrelated

[05:39:01.0000] <Hixie>
one is gone the other is not

[05:39:02.0000] <Hixie>
fixed the bb issue

[05:39:03.0000] <annevk>
oh, there's a separate Command interface?

[05:40:00.0000] <Hixie>
yeah

[05:40:01.0000] <Hixie>
not quite sure what i'm gonna do with it

[05:40:02.0000] <Lachy>
ah, apparently I need to register as a beta tester to do it. http://blogs.msdn.com/ie/archive/2008/07/30/wanted-ie8-beta-testers.aspx

[05:40:03.0000] <Hixie>
we'll see

[05:40:04.0000] <hsivonen>
bah. SVG elements don't have innerHTML in Gecko.

[05:41:00.0000] <annevk>
Hixie, I think the thing with the icons is that it takes some time to make the request to all these other servers, which gives quite a bit of overhead over it all coming from whatwg.org

[05:41:01.0000] <Hixie>
annevk: so?

[05:42:00.0000] <annevk>
in Opera that delays search interaction with the document

[05:42:01.0000] <Hixie>
annevk: file a bug

[05:43:00.0000] <Hixie>
on a totally different note, wtf is up with my mac's menu widget things. every now and then, the time machine menu, the bluetooth menu, and the wifi menu go totally batshit crazy, reverting to their initial settings and losing all interaction.

[05:43:01.0000] <Hixie>
e.g. the wifi icon disappears and it claims to be offline in the menu, even though wifi is fine

[05:43:02.0000] <Hixie>
and the bluetooth menu reverts to kThisThat menu labels

[05:43:03.0000] <Hixie>
and the time machine menu says no backup ever ran

[05:43:04.0000] <hsivonen>
Gecko and WebKit have XHTML and SVG nested script execution interop

[05:43:05.0000] <hsivonen>
and XHTML and SVG are consistent

[05:43:06.0000] <hsivonen>
Opera does something crazy

[05:43:07.0000] <Philip`>
Lachy: For registering connect.microsoft.com, try not using Opera - I think I had problems with that

[05:44:00.0000] <Hixie>
seems to happen around the same time that iTunes freaks out over my iPod touch and starts not recognising it

[05:44:01.0000] <hsivonen>
demo: http://hsivonen.iki.fi/test/moz/script-execution.xhtml

[05:44:02.0000] <Hixie>
i wonder what on earth is up with this machine

[06:45:00.0000] <annevk>
/me almost replied to RB

[06:45:01.0000] <annevk>
I have like 400 e-mails in drafts

[06:45:02.0000] <annevk>
a lot of it is probably due to Opera's autosave feature though

[06:45:03.0000] <annevk>
haha

[06:45:04.0000] <annevk>
blog: http://news.bbc.co.uk/1/hi/england/hereford/worcs/7585098.stm

[06:45:05.0000] <gsnedders>
/me wonders whether to just call the specGen cameleon and be done with it

[06:45:06.0000] <Lachy>
gsnedders, chameleon is spent with an h, unless the misspelling is intentional

[06:45:07.0000] <Lachy>
in which case, I think it's a bad name

[06:45:08.0000] <gsnedders>
/me yawns

[06:45:09.0000] <Lachy>
pick a name that's easy to spell and is relevant

[06:45:10.0000] <gsnedders>
/me is half-asleep

[06:49:00.0000] <gsnedders>
Lachy: What is relevant though? htmlpostprocessor?

[06:49:01.0000] <Lachy>
call it sg

[06:49:02.0000] <gsnedders>
sg? Why?

[06:49:03.0000] <Lachy>
it's consistent with the linux approach of using the shortest possible, and totally unclear abbreviations for command names

[06:49:04.0000] <Lachy>
sg = Spec Gen

[06:49:05.0000] <annevk>
I thought alonis was quite good

[06:49:06.0000] <Philip`>
You have to have versioning information in case you release another one later, so you'd have to call it SG-1

[06:49:07.0000] <Lachy>
Philip`, that's what I was thinking :-)

[06:49:08.0000] <annevk>
SG-1 wfm

[06:49:09.0000] <Lachy>
I don't like the hyphen in command names though

[06:49:10.0000] <gsnedders>
/me bets SG-1 is a trademark

[06:49:11.0000] <annevk>
use alonis then and move on

[06:49:12.0000] <Lachy>
what is alonis?

[06:49:13.0000] <annevk>
or chameleon

[06:49:14.0000] <Philip`>
Trademarks don't have global scope

[06:49:15.0000] <gsnedders>
sg is bad anyway because it isn't really specific to specs

[06:49:16.0000] <Philip`>
so it's only a problem if someone has released similar software under a trademarked name SG-1

[06:49:17.0000] <annevk>
Lachy, a species

[06:49:18.0000] <gsnedders>
annevk: No anoles is

[06:49:19.0000] <Lachy>
really? what kind of animal?

[06:49:20.0000] <gsnedders>
http://en.wikipedia.org/wiki/Anoles

[06:49:21.0000] <gsnedders>
Lizard

[06:49:22.0000] <annevk>
oops

[06:49:23.0000] <gsnedders>
anolis is a genera of the anoles though

[06:49:24.0000] <gsnedders>
seemingly

[06:50:00.0000] <annevk>
right, I did mean Anolis

[06:50:01.0000] <annevk>
oh well, need to go somewhere

[06:50:02.0000] <gsnedders>
Anolis it is.

[06:50:03.0000] <Philip`>
Anoliser

[06:50:04.0000] <gsnedders>
But versioning may be useful

[06:52:00.0000] <annevk>
gsnedders, don't worry about versioning too much until it becomes an actual problem

[06:53:00.0000] <Lachy>
/me moans about the choice of a terrible name and moves on. Oh well.

[06:55:00.0000] <annevk>
deployment of this tool is not going to be widespread anyway

[07:24:00.0000] <Lachy>
Hixie, bug in the spec update notification script: "This specification has been updated. You are reading r2120 but the latest revision is r2119 ("Rename the 'irrelevant' attribute to 'hidden'.")"

[07:40:00.0000] <krijnh>
(Sorry for crappy connection & logs)

[07:40:01.0000] <gsnedders>
If anyone finds any bugs in <http://stuff.gsnedders.com/spec-gen/html5.html>, do tell me

[07:40:02.0000] <Lachy>
notice that it things 2120 is older than 2119

[07:40:03.0000] <Lachy>
woah, this mail gives some interesting insight into the way authors might abuse workers in inefficient ways. http://lists.w3.org/Archives/Public/public-html/2008Aug/0823.html

[07:40:04.0000] <Philip`>
gsnedders: The copyright statement is a smaller font than in the whatwg.org version

[07:40:05.0000] <Philip`>
Maybe that's because the whatwg.org version is in quirks mode?

[07:40:06.0000] <Lachy>
Philip`, is that in Opera? It's the same size in Firefox.

[07:40:07.0000] <Philip`>
Ooh, I've found a bug: The new spec-gen uses unquoted attributes and therefore the source looks hideous

[07:40:08.0000] <Philip`>
Lachy: Yes

[07:40:09.0000] <Lachy>
oh, yeah, it should use quoted attributed cause they do look better

[07:40:10.0000] <Lachy>
gsnedders, is the attribute quoting an optional feature?

[07:40:11.0000] <Lachy>
Philip`, in IE, the copyright statement is larger in the whatwg.org version :-)

[07:40:12.0000] <Lachy>
hooray for interopability! ;-)

[07:40:13.0000] <Philip`>
It's called "quirks" for a reason :-)

[07:40:14.0000] <Lachy>
hmm. weird. I just realised that Snickers in Norway have white-coloured nougat, wheras in Australia, they have chocolate-coloured nougat

[07:40:15.0000] <Philip`>
Does anyone happen to know how http://philip.html5.org/docs/quirks.txt would need to be updated for IE8?

[07:40:16.0000] <Lachy>
Philip`, no idea. But if you have a TC, I can run it for you

[07:40:17.0000] <kangax>
Lachy: white-colored in US as well

[07:40:18.0000] <Lachy>
what?! OMG! that's crazy

[07:40:19.0000] <Philip`>
Lachy: I have some code to test quirks/standards, but I'll need to extend that to detect the quirks/IE7-bugs/IE8-bugs modes, which I suppose I'll do at some point in the future

[07:41:00.0000] <Lachy>
kangax, is it the same with milky ways too?

[07:44:00.0000] <Lachy>
Anyway, just 24 days till I get back to eating normal coloured, correctly named chocolate bars :-) Woo Hoo!

[07:44:01.0000] <Philip`>
kangax: Most non-trivial canvas tests don't work in Excanvas at all

[07:45:00.0000] <krijn>
http://krijnhoetmer.nl/irc-logs/html-wg/20080828 - yuk! Does this happen a lot?

[07:45:01.0000] <Philip`>
Also I couldn't get Excanvas to work at all in IE8b1

[07:45:02.0000] <kangax>
damn

[07:45:03.0000] <Philip`>
In any case it's likely to be the same as in IE <8, because nobody uses VML so they wouldn't have put any work into optimising it, I assume

[07:46:00.0000] <kangax>
Philip`: have you heard of any progress about silverlight-based excanvas?

[07:46:01.0000] <kangax>
(their google group seems to be somewhat dead)

[07:47:00.0000] <Philip`>
kangax: I'm not aware of any IE canvas implementations that have got much beyond proof-of-concept stage

[07:51:00.0000] <kangax>
Philip`: I wish I had some time to tackle with it. Currently working on a parser that allows to output "svg path" on canvas.

[08:21:00.0000] <Lachy>
and that way you don't end up with some attributes having quotes where necessary, and others not, within the same document. Rather, you end up with everything being consistent

[08:22:00.0000] <gsnedders>
Hixie doesn't seem to quote everything either

[08:23:00.0000] <Philip`>
He's crazy too

[08:23:01.0000] <gsnedders>
Whoever thought any of us were sane though?

[08:23:02.0000] <Philip`>
Unquoted attributes are objectively ugly, and if you disagree then you're wrong

[08:25:00.0000] <jgraham>
Philip`: Time for your medication again I see ;)

[08:26:00.0000] <gsnedders>
/me wonders whether he can take more paracetamol yet

[08:26:01.0000] <jgraham>
/me has no objection to changing html5lib defaults here fwiw

[08:26:02.0000] <jgraham>
Nor to leaving them as is

[08:27:00.0000] <gsnedders>
/me notes that changing them will break backwards compatibility, as the option names will have to be changed

[08:27:01.0000] <gsnedders>
Or maybe not

[08:27:02.0000] <gsnedders>
/me headdesks

[08:27:03.0000] <gsnedders>
/me wakes up

[08:27:04.0000] <gsnedders>
The option names in parse.py at least, as --quote-attr-values no longer makes sense if it defaults to True if you want to disable it

[08:28:00.0000] <Philip`>
Add '--no-...' versions of all those options, so people can always be explicit and not rely on the defaults

[08:29:00.0000] <Lachy>
jgraham, if anyone here needs medication, it's the crazy people who leave attributes unquoted, and maybe the leprechauns running around this office making noise. But sane people like Philip` and I don't! 

[08:29:01.0000] <gsnedders>
Lachy: Real sprites, or not?

[08:30:00.0000] <Lachy>
gsnedders, what?

[08:30:01.0000] <gsnedders>
Lachy: leprechauns and sprites, no?

[08:30:02.0000] <gsnedders>
*are

[08:30:03.0000] <Lachy>
oh, I didn't realise it meant that too

[08:30:04.0000] <Lachy>
yeah

[08:31:00.0000] <gsnedders>
Because that would be kinda awesome if you did have sprites running around your office :)

[08:32:00.0000] <Lachy>
actually, I should have gone with trolls. I'm told they're a norwegian creation, so it would make more sense than an irish creation :-)

[08:32:01.0000] <gsnedders>
:)

[08:33:00.0000] <Philip`>
/me has a reasonably large collection of trolls (humanoid creatures made of some sort or rubber or plastic, with long brightly-coloured hair) in a box in a cupboard somewhere at home

[08:34:00.0000] <Lachy>
I never had a troll doll. Though they were reasonably popular when I was a kid

[08:34:01.0000] <Lachy>
I think they were more of a girly thing

[08:34:02.0000] <gsnedders>
What does that say about Philip`?

[08:35:00.0000] <Lachy>
gsnedders, nothing. I assumed she was a girl all along. Was I wrong?

[08:35:01.0000] <Philip`>
http://en.wikipedia.org/wiki/Troll_doll - aha

[08:35:02.0000] <Philip`>
They weren't girly! >:-(

[08:35:03.0000] <gsnedders>
Lachy: He doesn't look like one, at lesat

[08:35:04.0000] <gsnedders>
*least

[08:35:05.0000] <gsnedders>
Lachy: And his voice is a bit deep for a girl

[08:35:06.0000] <gsnedders>
s/He/It/

[08:35:07.0000] <gsnedders>
s/his/it/

[08:37:00.0000] <Lachy>
gsnedders, since I've never met him/her in person yet, I'll have to take your word for it. :-)

[08:37:01.0000] <gsnedders>
Lachy: :)

[08:37:02.0000] <Philip`>
I might have hired an actor to pretend to be me, to disguise my real identity

[08:37:03.0000] <gsnedders>
Lachy: Also, I can't show you a photo of it because its reaction to being photographed

[08:39:00.0000] <Lachy>
Philip`, sure, the pink hair and the fact that it was a doll definitely doesn't make it girly. Sorry ;-)

[08:41:00.0000] <Philip`>
Lachy: I don't think I had any with pink hair - they were mostly yellow and green, as far as I can remember

[08:42:00.0000] <Philip`>
And I don't think I ever heard them called "troll dolls" before today - they were just trolls, in the same way that you get Action Mans and not Action Man dolls

[08:44:00.0000] <Lachy>
OMG! found more UI bugs in IE8. this is terrible

[08:45:00.0000] <Philip`>
And anway I think I mostly made cars crash into them, rather than combing their hair or whatever girls would do

[08:45:01.0000] <Philip`>
s/terrible/beta/

[08:45:02.0000] <Lachy>
s/beta/lower-than-alpha/

[08:46:00.0000] <Lachy>
this latest bug is: Resize the window, press Ctrl+W to close all tabs till the window itself closes. Then restart IE, and it reopens at the size it was before resizing it.

[08:47:00.0000] <Lachy>
Do the same, but use the close button instead of Ctrl+W, and it remembers the last size properly

[08:50:00.0000] <gsnedders>
/me had one or two trolls, and did what Philip` didn't, and did what girls would do

[08:50:01.0000] <gsnedders>
/me ducks

[08:51:00.0000] <Lachy>
gsnedders, fair enough. I'm sure they helped you practice before messing with your own hair.

[08:51:01.0000] <Lachy>
/me ducks

[08:51:02.0000] <gsnedders>
hmm…

[08:51:03.0000] <gsnedders>
/me does have relatively long hair at the moment

[08:51:04.0000] <Lachy>
you should dye it bright green and take a photo for us

[08:51:05.0000] <Lachy>
then I can use that in my next presentation

[08:52:00.0000] <Lachy>
I'll need it before about the end of september. My presentation is on October 7th

[08:52:01.0000] <gsnedders>
A girl at school was saying I ought to dye it green

[08:52:02.0000] <gsnedders>
(bright green, that is)

[08:52:03.0000] <gsnedders>
I don't think my mother would approve, though

[08:53:00.0000] <Philip`>
Wear a hat, so she wouldn't notice

[08:53:01.0000] <Lachy>
well, you've got to give into peer pressure. You don't have a choice about that.

[08:53:02.0000] <gsnedders>
:S

[08:53:03.0000] <gsnedders>
*:D

[08:53:04.0000] <Philip`>
/me also has a large cuddly troll with a hood

[08:53:05.0000] <gsnedders>
/me wonders where he could get green hair dye

[08:54:00.0000] <Lachy>
gsnedders, hair salon

[08:54:01.0000] <gsnedders>
/me realizes he knows people at school who would know

[08:54:02.0000] <gsnedders>
Like, specifically around here

[08:55:00.0000] <Lachy>
or perhaps you can get that spray on stuff from the supermarket that kids often used to wear to school swimming/athletics carnivals at my school, based on their team colours

[08:55:01.0000] <gsnedders>
Lachy: That's boring.

[08:56:00.0000] <Lachy>
/me wonders if gsnedders is actually going to go through with it if he finds out where to get it

[08:56:01.0000] <gsnedders>
Lachy: Quite possibly :)

[08:56:02.0000] <gsnedders>
But it'll probably end up like all kinds of things I've said I'll do but never do

[08:56:03.0000] <Lachy>
great! I look forward to it.

[08:56:04.0000] <Lachy>
/me changes topic to 'WHATWG (HTML5) -- http://www.whatwg.org/ -- Logs: http://krijnhoetmer.nl/irc-logs/ -- Please leave your sense of logic at the door, thanks! - gsnedders promised to dye his hair green, photos coming soon :-)'

[08:57:00.0000] <gsnedders>
hah.

[08:57:01.0000] <Philip`>
gsnedders: Use radium

[08:57:02.0000] <gsnedders>
Philip`: Slightly risky.

[08:58:00.0000] <Philip`>
gsnedders: Sure, but it's worthwhile for glow-in-the-dark hair

[08:58:01.0000] <Lachy>
I think it's home time. I'm going to go pull out my transformers and play with some real boys toys!

[08:58:02.0000] <gsnedders>
Lachy: Like hair straighteners?

[08:59:00.0000] <Lachy>
gsnedders, no. Transformers!

[09:01:00.0000] <gsnedders>
/me notes the girl at school also said to wait for my hair to longer first

[09:01:01.0000] <gsnedders>
s/my/his/

[09:01:02.0000] <Philip`>
s//$verb/

[09:02:00.0000] <gsnedders>
/me notes girls are crazy

[09:02:01.0000] <gsnedders>
/me ducks

[09:02:02.0000] <mpt>
/me flees as the channel fills with an infinite number of verbs

[09:03:00.0000] <Philip`>
mpt: I didn't use /g :-p

[09:04:00.0000] <Philip`>
(and it wouldn't be infinite anyway unless there was an external loop)

[09:04:01.0000] <Philip`>
and anyway these are context-sensitive regexps that only apply to the (unspecified) appropriate point of the appropriate line

[09:05:00.0000] <Philip`>
(I don't think even Perl has implemented that much DWIM yet, sadly)

[09:18:00.0000] <gsnedders>
If I'm going to do this, I may as well do everything else that I wanted to do to change how I look that my mother will disapprove of too

[09:18:01.0000] <gsnedders>
Lachy: No, I haven't yet

[09:19:00.0000] <Philip`>
gsnedders: That does not necessarily make sense, unless she simply has a boolean "approve"/"disapprove" thought process

[09:20:00.0000] <gsnedders>
Philip`: I am aware it doesn't really make sense. But nothing else that I would do to change how I look would have any worse reaction that dying my hair green.

[09:24:00.0000] <BenMillard>
gsnedders, if you get it done then make sure it's expensive

[09:24:01.0000] <Philip`>
gsnedders: That doesn't make sense either, since disapproval is usually based on an additive function rather than a max function, so one extreme act plus one minor act is worse than just the extreme act by itself

[09:24:02.0000] <BenMillard>
otherwise it's likely to be a lousy job that sucks badly

[09:24:03.0000] <gsnedders>
Philip`: Peh. It gets it over and done with.

[09:24:04.0000] <Philip`>
BenMillard: At least it'd be a lousy job that sucks badly and was cheap

[09:24:05.0000] <gsnedders>
BenMillard: But I have next to no money :P

[09:25:00.0000] <gsnedders>
If I can raise £n, I will dye my hair!

[09:25:01.0000] <gsnedders>
:P

[09:25:02.0000] <BenMillard>
gsnedders, then your finances can act as a throttle to your mother's disapproval :)

[09:25:03.0000] <BenMillard>
i.e. you can only annoy her to the extent you can afford it

[09:25:04.0000] <gsnedders>
BenMillard: :)

[09:26:00.0000] <BenMillard>
but I should point out that you could try asking first

[09:27:00.0000] <BenMillard>
I imagine she grew up during the 1980's so doesn't have much of a case against you adopting a crazy appearance

[09:28:00.0000] <gsnedders>
BenMillard: My parents are old. Really old.

[09:28:01.0000] <BenMillard>
in that case the playing field is even more in your favour :)

[09:28:02.0000] <BenMillard>
until you get pre1960's, at least

[09:28:03.0000] <gsnedders>
BenMillard: My mother is during WWII :)

[09:29:00.0000] <BenMillard>
oh, lol

[09:29:01.0000] <gsnedders>
BenMillard: And there should be no apos. in either 1960s or 1980s

[09:29:02.0000] <gsnedders>
my father is months after the end

[09:29:03.0000] <gsnedders>
BenMillard: I did say old:)

[09:29:04.0000] <gsnedders>
(hi mum!)

[09:30:00.0000] <BenMillard>
"... all intranet websites display in Internet Explorer 7 Standards mode ..." -- http://blogs.msdn.com/ie/archive/2008/08/27/introducing-compatibility-view.aspx

[09:30:01.0000] <BenMillard>
so much for competition

[09:56:00.0000] <Dashiva>
The metadata will save us (after the tools save us from the metadata)

[13:29:00.0000] <annevk>
hmm, IE8 standards mode killed the createElement workaround?

[13:30:00.0000] <Lachy>
annevk, do you mean the one that fixes the parsing of unknown elements?

[13:30:01.0000] <Xenos>
annevk: Doesn't have as much ring to it as "video killed the radio star", but good try

[13:31:00.0000] <annevk>
omg, IE8 killed <canvas>

[13:31:01.0000] <Lachy>
/me tries it

[13:31:02.0000] <annevk>
(though that's not entirely true, <canvas> workarounds existed before the createElement workaround was discovered (again))

[13:33:00.0000] <Xenos>
How did IE8 kill <canvas>? It's not like IE's had any support for it so far?

[13:33:01.0000] <annevk>
people used createElement as part of <canvas> workarounds

[13:33:02.0000] <Lachy>
annevk, it works for me

[13:33:03.0000] <Lachy>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cbody%3E%0D%0A%3Cscript%3Edocument.createElement(%22xxx%22)%3B%3C%2Fscript%3E%0D%0A%3Cxxx%3E%3C%2Fxxx%3E

[13:34:00.0000] <annevk>
ok, good

[13:34:01.0000] <Lachy>
in what way did it kill canvas?

[13:34:02.0000] <annevk>
/me read it here http://intertwingly.net/blog/2008/08/28/Improved-Namespace-Support#c1219939338

[13:34:03.0000] <annevk>
Lachy, see above

[13:35:00.0000] <Philip`>
annevk: What people used createElement for that?

[13:35:01.0000] <annevk>
(the <canvas> thing was just a more dramatic statement of my initial one for Xenos)

[13:35:02.0000] <Philip`>
(Excanvas just looks for the CANVAS and /CANVAS elements, and replaces that range with a new element)

[13:35:03.0000] <annevk>
Philip`, someone at Google? dunno, only heard this through proxy

[13:36:00.0000] <Philip`>
/me really wishes that pressing tab in IE's address bar would focus the search box, not the refresh button

[13:36:01.0000] <Lachy>
oh, crap. http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cbody%3E%0D%0A%3Cstyle%3Exxx%20%7B%20color%3A%20green%3B%20%7D%3C%2Fstyle%3E%0D%0A%3Cscript%3Edocument.createElement(%22xxx%22)%3B%3C%2Fscript%3E%0D%0A%3Cxxx%3Etest%3C%2Fxxx%3E

[13:37:00.0000] <Lachy>
while the createElement workaround fixes the parsing, it doesn't seem possible to style it

[13:37:01.0000] <annevk>
meh

[13:39:00.0000] <Philip`>
It seems you can't style psuedonamespaced elements either

[13:39:01.0000] <Philip`>
like <style>a\:b{...}</style><body xmlns:a><a:b>...</a:b>

[13:39:02.0000] <Philip`>
s/ue/eu/

[13:41:00.0000] <Lachy>
ok. We'll need to make sure they get bugs about these, especially the ones that aren't very forwards compatible with HTML5

[13:42:00.0000] <Lachy>
Philip`, if could you do it, since you're able to? I would, but they haven't got back to me yet

[13:43:00.0000] <Xenos>
https://connect.microsoft.com/IE/Feedback is the place to do it, I think

[13:43:01.0000] <Lachy>
I get page not found

[13:43:02.0000] <Xenos>
And then spam the people here and on the mailinglist to rate up the bugs we consider important 

[13:44:00.0000] <Xenos>
Hm, weird. I got there by following a link on https://connect.microsoft.com/site/sitehome.aspx?SiteID=136

[13:44:01.0000] <Philip`>
Lachy: Can do

[13:45:00.0000] <Lachy>
Xenos, it may be because I'm not an authorised beta tester yet. Are you?

[13:45:01.0000] <Xenos>
Nope

[13:45:02.0000] <Lachy>
I get 404 following that other link as well

[13:45:03.0000] <Xenos>
just googled for "internet explorer beta 2 feedback" and clicked the third result, which led me to page above which led on to first link I posted

[13:46:00.0000] <Lachy>
the first result I get for that search is http://blogs.msdn.com/ie/archive/2008/07/30/wanted-ie8-beta-testers.aspx

[13:46:01.0000] <annevk>
Lachy, you said "No, it is not a good idea to place constraints on the development of HTML based on the flaws of other tools/langauges", just on the flaws of browser tools then?

[13:46:02.0000] <Xenos>
Lachy: third result

[13:46:03.0000] <Xenos>
And try http://connect.microsoft.com/IE/Feedback instead, might be some https issue perhaps?

[13:47:00.0000] <Lachy>
annevk, that was in realtion to XSLT, and I meant authoring tool design flaws

[13:48:00.0000] <annevk>
I don't see how those shouldn't affect us

[13:48:01.0000] <annevk>
in fact, they already have, see eg />

[13:48:02.0000] <Lachy>
well, it's 7th result for me, and I still get 404

[13:48:03.0000] <annevk>
or xmlns="..."

[13:48:04.0000] <Xenos>
Lachy: Much strangeness.

[13:49:00.0000] <Lachy>
they're still complaining about not being able to output <!DOCTYPE html> using XSLT

[13:49:01.0000] <Lachy>
even though it can be done by disabling output escaping and doing it literally

[13:50:00.0000] <Philip`>
http://philip.html5.org/tests/ie8/cases/unknown-element-styling.html

[13:50:01.0000] <annevk>
wouldn't it be better to get IE fixed so you can simply emit XML from XSLT?

[13:50:02.0000] <Philip`>
http://philip.html5.org/tests/ie8/x-ua-ie7/cases/unknown-element-styling.html

[13:50:03.0000] <annevk>
what's the use of emitting text/html out of XSLT over text/xml ?

[13:51:00.0000] <Lachy>
Philip`, the 2nd TC passes in IE8

[13:51:01.0000] <Lachy>
the 1st fails

[13:51:02.0000] <Lachy>
annevk, I didn't think anyone said anything about text/xml

[13:51:03.0000] <Philip`>
Lachy: Both pass in IE7

[13:52:00.0000] <Philip`>
and the second passing in IE8 indicates that IE8's IE7 compatibility mode is compatible with IE7, which is good

[13:53:00.0000] <annevk>
Lachy, I'm just wondering

[13:55:00.0000] <Lachy>
both TCs pass in IE6 too

[13:57:00.0000] <Lachy>
the annoying thing about switching to IE running in VMware is making my fingers press the right keyboard shortcuts

[13:59:00.0000] <Lachy>
ha! microsoft.com is in the default list of websites that render in IE7 mode

[13:59:01.0000] <Philip`>
Does anyone have IE8b1 to test in?

[13:59:02.0000] <Xenos>
I must disagree. "the annoying thing about switching to IE" is leaving behind half your IQ :p

[13:59:03.0000] <Lachy>
Philip`, I can install it quickly, if you like

[14:00:00.0000] <Lachy>
well, not quickly. I need to make a copy of the VM, find the installation file if I've still got it, and then spend 20 mintes waiting for it to finsih

[14:00:01.0000] <Philip`>
Lachy: Hmm, probably no need, since I think someone would have noticed if that test case wouldn't have worked in IE8b1

[14:01:00.0000] <Xenos>
No wonder, Lachy: Having microsoft.com working well is a pretty important PR thing, and redeveloping microsoft.com for each beta would be pretty damn expensive :p

[14:01:01.0000] <Lachy>
I should probably install it anyway. But I have a problem, cause I need to maintain separate virtual machines to run each version of IE: IE6, 7, 8b1 and 8b2

[14:02:00.0000] <Lachy>
/me removes it from the list and takes a look...

[14:05:00.0000] <Lachy>
hmm, interesting. Regardless of whether it's in the list, it seems to always render in compatibility mode and the option to switch it is disabled.

[14:05:01.0000] <Lachy>
I wonder if they've set X-UA-Compatible or something to make it do that (which would be a more forward compatible approach, than effectively hard coding it into the browser

[14:06:00.0000] <Lachy>
yes: X-UA-Compatible: IE=EmulateIE7

[14:07:00.0000] <Lachy>
oh, they also have an option to display all intranet sites in compatibility mode, selected by default, or all websites in compat mode, unseleced by default

[14:23:00.0000] <Philip`>
https://connect.microsoft.com/IE/feedback/ViewFeedback.aspx?FeedbackID=364356

[14:25:00.0000] <annevk>
nice

[14:27:00.0000] <Philip`>
Does http://philip.html5.org/tests/ie8/cases/dynamic-class-sibling-restyle.html look green to anyone?

[14:27:01.0000] <Philip`>
(in IE8)

[14:43:00.0000] <Philip`>
/me wonders what the point is in giving credit to people using two letter initials, so that nobody knows who it is and you're bound to get collisions

[14:44:00.0000] <annevk>
I suspect Hixie can extrapolate the full name

[14:45:00.0000] <annevk>
mine is three letter btw

[14:45:01.0000] <gsnedders>
mine is two letter

[14:45:02.0000] <hober>
/me contemplates "<brb>" for the browser button element... :)

[14:46:00.0000] <annevk>
<browserbutton> would work

[14:46:01.0000] <annevk>
it's not like you'd use it all over

[14:49:00.0000] <annevk>
http://www.ibm.com/developerworks/blogs/page/JohnBoyer?entry=the_ubiquity_strategy_for_promoting

[14:49:01.0000] <annevk>
"More importantly, this means that the W3C, not the browser makers, defines the web and its standards."

[14:50:00.0000] <annevk>
(the problem with all his cheering of course is that if you build everything on top of AJAX, you're still limited by what browser engines support)

[14:51:00.0000] <annevk>
/me replies

[14:51:01.0000] <Philip`>
If you build everything on top of AJAX, you don't need standards

[14:52:00.0000] <annevk>
that too :)

[14:52:01.0000] <Philip`>
You can just bundle the application with its AJAX scripts and there's no need to put up with the pain of standardisation

[14:52:02.0000] <annevk>
(though you need them for AJAX, to make sure it works in browsers)

[14:52:03.0000] <gsnedders>
I'm not built on top of AJAX, but there's no human standard

[14:52:04.0000] <annevk>
fail

[14:54:00.0000] <annevk>
Philip`, heh, thanks for backing up thoughts by facts :)

[14:55:00.0000] <Philip`>
Facts are easier than thoughts

[14:55:01.0000] <Philip`>
As long as you don't analyse them, you can't be wrong

[15:00:00.0000] <Philip`>
/me wonders why he keeps getting given mod points on Slashdot even though he let the last ~75 points expire

[15:01:00.0000] <Lachy>
Philip`, go use them to mod me up

[15:03:00.0000] <gsnedders>
Philip`: Because you comment around the median and have good karma

[15:03:01.0000] <Philip`>
Lachy: That would be immoral :-p

[15:04:00.0000] <gsnedders>
/me changes topic to 'WHATWG (HTML5) -- http://www.whatwg.org/ -- Logs: http://krijnhoetmer.nl/irc-logs/ -- Please leave your sense of logic at the door, thanks! -- gsnedders promised to dye his hair green, photos coming soon :-)'

[15:05:00.0000] <gsnedders>
Lachy: Double hyphens are used everywhere else in the topic. Please be consistent changing it!

[15:05:01.0000] <Lachy>
Philip`, here's an interesting comment from a user I don't know that should be modded up. http://ask.slashdot.org/comments.pl?sid=914095&cid=24784973

[15:05:02.0000] <gsnedders>
/me had mod points till yesterday

[15:05:03.0000] <Lachy>
gsnedders, sorry, I didn't realise

[15:05:04.0000] <gsnedders>
Lachy: heh.

[15:05:05.0000] <annevk>
 /. still popular?

[15:05:06.0000] <annevk>
/me thought it was all reddit

[15:05:07.0000] <Lachy>
annevk, no

[15:06:00.0000] <Lachy>
I never read reddit

[15:06:01.0000] <gsnedders>
/. > digg > reddit

[15:06:02.0000] <Lachy>
http://yro.slashdot.org/article.pl?sid=08/08/28/1855213

[15:06:03.0000] <annevk>
digg over reddit?

[15:07:00.0000] <annevk>
crazy

[15:07:01.0000] <gsnedders>
annevk: The content that gets on digg is better, even if the comments are even worse

[15:07:02.0000] <Philip`>
Lachy: Why do you think hard drives would last any longer while unused than DVDs?

[15:07:03.0000] <annevk>
/me looks on /.

[15:07:04.0000] <annevk>
/me finds http://yro.slashdot.org/yro/08/08/28/1855213.shtml

[15:07:05.0000] <Lachy>
annevk, I just posted that link

[15:08:00.0000] <Philip`>
/me has disabled display of the whole YRO section

[15:08:01.0000] <Lachy>
Philip`, they last longer because as I move the data around to different drives, the data is copied and backed up

[15:08:02.0000] <gsnedders>
/me thinks Philip` has no rights online

[15:08:03.0000] <gsnedders>
You girl!

[15:09:00.0000] <Lachy>
also, after I get a drobo, it will take care of maintaining the data and redundancy for me

[15:09:01.0000] <Philip`>
Lachy: Ah

[15:09:02.0000] <Philip`>
Lachy: -1 Off-topic, since copying data between drives means your point is irrelevant when considering 25-year storage :-p

[15:10:00.0000] <Lachy>
my advice wasn't to use the hard drives in the time capsule. My advice was to stick USB thumb drives in there, and also maintain an unburied copy to cheat

[15:11:00.0000] <Philip`>
Because the tiny electronic charge on flash memory is going to last for decades?

[15:13:00.0000] <Lachy>
Philip`, yes

[15:13:01.0000] <Lachy>
well, I think so. I can't remember what the life of them is, but I thought it was pretty long

[15:14:00.0000] <annevk>
I just hope my various hosting providers copy it all losslessly year after year

[15:14:01.0000] <annevk>
and if that doesn't happen, oh well

[15:15:00.0000] <gsnedders>
/me wonders how much eyeliner costs

[15:16:00.0000] <Lachy>
gsnedders, don't tell me someone talked you into wearing eyeliner as well as dying your hair?

[15:17:00.0000] <gsnedders>
Lachy: No, but if I do one, I'm doing both.

[15:17:01.0000] <gsnedders>
I wanted to do both before anyone started trying to talk me into it

[15:18:00.0000] <Lachy>
gsnedders, don't spoil the fun for us by ruining my delusion that peer pressure made you do it

[15:18:01.0000] <gsnedders>
:P

[15:18:02.0000] <gsnedders>
Lachy: The big duo of you and a girl from school

[15:19:00.0000] <gsnedders>
(Actually, she's quite a bit shorter than me, and I don't know how tall you are)

[15:19:01.0000] <Lachy>
I'm 2m tall

[15:19:02.0000] <Lachy>
I assume you're quite a bit shorter than me :-)

[15:21:00.0000] <gsnedders>
around 20cm shorter

[15:24:00.0000] <gsnedders>
I ought to write a more sensible about me page

[15:50:00.0000] <annevk>
Would the RDF idea still work good enough if you take out URIs and make it work using the CSS prefix-name model?

[15:57:00.0000] <sicking>
Hixie, ping

[15:58:00.0000] <Hixie>
heya

[15:58:01.0000] <Hixie>
wassup

[16:02:00.0000] <jgraham>
I was just thinking "oh were it that I had a whole pile of email about RDFa, why then I could be truly happy" 

[16:02:01.0000] <jgraham>
and I have

[16:02:02.0000] <annevk>
Hixie, <browserbutton> seems simple enough

[16:03:00.0000] <Hixie>
jgraham: bet you my pile is bigger (http://www.whatwg.org/issues/#rdfa) -- and i have to reply to mine...

[16:03:01.0000] <Hixie>
annevk: i guess

[16:03:02.0000] <jgraham>
Hixie: Fair enough :)

[16:03:03.0000] <jgraham>
But you get paid to do it :)

[16:07:00.0000] <annevk>
Hixie, fwiw, when you replied to the e-mail, did you fix the example typo?

[16:07:01.0000] <annevk>
doesn't seem like you did

[16:07:02.0000] <annevk>
/me reported a typo and a suggestion to change the name

[16:09:00.0000] <Hixie>
annevk: you were the third person to report the typo and i have fixed it in the source but not regenned

[16:18:00.0000] <annevk>
Hixie, the third, how nice

[16:19:00.0000] <annevk>
Hixie, based on some status thing it was reported twice internally too :)

[16:19:01.0000] <Hixie>
:-P

[16:19:02.0000] <annevk>
your typoes cost a lot of money

[16:19:03.0000] <annevk>
:)

[16:21:00.0000] <BenoitRen>
Hello.

[16:22:00.0000] <BenoitRen>
Is it all right to ask for help with HTML semantics here?

[16:22:01.0000] <annevk>
#html might be better for that, this channel is mainly concerned with developing the next version of HTML

[16:22:02.0000] <annevk>
though feel free to ask

[16:23:00.0000] <BenoitRen>
I'm already using HTML5 semantics in a way, so...

[16:23:01.0000] <Hixie>
you can ask help questions, you might just get turned into a lab rat as part of it :-)

[16:24:00.0000] <annevk>
BenoitRen, ah, HTML5 semantic questions prolly best go here, yes :)

[16:24:01.0000] <BenoitRen>
I've been working on and off on my site that hosts the script to a video game. I've been marking up the dialogue with the dl element.

[16:24:02.0000] <BenoitRen>
Basically, using it like <dialog>.

[16:24:03.0000] <annevk>
/me is completely confused by Kristof latest

[16:25:00.0000] <annevk>
/me is glad he stepped out of the thread

[16:25:01.0000] <BenoitRen>
Apart from plain dialogue, you can also talk to non-playable characters, which I mark up by using an unsorted list in the <dd> element.

[16:25:02.0000] <annevk>
Kristof's* grmbl

[16:26:00.0000] <BenoitRen>
My current problem is that during talk, things might be happening. Like, a character moves, or mist appears, etc.

[16:26:01.0000] <annevk>
ah yeah, I think we have some open issue on that

[16:27:00.0000] <BenoitRen>
I'm not sure how to handle that. Do I use close the dialog and make a <p>aragraph about it? Do I put it on a separate line in the character's dialog? Behind the line?

[16:27:01.0000] <annevk>
I think Hixie punted on it for now

[16:28:00.0000] <Hixie>
right now you have to close the dialog

[16:29:00.0000] <Hixie>
and put a paragraph

[16:29:01.0000] <BenoitRen>
"right now"?

[16:29:02.0000] <Hixie>
as in, the way the spec stands today

[16:30:00.0000] <Hixie>
i see the spec has no example of that

[16:31:00.0000] <BenoitRen>
<dl><dt>Man</dt><dd><ul><li>I like flowers.</li></ul></dd></dl><p>A pig flies by.</p><dl><dt>Man</dt><dd><ul><li>What was that?</li></ul></dd></dl><p>An explosion occurs.</p><dl><dt>Man</dt><dd><ul><li>What's happening?</li></ul></dd></dl>

[16:32:00.0000] <annevk>
why  the <ul><li> ?

[16:33:00.0000] <Hixie>
BenoitRen: that's right, except using <p> instead of <ul><li> :-)

[16:33:01.0000] <Hixie>
unless the guy is really saying a list of items :-)

[16:33:02.0000] <Hixie>
like reciting a shopping list or something :-)

[16:34:00.0000] <BenoitRen>
Because as I said above, NPCs can say multiple things by talking to them more than once, which I denote with a list. To be consistent I also use with dialogue between multiple people.

[16:34:01.0000] <annevk>
(and the <p> is semantically implied, so you could drop that too if you don't care much for styling it)

[16:34:02.0000] <BenoitRen>
*also use it with

[16:34:03.0000] <BenoitRen>
As in:

[16:34:04.0000] <Hixie>
BenoitRen: you mean multiple different things and you only pick one?

[16:35:00.0000] <BenoitRen>
Hixie: No. You talk to an NPC the first time, she says A. You talk a second time, and she says B.

[16:35:01.0000] <BenoitRen>
<dl><dt>Woman</dt><dd><ul><li>A</li><li>B</li></ul></dd></dl>

[16:36:00.0000] <Hixie>
ah, interesting

[16:36:01.0000] <annevk>
multiple <dd> would actually be better, but that's probably not allowed right now

[16:36:02.0000] <Hixie>
so is this just a data storage mechanism, or is it the actual elements that the user sees?

[16:37:00.0000] <BenoitRen>
Yeah, I recently learned you can have multiple <dd>s with one <dt>, but the HTML5 spec disallows that in <dialog>.

[16:37:01.0000] <BenoitRen>
It's for an actual web page.

[16:37:02.0000] <BenoitRen>
So yes, the user sees it.

[16:37:03.0000] <Hixie>
oh it's like a walkthrough and you're showing the various things someone says?

[16:37:04.0000] <Hixie>
interesting

[16:38:00.0000] <Hixie>
not sure we really have a good way to mark that up today. i guess <ul> is an ok workaround for this case.

[16:38:01.0000] <BenoitRen>
I guess you could see it like a walkthrough, yes. You see the entire game, like a script.

[16:38:02.0000] <Hixie>
cool

[16:39:00.0000] <Hixie>
yeah i didn't really even consider that case when writing the spec for it

[16:40:00.0000] <BenoitRen>
I first used plain text files. But then I had to start using coloured text as the game used them in special quests, so I just put it in a plain page with <pre>. and used <font>. This was back when I didn't know about the right way to do things.

[16:40:01.0000] <Hixie>
hehe

[16:40:02.0000] <BenoitRen>
Then I used <cite> with <blockquote>, and finally now <dl>.

[16:40:03.0000] <Hixie>
yeah that's why we introduced <dialog>, to make it less confusing about what the heck to use for this kind of stuff :-)

[16:41:00.0000] <BenoitRen>
Though my website only reflects my dark ages days...

[16:41:01.0000] <BenoitRen>
Yeah, it's a good idea. :)

[16:51:00.0000] <gsnedders>
Hixie: Re: <bb>, how about <bob>?

[16:51:01.0000] <gsnedders>
it's only one letter more

[16:52:00.0000] <sicking>
annevk, so am I reading the spec wrong, or does AC only let you specify one name in the Access-Control-Allow-Methods/Headers headers?

[16:52:01.0000] <Hixie>
gsnedders: heh

[16:53:00.0000] <gsnedders>
Hixie: On another note, I shipped Anolis (the new name for the spec-gen) 1.0RC1 today

[16:53:01.0000] <Hixie>
woo

[16:53:02.0000] <Hixie>
is it ready for use?

[16:53:03.0000] <gsnedders>
yup.

[16:53:04.0000] <Hixie>
sweet

[16:53:05.0000] <Hixie>
uri?

[16:53:06.0000] <gsnedders>
The only way any bugs that haven't all ready been found are going to be found is by getting thousands of people looking at a doc processed by it :)

[16:53:07.0000] <gsnedders>
http://hg.gsnedders.com/anolis/

[16:54:00.0000] <gsnedders>
I need to fix a bug on the real website, so that'll have to suffice

[16:54:01.0000] <Philip`>
Let me know if you're going to change the HTML5 spec to use it, since I'll have to rewrite all the spec-splitter's section listing

[16:55:00.0000] <gsnedders>
Philip`: Re-write it to operate as a process within anolis!

[16:55:01.0000] <gsnedders>
Philip`: Thereby not needing to parse the entire spec again

[16:56:00.0000] <BenoitRen>
So for the current spec I have to close <dialog> and use a <p>aragraph. But I'm interested in other possible solutions that could get included in the spec later. Ideas? I can't come up with more than <span class="note">earthquake</span>, and that's not semantic...

[16:56:01.0000] <Philip`>
gsnedders: That'd be trivial if there's just a lxml tree, I think

[16:56:02.0000] <Hixie>
Philip`: i'm going to change the HTML5 spec to use it

[16:56:03.0000] <gsnedders>
Philip`: There is

[16:57:00.0000] <Philip`>
Hixie: I should have said "Let me know when ..." :-)

[16:57:01.0000] <gsnedders>
Hixie: I have a draft email to you about moving to use it

[16:57:02.0000] <Hixie>
so does anyone want to host this as a web service somewhere?

[16:57:03.0000] <Hixie>
Philip`: my plan was now :-)

[16:57:04.0000] <Hixie>
Philip`: i can keep on using the old one for the spec splitter until you're ready

[16:57:05.0000] <Philip`>
gsnedders: Is there some way to plug such a processor into your spec-gen?

[16:58:00.0000] <gsnedders>
Hixie: I'll host it, if someone writes a web interface :)

[16:58:01.0000] <Philip`>
(since it shouldn't be an integrated part, because it's too hard-coded for the HTML5 spec)

[16:58:02.0000] <Hixie>
gsnedders: no need for a web interface, i just need something i can do a GET on (with a URI in a parameter) and have it return the resulting html

[16:58:03.0000] <gsnedders>
Philip`: See second paragraph http://hg.gsnedders.com/hgwebdir.cgi/anolis/raw-file/1c14b53c4a45/README.html#using-anolis

[16:59:00.0000] <Hixie>
personally i'd prefer if it wasn't actually part of the same script, to be honest :-)

[16:59:01.0000] <Hixie>
i like being able to generate te spec separte from splitting it :-)


2008-08-29
[17:01:00.0000] <Philip`>
gsnedders: If the spec-gen output well-formed XML, then the spec-splitter could load it in approximately no time and so the parsing cost would no longer be a problem

[17:01:01.0000] <annevk>
sicking, where does it say that?

[17:01:02.0000] <gsnedders>
Philip`: It can't easily output both

[17:01:03.0000] <annevk>
sicking, #Method means comma-separated list of Method names per RFC 2616

[17:01:04.0000] <annevk>
sicking, same for #field-name

[17:01:05.0000] <gsnedders>
Hixie: Creating a really basic form isn't that much more than what I meant by a web UI :)

[17:01:06.0000] <Philip`>
gsnedders: Output one file that is a subset of both HTML and XHTML :-)

[17:01:07.0000] <annevk>
sicking, # has special meaning

[17:02:00.0000] <sicking>
annevk, ah, i missed the # thing

[17:02:01.0000] <Philip`>
Hixie: I probably won't be able to do anything with the spec-splitter in the next few days, but then I'll probably forget so someone will have to remind me

[17:03:00.0000] <Hixie>
k

[17:03:01.0000] <Hixie>
np

[17:03:02.0000] <sicking>
btw, has anyone checked what XDR in IE8b2 does?

[17:04:00.0000] <annevk>
Hixie, I'm interested in the Web service

[17:04:01.0000] <annevk>
(assuming it works with curl or something)

[17:05:00.0000] <annevk>
/me doesn't like how WS-Pain took "Web Services"

[17:05:01.0000] <Hixie>
gsnedders: http://junkyard.damowmow.com/340

[17:05:02.0000] <Hixie>
gsnedders: there's your ui, fill in the "..." bit as needed :-)

[17:06:00.0000] <gsnedders>
Hixie: :D

[17:06:01.0000] <annevk>
Hixie, doesn't work

[17:06:02.0000] <gsnedders>
Hixie: The problem is I've never written any web thing before in Python, and I don't particularly care for a web ui

[17:06:03.0000] <annevk>
Hixie, http://junkyard.damowmow.com/...?url=http%3A%2F%2Fdev.w3.org%2F2006%2Fwaf%2Faccess-control%2FOverview.src.html

[17:06:04.0000] <Hixie>
annevk: :-P

[17:06:05.0000] <gsnedders>
annevk: look at the source.

[17:06:06.0000] <gsnedders>
annevk: It's a UI. Nothing more.

[17:06:07.0000] <gsnedders>
<form action="..." method=get>

[17:06:08.0000] <Hixie>
gsnedders: ok i can hook something up

[17:07:00.0000] <annevk>
oh, I see, lol

[17:07:01.0000] <Hixie>
gsnedders: is there a tarball somewhere?

[17:07:02.0000] <Hixie>
annevk: i said it was the ui, nothing else :-P

[17:07:03.0000] <gsnedders>
Hixie: See the download links from http://anolis.gsnedders.com/

[17:07:04.0000] <gsnedders>
Hixie: See the "anolis" and "runtests.py" files

[17:08:00.0000] <gsnedders>
Hixie: If you wanted to be über-clever, you could probably actually run it as a pre-commit hook in SVN

[17:09:00.0000] <gsnedders>
Anyhow, I'm off to bed

[17:10:00.0000] <gsnedders>
http://stuff.gsnedders.com/spec-gen/html5.html is a version of the spec from earlier today

[17:11:00.0000] <Hixie>
anyone know how i install to a non-default location, like "." ?

[17:11:01.0000] <gsnedders>
Hixie: I could just send you the email I wrote a while ago, though one or two bits are a bit outdated, but if you ignore the stuff about 1.0b2, I think it's fine

[17:11:02.0000] <Hixie>
sure

[17:11:03.0000] <gsnedders>
Hixie: that's in the email

[17:11:04.0000] <Hixie>
send send send! :-D

[17:11:05.0000] <gsnedders>
/me sends

[17:11:06.0000] <Hixie>
wooo

[17:11:07.0000] <gsnedders>
it also refers to the program as spec-gen, while it's now anolis

[17:14:00.0000] <Hixie>
byte-compiling /home/ianh/bin/anolis-spec-gen/lib/python2.3/site-packages/anolislib/processes/toc.py to toc.pyc

[17:14:01.0000] <Hixie>
  File "/home/ianh/bin/anolis-spec-gen/lib/python2.3/site-packages/anolislib/processes/toc.py", line 191

[17:14:02.0000] <Hixie>
    sections.extend((child_section, depth + 1) for child_section in reversed(section))

[17:14:03.0000] <Hixie>
                                                 ^

[17:14:04.0000] <Hixie>
SyntaxError: invalid syntax

[17:14:05.0000] <annevk>
(Anolis name flames can be e-mailed directly to annevan...)

[17:16:00.0000] <Hixie>
i'd fix that line but i have no idea what it's trying to do

[17:17:00.0000] <annevk>
create a list of tuples?

[17:17:01.0000] <gsnedders>
Hixie: Py 2.3 :(

[17:17:02.0000] <annevk>
Hixie, ask markp :)

[17:17:03.0000] <annevk>
prolly easiest

[17:18:00.0000] <gsnedders>
Hixie: Trying putting another ( at the beginning, and ), before the final )

[17:18:01.0000] <gsnedders>
(note the comma)

[17:18:02.0000] <gsnedders>
Actually, that's nonsense

[17:18:03.0000] <gsnedders>
[ and ,]

[17:19:00.0000] <gsnedders>
peh.

[17:19:01.0000] <gsnedders>
I'll deal with this tomorrow

[17:19:02.0000] <gsnedders>
Hixie: Install a newer version of python :P

[17:19:03.0000] <annevk>
for child_section in reversed(section):

[17:19:04.0000] <annevk>
temp_list.append((child_section, depth + 1))

[17:19:05.0000] <annevk>
sections.extend(temp_list)

[17:20:00.0000] <gsnedders>
annevk: I lost the whitespace in that

[17:20:01.0000] <annevk>
i didn't put any in

[17:20:02.0000] <Hixie>
annevk: i don't need to declare temp_list or something?

[17:20:03.0000] <annevk>
yeah, that too

[17:20:04.0000] <annevk>
temp_list = []

[17:20:05.0000] <annevk>
and some spaces before temp_list.append

[17:20:06.0000] <gsnedders>
There probably is a better way of fixing that though

[17:21:00.0000] <Hixie>
woot it compiled

[17:21:01.0000] <annevk>
yay for remote debugging

[17:21:02.0000] <gsnedders>
sections.extend([(child_section, depth + 1) for child_section in reversed(section)])

[17:21:03.0000] <annevk>
(aka guessing)

[17:21:04.0000] <gsnedders>
That should work

[17:22:00.0000] <annevk>
isn't it just that inline for doesn't work in Python 2.3?

[17:22:01.0000] <gsnedders>
(I think)

[17:22:02.0000] <gsnedders>
/me shrugs

[17:22:03.0000] <gsnedders>
I only claim to support 2.5 anyway :P

[17:22:04.0000] <gsnedders>
Mainly because I'm too lazy to test

[17:23:00.0000] <gsnedders>
Umm…

[17:23:01.0000] <Hixie>
something about a missing xProfile

[17:23:02.0000] <gsnedders>
It will fail to run on 2.3

[17:23:03.0000] <Hixie>
cProfile even

[17:24:00.0000] <gsnedders>
reversed() doesn't exist either

[17:24:01.0000] <Hixie>
awww, failed upgrade. :-(

[17:24:02.0000] <Hixie>
oh well

[17:25:00.0000] <Hixie>
i look forward to RC2!

[17:25:01.0000] <annevk>
there's a way to get newer python versions on dreamhost

[17:25:02.0000] <Hixie>
oh?

[17:25:03.0000] <gsnedders>
annevk: sections.extend() (above) is a generator expression, which is 2.4

[17:25:04.0000] <annevk>
/me looks for a wiki page

[17:25:05.0000] <gsnedders>
Hixie: supporting Py2.3 would mean moving back on to deprecated features

[17:26:00.0000] <Hixie>
well i have some stuff to do now anyway

[17:26:01.0000] <Hixie>
i'll be back in a bit

[17:26:02.0000] <annevk>
eg http://wiki.dreamhost.com/Python#Building_a_custom_version_of_Python

[17:27:00.0000] <jgraham>
/me has python2.5 at james.html5.org

[17:27:01.0000] <gsnedders>
It should be fairly easy to get it working on 2.4, it probably already does

[17:28:00.0000] <jgraham>
gsnedders: FWIW it would be nice if there were a utility function in anolislib that took a file-like object and a set of options and returned a processed document

[17:28:01.0000] <jgraham>
So the web interface would just be

[17:28:02.0000] <jgraham>
doc = urllib2.urlopen(url)

[17:28:03.0000] <gsnedders>
Yeah, it would be nice

[17:28:04.0000] <gsnedders>
A lot of things would be nice :)

[17:28:05.0000] <jgraham>
anolislib.gen(doc)

[17:29:00.0000] <jgraham>
gsnedders: Well all the code for thi thing is already in anolis. It's just refactoring

[17:29:01.0000] <gsnedders>
jgraham: The processed doc being a StringIO object, or an ElementTree, or…?

[17:29:02.0000] <annevk>
btw, the Web interface (whoever will set it up) needs file input as well

[17:29:03.0000] <annevk>
the Web service rather

[17:30:00.0000] <jgraham>
gsnedders: Don't really care. I guess an ElementTree is best but a string is more convenient

[17:30:01.0000] <jgraham>
(a StringIO would be weird)

[17:35:00.0000] <gsnedders>
/me heads off, again

[17:41:00.0000] <annevk>
Hixie, using #!/usr/bin/python2.5 at the start of the file should do the trick

[17:41:01.0000] <annevk>
Hixie, through shell I suppose you need to use python2.5 as programm

[17:42:00.0000] <annevk>
s/programm/program/

[17:44:00.0000] <annevk>
/me -> bed

[19:06:00.0000] <Hixie>
jgraham: if you're still around, i could totally use a service on james.html5.org if you set one up

[19:06:01.0000] <Hixie>
i'd rather not have to set things up so that the script runs on my machine, for load reasons

[20:30:00.0000] <sicking>
Hixie, can you convince the gmail team to write a tool to import IMAP mail?

[20:30:01.0000] <sicking>
Hixie, pretty please?

[20:53:00.0000] <MikeSmith>
Hixie: <butt>

[20:53:01.0000] <MikeSmith>
I don't think there is likely much existing content with that in it

[20:57:00.0000] <roc>
sicking: gmail supports imap, so can't you just pull up an IMAP client (say Thunderbird), open both accounts, and drag-and-drop from one to the other?

[20:58:00.0000] <sicking>
roc, yeah, that should work in theory

[20:58:01.0000] <sicking>
roc, though last i tried thunderbird didn't like connecting to gmail as IMAP, dunno why

[20:59:00.0000] <sicking>
roc, and last i copied mail between accounts using thunderbird i basically lost the majority of mails due to thunderbird bugs

[20:59:01.0000] <roc>
hehe

[20:59:02.0000] <roc>
just write a script to talk IMAP directly to the two servers

[20:59:03.0000] <roc>
how hard could it be

[20:59:04.0000] <sicking>
dunno how hard imap is

[20:59:05.0000] <sicking>
you might be right

[20:59:06.0000] <roc>
it's ugly

[21:00:00.0000] <sicking>
i hear python solves all programming problems ever

[21:00:01.0000] <sicking>
import IMAP-simple

[21:00:02.0000] <sicking>
?

[21:22:00.0000] <jacobolus>
roc, sicking: yes, opening up an imap client and moving emails into Gmail works just fine

[21:22:01.0000] <jacobolus>
sicking: if you want you can just add emails there using the IMAP protocol directly

[21:22:02.0000] <jacobolus>
write a little script or something

[21:22:03.0000] <sicking>
yeah

[21:23:00.0000] <sicking>
i just found that thunderbird has a copy function though, so i can probably use that

[21:23:01.0000] <sicking>
rather than the move which i tried last i did something like this

[21:59:00.0000] <roc>
is "seeked" really a word?

[22:00:00.0000] <jacobolus>
yes

[22:00:01.0000] <jacobolus>
erm, no

[22:00:02.0000] <jacobolus>
sought

[22:00:03.0000] <sicking>
sunk?

[22:00:04.0000] <sicking>
:)

[22:01:00.0000] <roc>
Hixie thinks it is :-) http://www.whatwg.org/specs/web-apps/current-work/#seeked

[22:02:00.0000] <jacobolus>
yeah, seeked sucks as a name for that

[22:02:01.0000] <jacobolus>
I'm not sure sought is much better though

[22:02:02.0000] <jruderman>
'doneseeking'?

[22:05:00.0000] <jacobolus>
roc: then again, networking specs have a proud tradition of mis-spelling words, e.g. 'referer'

[22:13:00.0000] <Hixie>
seeked isn't misspelt. it's just a made up word.

[22:13:01.0000] <Hixie>
onsought would confusing the heck out of people

[22:14:00.0000] <Hixie>
confuse, even

[22:14:01.0000] <sicking>
doneseeking sounds like a good idea to me

[22:14:02.0000] <GregHouston>
Could throw some nice archaic terms in there like beseeched.

[22:14:03.0000] <Hixie>
what's wrong with 'onseeked'?

[22:15:00.0000] <sicking>
it's a made up word

[22:17:00.0000] <Hixie>
so? half of our industry uses made up words

[22:17:01.0000] <sicking>
no other event names do

[22:17:02.0000] <sicking>
s/do/are/

[22:18:00.0000] <sicking>
why is onseeked better than ondoneseeking?

[22:19:00.0000] <Hixie>
there isn't a single other event of the form ondoneX

[22:20:00.0000] <sicking>
that seems like less of an addition than adding made up words

[22:20:01.0000] <Hixie>
we have lots of made up words :-)

[22:21:00.0000] <sicking>
we do? and not in event names

[22:21:01.0000] <sicking>
and are those truly made up, or using commonly used but not in marriam webster, words?

[22:22:00.0000] <GregHouston>
One convention is to add onsuccess or oncomplete to the event name, e.g., seekoncomplete, not that seeked bothers me any.

[22:23:00.0000] <sicking>
onseekcomplete sounds good too

[22:23:01.0000] <roc>
afterseek?

[22:23:02.0000] <GregHouston>
Sounds like a nightclub roc. :)

[22:24:00.0000] <Hixie>
i have approximately zero interest in worrying about event names, so if y'all realy do have a problem with 'onseeked', then implement something else and tell me what it is

[22:24:01.0000] <roc>
I don't care all that much either to be honest

[22:24:02.0000] <sicking>
same here, but it seems easy to fix

[22:24:03.0000] <Hixie>
(make sure the apple guys implement it too)

[22:24:04.0000] <Hixie>
(and opera i guess)

[22:25:00.0000] <sicking>
do either of them do seeked?

[22:25:01.0000] <roc>
Hixie, from my reading of this seek spec, if the seek fails for some reason, we should just fire a 'seeked' event and its up to the script to figure out that the seek didn't get to the right place

[22:25:02.0000] <Hixie>
sicking: no idea :-)

[22:25:03.0000] <roc>
is that correct?

[22:25:04.0000] <roc>
I mean if it fails around step 13

[22:26:00.0000] <Hixie>
roc: correct; the event is really only meant for putting up "Seeking..." UI

[22:26:01.0000] <roc>
ok

[22:28:00.0000] <roc>
sicking: you might want to avoid posting with the name "Domain Admin"

[22:29:00.0000] <sicking>
roc, ugh!

[22:29:01.0000] <sicking>
roc, thanks

[23:40:00.0000] <Dashiva>
This has to be a new record in pointlessness: http://lists.w3.org/Archives/Public/public-html/2008Aug/0842.html

[23:41:00.0000] <jacobolus>
Dashiva: don't tempt the internets

[23:43:00.0000] <Dashiva>
I wear hubris like a crown already

[00:12:00.0000] <MikeSmith>
Dashiva: If they had a King of Fools, then I could wear that crown... and you can all die laughing because I'll wear it proudly.

[00:13:00.0000] <MikeSmith>
wow, a plus one with double exclamation

[00:13:01.0000] <MikeSmith>
I think that earns extra points in the olympics

[00:14:00.0000] <MikeSmith>
at least with the east German judges

[00:15:00.0000] <GregHouston>
It's really our fault as developers. We encourage everyone to grade every bit of content with little ajax star selectors.

[00:16:00.0000] <jacobolus>
GregHouston: as in, five stars, $('* * * * *').… ??

[00:19:00.0000] <jacobolus>
ouch: http://intertwingly.net/blog/2008/08/28/Improved-Namespace-Support

[00:19:01.0000] <GregHouston>
Right, !! probably equates to "I wish I could have given it six stars!"

[00:20:00.0000] <jacobolus>
i guess ppl already saw that

[00:25:00.0000] <GregHouston>
Jacob, I'm a little slow, I just got the CSS selector part of your joke.

[00:26:00.0000] <jacobolus>
GregHouston: I thought it was *your* joke?!

[00:26:01.0000] <GregHouston>
Lol, I think you took my joke a step further and made it geekier. Haha.

[00:30:00.0000] <hdh>
howcome two votes make a 4.7 score? does microsoft round down?

[00:31:00.0000] <GregHouston>
Lol. After making fun of voting. I actually made that second vote, and I asked myself the same question. I thought maybe I didn't have cursor far enough over to the right when I clicked on the fifth star.

[00:34:00.0000] <GregHouston>
hdh, I'm assuming you followed the same path I did and landed here: https://connect.microsoft.com/IE/feedback/ViewFeedback.aspx?FeedbackID=364356

[00:35:00.0000] <hdh>
right; that's the only thing interesting (apart from I can write ;-- in Connect password)

[00:50:00.0000] <Dashiva>
"So we should raise the conformance standards for accessibility, but also raise the conformance standard for authoring tool so that they re even required to produce non-conforming HTML whenever the author provides insufficient information to make the document conforming."

[00:50:01.0000] <Dashiva>
"We'll also raise the conformance standard for spec writers, by making it non-conforming not to shoot yourself in the shoot."

[00:57:00.0000] <GregHouston>
In Javascript when I shoot myself in the shoot I get a too much recursion error.

[00:58:00.0000] <Dashiva>
In spec-writing, that's just an implementation detail

[01:01:00.0000] <Dashiva>
Hixie: Has there been much brainstorming about <bb> names?

[01:02:00.0000] <Hixie>
some

[01:02:01.0000] <Hixie>
see the logs

[01:03:00.0000] <Dashiva>
/me tries to search for <bb> and gets every mention of bb.online.no

[01:03:01.0000] <zcorpan>
jgraham: fwiw, it's possible to exploit a security parsing bug in older operas (pre-9.52 i think) with unquoted attributes and html5lib, so using quoted attributes by default (at least if it contains any offending characters) would be good

[01:04:00.0000] <Hixie>
oooh, hsivonen's idea of an XSLT-specific PUBLIC FPI makes sense

[01:04:01.0000] <Dashiva>
Hixie: What about <ui> or <uibtn> or <uibutton> or something like that? Would that be misrepresenting the purpose?

[01:05:00.0000] <zcorpan>
jgraham: consider <a href=#{U+000A}onclick=doSomethingEvil()> ... or hmm, does html5lib emit quotes if the value contains an = ?

[01:05:01.0000] <Hixie>
Dashiva: maybe

[01:05:02.0000] <Hixie>
Dashiva: send mail on the thread, it's not really something i want to worry about now

[01:06:00.0000] <Dashiva>
Hey, not outright rejection. That's a good start :)

[01:06:01.0000] <Hixie>
:-)

[01:07:00.0000] <Hixie>
i'd want something that is clearly different from <command> and <button> if possible

[01:08:00.0000] <zcorpan>
<differentfromcommandandbutton>

[01:09:00.0000] <zcorpan>
clear enough?

[01:09:01.0000] <Hixie>
no :-P

[01:09:02.0000] <Hixie>
hsivonen: maybe "XSLT-generated"?

[01:10:00.0000] <zcorpan>
<clearlydifferentfromcommandandbutton>

[01:10:01.0000] <Hixie>
too long

[01:10:02.0000] <Hixie>
:-P

[01:10:03.0000] <zcorpan>
:(

[01:10:04.0000] <zcorpan>
<cdfcab>

[01:11:00.0000] <Dashiva>
<cdfcab>

[01:11:01.0000] <Dashiva>
dang

[01:12:00.0000] <Dashiva>
That's not so bad, either. CDF is a popular acronym, and it drives around in a cab.

[01:12:01.0000] <GregHouston>
It's sort of like YAML, "YAML Ain't a Markup Language"

[01:12:02.0000] <hdh>
CDF Cabinet, to go with the zipped-XML fever

[01:14:00.0000] <zcorpan>
hmm, crappy that one needs to reboot for ie8b2

[01:14:01.0000] <Dashiva>
On the other hand, it might make people think it's for embedding widgets

[01:16:00.0000] <hdh>
I heard someone rebooted twice, and contemplated clicking "Cancel" after the old IE is gone and IE8 not yet installed

[01:24:00.0000] <zcorpan>
hmm. how does one know if ie8b2 was installed successfully? any characteristics to check for that differentiates it from ie8b1?

[01:25:00.0000] <zcorpan>
:first-line and :first-letter doesn't seem to work in standards mode

[01:28:00.0000] <Lachy>
zcorpan, Help>About

[01:28:01.0000] <Lachy>
you may need to turn on the menu bar. it's off by default

[01:29:00.0000] <zcorpan>
Lachy: it just says "Beta"

[01:30:00.0000] <zcorpan>
version 8.0.6001.17184

[01:30:01.0000] <zcorpan>
hmm that's beta 1 it seems

[01:30:02.0000] <zcorpan>
why wasn't beta 2 installed then

[01:31:00.0000] <jacobolus>
the menu bar is off by default?

[01:31:01.0000] <jacobolus>
how does that work?

[01:32:00.0000] <zcorpan>
i unchecked the "install updates" during the install process, did that make the installer not do anything?

[01:33:00.0000] <Lachy>
zcorpan, mine says "Update versions: beta 2"

[01:33:01.0000] <Lachy>
Version 8.0.6001.18241

[01:36:00.0000] <zcorpan>
Lachy: ok, running the installer again, this time with the checkbox checked

[01:37:00.0000] <zcorpan>
i would have presumed that unchecking the checkbox meant "only install ie8b2, don't update windows and everything else"

[01:37:01.0000] <zcorpan>
and not "don't do anything"

[01:38:00.0000] <hdh>
http://homepage.mac.com/bradster/iarchitect/controls.htm#element(/1/2/19) happened before

[01:38:01.0000] <hdh>
search for "Grrr"

[01:50:00.0000] <zcorpan>
still no luck :(

[01:50:01.0000] <zcorpan>
oh well

[01:55:00.0000] <Lachy>
zcorpan, try uninstalling beta 1 first, and then installing beta 2

[02:00:00.0000] <collinjackson>
should the sandbox attribute of iframes be settable dynamically?

[02:18:00.0000] <zcorpan>
does someone in the html5 community know stuff about silverlight?

[02:19:00.0000] <Hixie>
what stuff?

[02:19:01.0000] <zcorpan>
i mean if we don't know anything about it how are we supposed to make html5 competitive?

[02:20:00.0000] <zcorpan>
/me rebooting

[02:22:00.0000] <roc>
people know things about it

[02:23:00.0000] <hsivonen>
it seems to me that the Open Web is lacking a killer IDE compared to Flash and Silverlight

[02:24:00.0000] <roc>
tools are definitely a big problem

[02:24:01.0000] <Hixie>
yes

[02:24:02.0000] <Hixie>
GWT and Firebug are basically it right now

[02:25:00.0000] <hsivonen>
GWT doesn't do vector graphics yet, though

[02:25:01.0000] <hsivonen>
and using GWT with Eclipse doesn't let you arrange SVG stuff graphically

[02:26:00.0000] <roc>
it's hard to figure out what to do in the tools space

[02:27:00.0000] <zcorpan>
so there's obviously a hole in the market for a standards based IDE

[02:27:01.0000] <zcorpan>
why isn't it filled?

[02:27:02.0000] <zcorpan>
i mean

[02:27:03.0000] <hsivonen>
HTML, SVG and CSS are too expressive to make a GUI tool import everything

[02:27:04.0000] <zcorpan>
an IDE for standards based apps

[02:28:00.0000] <hsivonen>
however, polished tool could output to HTML, SVG, CSS and JS

[02:28:01.0000] <roc>
there's no money in it and it's a lot of work

[02:29:00.0000] <zcorpan>
why is there no money in it?

[02:30:00.0000] <zcorpan>
would it be possible to write a plugin for existing IDEs that doesn't change the editing process, just the output?

[02:30:01.0000] <roc>
most tools are free

[02:31:00.0000] <roc>
only the most entrenched tools have people paying for them

[02:32:00.0000] <hsivonen>
we need a google search field in IDEs. :-)

[02:32:01.0000] <roc>
you might want to check this out: http://www.evolus.vn/Pencil/

[02:32:02.0000] <zcorpan>
/me ponders whether it would be Good for Google to develop an IDE for standards based apps

[02:33:00.0000] <hsivonen>
GWT lacks an Interface Builder with flashy vector graphics eye candy goodness

[02:33:01.0000] <roc>
another problem is that for Web apps you usually have a big server piece

[02:33:02.0000] <Hixie>
there are plenty of wysiwyg tools

[02:34:00.0000] <roc>
so you need to integrate with PHP or Java or ASP.NET or whatever

[02:34:01.0000] <Hixie>
dreamweaver

[02:34:02.0000] <Hixie>
frontpage

[02:34:03.0000] <Hixie>
nvu

[02:34:04.0000] <Hixie>
etc

[02:34:05.0000] <roc>
but they suck

[02:34:06.0000] <Hixie>
but they suck, as you say

[02:34:07.0000] <roc>
NVu is obsolete

[02:34:08.0000] <hsivonen>
roc: but Flash and Silverlight apps need a server side whenever an Open Web app would need a server side

[02:34:09.0000] <Hixie>
what's needed is an IDE (app creator, not a web page creator) that does both client side and server side and does a good job

[02:35:00.0000] <Hixie>
that's non-trivial

[02:35:01.0000] <zcorpan>
Hixie: wouldn't providing such an IDE be Good for Google?

[02:35:02.0000] <Hixie>
yes

[02:35:03.0000] <Hixie>
google has GWT and App Engine in this space

[02:36:00.0000] <Hixie>
but we've only just come into this space

[02:36:01.0000] <Hixie>
i mean, in the last year or so

[02:36:02.0000] <Hixie>
we're still learning the ropes

[02:36:03.0000] <hsivonen>
interestingly, all of Flash, Silverlight and GWT treat semantic HTML as a bug

[02:36:04.0000] <Hixie>
and IDE's aren't our core competency (search and ads)

[02:36:05.0000] <Hixie>
yeah, actually doing it right (semantically) in a wysiwyg environment is an unsolved problem

[02:37:00.0000] <hsivonen>
I like the GWT compiler, but I have reservations about the toolkit side

[02:37:01.0000] <hsivonen>
but compilers are easier than user-facing stuff in a way

[02:37:02.0000] <jgraham>
/me notes that seked is a real word per OED

[02:37:03.0000] <jgraham>
s/seked/seeked/

[02:37:04.0000] <hsivonen>
for example, the Google Docreader (a GWT app) breaks Google indexability

[02:38:00.0000] <hsivonen>
there's some irony there

[02:38:01.0000] <roc>
what I'd really like to do is something a bit more disruptive than taking on the IDE space head-on

[02:38:02.0000] <Hixie>
yeah i don't like the docreader

[02:39:00.0000] <Hixie>
roc: like what?

[02:39:01.0000] <roc>
and that is, leverage the browser by improving contenteditable/designmode

[02:39:02.0000] <roc>
get SVG into wikis and email that way

[02:39:03.0000] <Hixie>
well i'm all about that, but i think that would be solving a different problem

[02:39:04.0000] <roc>
user-created content

[02:40:00.0000] <roc>
part of the problem with IDEs is that a lot of the market is a big enterprise nightmare

[02:41:00.0000] <roc>
lots of stupid checkbox requirements, integration with byzantine processes, etc

[02:42:00.0000] <roc>
I should find out what the environments for developing for App Engine and Facebook are like

[02:42:01.0000] <Hixie>
hsivonen: btw i commented on the doctype idea in the logs while you were gone -- basically i think your idea is a good one and i'd suggest using "XSLT-generated" as the name

[02:43:00.0000] <hsivonen>
Hixie: nice. then one issue can be closed

[02:47:00.0000] <annevk>
(people might argue it doesn't quite close the issue with respect to <source>, <eventsource>, <command>, <embed>, etc.)

[02:47:01.0000] <Hixie>
well there's nothing we can do about those

[02:48:00.0000] <Hixie>
especially embed

[02:48:01.0000] <hsivonen>
Hixie: there is something you could do about the others but <embed>. it just wouldn't be elegant

[02:50:00.0000] <zcorpan>
what's special about embed?

[02:50:01.0000] <Hixie>
it's already implemented

[02:50:02.0000] <annevk>
actually, can't XSLT output <source />

[02:50:03.0000] <annevk>
?

[02:50:04.0000] <annevk>
because that would toaly be compliant

[02:50:05.0000] <Hixie>
/me admits not caring one iota :-)

[02:50:06.0000] <hsivonen>
annevk: it can, if you use XML output and you are very careful, IIRC

[02:51:00.0000] <annevk>
/me cares slightly about the endless whining

[02:51:01.0000] <zcorpan>
Hixie: you could allow an end tag that doesn't do anything

[02:51:02.0000] <Hixie>
wouldn't be elegant

[02:51:03.0000] <zcorpan>
indeed

[02:51:04.0000] <annevk>
and confusing as hell

[02:51:05.0000] <zcorpan>
indeed

[02:52:00.0000] <Hixie>
i find mail filters to be more effective at stopping the whining than spec changes

[02:52:01.0000] <zcorpan>
though, people already write <embed ...></embed>

[02:52:02.0000] <Hixie>
:-)

[02:52:03.0000] <hsivonen>
/me mumbles something about </p> being confusing

[02:52:04.0000] <zcorpan>
</br> anyone?

[02:52:05.0000] <zcorpan>
actually </p> is more confusing

[02:53:00.0000] <zcorpan>
with <p><table></table></p> being the winner

[02:53:01.0000] <zcorpan>
or <p><object></p></object>

[02:54:00.0000] <hsivonen>
zcorpan: that's Hixie's fault even! :-)

[02:54:01.0000] <hsivonen>
<p><table></table></p> that is

[02:55:00.0000] <Hixie>
yeah yeah

[02:55:01.0000] <Hixie>
:-P

[02:55:02.0000] <jgraham>
couldn't we use "legacy-doctype-compat" or something rather than "XSLT-compat" or "XSLT-generated" as the name. There are other tools that insist on having PUBLIC sections in the doctype

[02:55:03.0000] <Hixie>
i cared about standards back when i wrote acid2...

[02:55:04.0000] <zcorpan>
hsivonen: <p><ol></ol></p>

[02:55:05.0000] <Hixie>
jgraham: like which?

[02:55:06.0000] <jgraham>
Hixie: I think the serilization functions in libxml2 (or at least lxml), for example

[02:56:00.0000] <zcorpan>
actually i'm fine with table acting like ol if we can make it do so in quirks mode too

[02:56:01.0000] <Dashiva>
The RDF thread reaches new heights. "If the domain becomes unavailable, you just change the namespace!"

[02:56:02.0000] <Hixie>
hsivonen, zcorpan, Philip`: there was a thread about making unquoted attribute values containing "=" conforming. i'm thinking of going with no change (leave it non-conforming), based on the data in that thread showing that it does catch some subtle authoring mistakes. any objections?

[02:56:03.0000] <jgraham>
Hixie: btw I'm happy to run anolis off jame.html5.org 

[02:56:04.0000] <Hixie>
jgraham: that would rock

[02:56:05.0000] <zcorpan>
Hixie: ok with me

[02:57:00.0000] <Philip`>
What limitations does html5.org have on long-running processes?

[02:57:01.0000] <annevk>
Dashiva, yeah, and then you take a third, unrelated domain, to describe the mapping

[02:57:02.0000] <Hixie>
jgraham: i'm not so worried about libxml2 serialisation, it doesn't have the high profile in w3c space that xslt does, and can be much more easily fixed (since it's just software, not a standard)

[02:57:03.0000] <jgraham>
Philip`: It's not particularly long running is it?

[02:57:04.0000] <Philip`>
jgraham: It takes tens of CPU-seconds just to parse the spec

[02:58:00.0000] <annevk>
I don't think I put limits in place...

[02:58:01.0000] <Hixie>
hsivonen, zcorpan, Philip`: ok i'm not making the aforementioned change. i'm also not replying to your e-mails on the subject. :-)

[02:58:02.0000] <Hixie>
(nobody else posted on the thread)

[02:58:03.0000] <Dashiva>
annevk: And you need a fourth domain to host the address of the third?

[02:58:04.0000] <annevk>
but I don't have DreamHost PS either

[02:59:00.0000] <annevk>
Dashiva, I think so, and then you google for prefix:localname to locate them all

[02:59:01.0000] <Philip`>
Dashiva: Have a persistent DNS service, which makes whateveryouwant.pdns.org be a CNAME to the actual domain the namespace content is hosted on, so it can be easily changed by someone other than the owner of the hosting domain

[02:59:02.0000] <jgraham>
Hixie: Well it seems to me that having a slightly more generic name that still emphasises legacy deals with any XSLT-like cases, thus not promoting XSLT as somehow special, but I don't care very much :)

[03:00:00.0000] <Dashiva>
It's like a giant game of pass-the-SPOF-buck

[03:00:01.0000] <annevk>
haha, jgraham is using more MB than everything I have on DreamHost combined

[03:00:02.0000] <zcorpan>
Hixie: nice to remove emails from your folders without changing the spec or replying to them, isn't it? :)

[03:01:00.0000] <Hixie>
zcorpan: :-D

[03:01:01.0000] <jgraham>
annevk: I have an entire custom python environment on there :)

[03:01:02.0000] <Hixie>
actually i'm going to reply to one of them after all

[03:01:03.0000] <annevk>
Philip` is slightly behind me

[03:01:04.0000] <Hixie>
since you proposed making <foo bar=> non-conforming, and that seems like a good idea

[03:01:05.0000] <zcorpan>
/me doesn't remember why he argued to make that non-conforming

[03:02:00.0000] <Hixie>
<foo bar=> add baz=1 becomes <foo bar= baz=1>

[03:02:01.0000] <annevk>
everything together it's still below 800MB out of 350GB

[03:02:02.0000] <zcorpan>
ah

[03:03:00.0000] <zcorpan>
wonder if i should set up a filter for "RDFa"

[03:04:00.0000] <Hixie>
doesn't work

[03:04:01.0000] <hsivonen>
Hixie: leaving = in unquoted attribute values non-conforming makes sense to me

[03:04:02.0000] <Hixie>
not enough of these e-mails mention the term rdfa

[03:04:03.0000] <Philip`>
Hixie: Unquoted attributes are ugly and cause errors, so I'm quite happy for any subset of them to be non-conforming :-)

[03:04:04.0000] <Hixie>
hsivonen: yeah, it was your e-mail arguing that that convinced me :-)

[03:04:05.0000] <zcorpan>
Hixie: they do in the subject line

[03:05:00.0000] <zcorpan>
hmm not all though

[03:05:01.0000] <zcorpan>
i guess holding down space is good enough for me

[03:06:00.0000] <Hixie>
gmail's "m" feature would come in handy for this kind of thing if i used gmail

[03:06:01.0000] <Dashiva>
At the moment, it's more like "every thread on whatwg that isn't about workers"

[03:06:02.0000] <zcorpan>
yeah

[03:06:03.0000] <zcorpan>
and i don't really care about workers either

[03:06:04.0000] <zcorpan>
:P

[03:06:05.0000] <annevk>
hah, html5.org has twice as much traffic as annevankesteren.nl in terms of bandwidth

[03:07:00.0000] <hsivonen>
Hixie: good to know I'm consistent enough that I don't need to recap what email I wrote :-)

[03:07:01.0000] <Hixie>
:-)

[03:07:02.0000] <Philip`>
annevk: Looks like a fifth of my space usage is logs, and half of the rest is canvas tests, and the rest is random stuff

[03:07:03.0000] <zcorpan>
annevk: from web-apps-tracker?

[03:07:04.0000] <Philip`>
(and it's only ~100MB in total)

[03:07:05.0000] <annevk>
looking into actual stats now

[03:07:06.0000] <Hixie>
oh hey

[03:07:07.0000] <Hixie>
anne

[03:07:08.0000] <Hixie>
while you're at it

[03:07:09.0000] <Hixie>
any chance you can set up a workers-tracker?

[03:07:10.0000] <annevk>
lately it has been 10.000 requests a day

[03:08:00.0000] <Hixie>
10000?! or 10?

[03:08:01.0000] <annevk>
10000

[03:08:02.0000] <Hixie>
wow

[03:08:03.0000] <Hixie>
oh wait do you have test cases on there that go through spartan?

[03:08:04.0000] <annevk>
no

[03:08:05.0000] <jgraham>
For everything?

[03:08:06.0000] <annevk>
yes, but excluding subdomains

[03:08:07.0000] <Hixie>
all spiders then? :-)

[03:10:00.0000] <Philip`>
wc -l for philip.html5.org's logs says: 6270 access.log.2008-08-27 7969 access.log.2008-08-28

[03:10:01.0000] <Philip`>
Uh, with newlines in appropriate places

[03:10:02.0000] <annevk>
most are actually browsers

[03:10:03.0000] <Philip`>
(Why does irssi keep stripping them out?)

[03:10:04.0000] <annevk>
with IE leading?!

[03:10:05.0000] <annevk>
lol, IE5 is leading, something is on crack

[03:10:06.0000] <zcorpan>
annevk: a bot masking as ie5?

[03:11:00.0000] <annevk>
I guess

[03:12:00.0000] <zcorpan>
i learned the other day that aria-hidden is semantically empty as implemented

[03:12:01.0000] <Hixie>
isn't all of aria semantically empty by definition?

[03:12:02.0000] <Philip`>
I see a couple of hundred requests from IE5 a day, almost all to http://philip.html5.org/tools/parser/

[03:13:00.0000] <Philip`>
which doesn't make much sense

[03:13:01.0000] <Philip`>
(Must be bots since there's no requests to that page's scripts)

[03:13:02.0000] <Hixie>
aria is the equivalent of presentational attributes for accessibility apis, no?

[03:13:03.0000] <zcorpan>
Hixie: i mean, the presence of aria-hidden doesn't do anything *at all*

[03:13:04.0000] <Hixie>
oh

[03:13:05.0000] <Hixie>
why not?

[03:13:06.0000] <annevk>
then they might as well drop it in favor of hidden

[03:14:00.0000] <hsivonen>
zcorpan: doesn't aria-hidden=true prune the accessible tree?

[03:14:01.0000] <zcorpan>
firefox just looks at what is being rendered

[03:14:02.0000] <zcorpan>
if it's display:none, it's dropped from the accessible tree

[03:14:03.0000] <zcorpan>
"aria-hidden" isn't checked for in the code

[03:14:04.0000] <hsivonen>
wow

[03:14:05.0000] <zcorpan>
aria-hidden is for firevox basically

[03:15:00.0000] <Hixie>
i don't understand why aria is so popular

[03:15:01.0000] <Philip`>
A bug / missing feature in an early implementation of a new technology? It's terrible :-(

[03:15:02.0000] <Hixie>
it's a bandaid

[03:16:00.0000] <annevk>
some accessibility people seem to love bolton stuff

[03:16:01.0000] <Philip`>
There's probably a billion-dollar industry in bandaids

[03:16:02.0000] <jgraham>
Hixie: There seems to be something of a culture of liking bandaids because they are nicely visible

[03:16:03.0000] <annevk>
Philip`, consultancy certainly benefits from it

[03:17:00.0000] <jgraham>
Whereas just fixing stuff properly doen't look like you're trying so much

[03:17:01.0000] <annevk>
jgraham, so you can pat yourself on the back?

[03:17:02.0000] <zcorpan>
for anyone interested in aria (or at least the moz ua impl guide document), there's a new googlegroup: aria-ua-impl

[03:18:00.0000] <jgraham>
annevk: Well I don' know about the authors point of view I guess but that seems to be the approach of a number of a11y types

[03:18:01.0000] <Philip`>
Hmph, I can't find any data on the size of the bandaid industry, since searches keep turning up bandaid metaphors instead

[03:18:02.0000] <annevk>
so the issue with workers-tracker is that in order to do it properly you actually want to abstract out the tracker stuff and just import it or something

[03:18:03.0000] <zcorpan>
also: http://monotonous.org/2008/08/25/lets-all-do-it-right-the-first-time/ should be nice for testing aria

[03:18:04.0000] <zcorpan>
haven't had time to look into it properly yet though

[03:19:00.0000] <jgraham>
/me should go do something useful

[03:20:00.0000] <hsivonen>
is there a simple test case that shows if IE8 is in IE7 standards mode or in IE8 mode?

[03:20:01.0000] <annevk>
"Let’s all do it right the first time" and ARIA in one post is sort of ironic :)

[03:21:00.0000] <Hixie>
in the last checkin comment, i meant DOS when i wrote DOM. before anyone points it out.

[03:21:01.0000] <annevk>
(how do you paste in a terminal?)

[03:21:02.0000] <Hixie>
annevk: :-/

[03:22:00.0000] <Hixie>
annevk: unix?

[03:22:01.0000] <Hixie>
annevk: middle click pastes selection if you're in X Windows

[03:22:02.0000] <annevk>
Ubuntu? :)

[03:22:03.0000] <annevk>
/me is such a noob

[03:23:00.0000] <hsivonen>
IIRC, Gnome Terminal has a context menu

[03:24:00.0000] <Hixie>
ew

[03:24:01.0000] <Hixie>
gnome terminal

[03:24:02.0000] <Philip`>
hsivonen: document.compatMode

[03:24:03.0000] <Hixie>
who the hell is 64.111.120.16 and why are they DOSsing me

[03:25:00.0000] <Hixie>
alameda.dreamhost.com

[03:25:01.0000] <Hixie>
is that one of you :-P

[03:25:02.0000] <zcorpan>
:roll:

[03:25:03.0000] <zcorpan>
j/k

[03:25:04.0000] <Hixie>
oh it's stopped now

[03:25:05.0000] <Philip`>
That's where html5.org is

[03:26:00.0000] <Hixie>
thought it might be :-)

[03:26:01.0000] <Philip`>
But I don't run any interesting scripts there so it's not my fault :-)

[03:26:02.0000] <Hixie>
a lot of hits to svn.whatwg.org, made the load on hixie go up to 24

[03:26:03.0000] <zcorpan>
perhaps it's the hidden feature of webapps-tracker?

[03:26:04.0000] <Hixie>
i blame anne :-P

[03:29:00.0000] <annevk>
hsivonen, it does, but using that is annoying

[03:30:00.0000] <Philip`>
/me always uses shift+insert to paste

[03:31:00.0000] <hsivonen>
/me exhausted a hard drive while storing validation messages

[03:31:01.0000] <Philip`>
/me always turns off middle-click paste when possible, because it's far too likely he'll click accidentally and send random clipboard content to random people

[03:32:00.0000] <hsivonen>
I guess I should have used shorter markers. like Q instead of QUIRKS

[03:33:00.0000] <Hixie>
heh

[03:33:01.0000] <hsivonen>
to fit more data onto the disk

[03:33:02.0000] <Hixie>
Philip`: i do the same with the clipboard in general

[03:33:03.0000] <Philip`>
If you're compressing, that shouldn't make a difference

[03:33:04.0000] <Philip`>
s/a/much/

[03:33:05.0000] <hsivonen>
or perhaps I should have omitted the file system

[03:34:00.0000] <Philip`>
Perhaps you should have done a small sample and measured how much space it used, and extrapolated that to the whole dataset, rather than doing as much as possible until you discover you've run out of space? :-)

[03:34:01.0000] <Hixie>
hah

[03:35:00.0000] <Hixie>
that's a pessimistic approach

[03:35:01.0000] <hsivonen>
Philip`: the dataset *almost* fit on the drive

[03:35:02.0000] <hsivonen>
missed last 6000 pages or so

[03:35:03.0000] <Hixie>
a better approach would have been to use a drobo, and as the capacity was reached, rushed to the store and slammed in a bigger capacity drive at the last minute :-)

[03:35:04.0000] <Philip`>
hsivonen: Did it get that close through planning, or through luck? :-)

[03:36:00.0000] <hsivonen>
Philip`: luck

[03:36:01.0000] <hsivonen>
Philip`: my planning said 60 GB

[03:36:02.0000] <hsivonen>
reality was closer to 150 GB

[03:56:00.0000] <Philip`>
Hixie: I think the text-coalescence change looks kind of reasonable (though I reserve the right to change my mind in the future)

[03:56:01.0000] <Hixie>
k

[03:56:02.0000] <Hixie>
that was pretty much my reaction

[03:57:00.0000] <annevk>
http://html5.org/tools/web-forms-tracker and http://html5.org/tools/web-workers-tracker now work

[03:57:01.0000] <Hixie>
woo

[03:57:02.0000] <Hixie>
thanks!

[03:58:00.0000] <annevk>
it was apparently quite easy to extract the functions into a trackerlib file and then invoke one function with a few arguments from the individual files

[03:58:01.0000] <Philip`>
I think it means a</li>b</li>c... will still be one text node, which is possible to optimise by being adequately clever, and a<table>b<td></td>c<td></td>d... will now be split into lots of text nodes, which is good since I don't think it was possible to optimise when you could have scripts inside the <td>s

[03:58:02.0000] <Hixie>
yeah the examples you gave in your original e-mail all work as you wanted them to definitely

[03:58:03.0000] <Hixie>
i'm just worried that there's some edge case where i have it defined in some ass-backwards way that conflicts with this

[03:59:00.0000] <annevk>
I should probably commit this back to the html5 SVN now

[03:59:01.0000] <Hixie>
is namespace-well-formed a subset of well-formed, or an orthogonal concept? hmm

[04:01:00.0000] <annevk>
done

[04:08:00.0000] <annevk>
shirt+insert works Philip`, but they are rather far apart

[04:08:01.0000] <annevk>
(on my keyboard)

[04:09:00.0000] <annevk>
prolly better than rightclicking and all though

[04:15:00.0000] <hsivonen>
Philip`: in IE8, what values for document.compatMode should I look for?

[04:17:00.0000] <Philip`>
hsivonen: Whoops

[04:17:01.0000] <Philip`>
I meant document.documentMode

[04:17:02.0000] <hsivonen>
Philip`: thanks

[04:17:03.0000] <Philip`>
which is 8 in IE8-mode

[04:17:04.0000] <Philip`>
and 5 in quirks mode

[04:18:00.0000] <hsivonen>
Philip`: thanks

[04:19:00.0000] <Philip`>
and 7 in IE7-mode

[04:19:01.0000] <Hixie>
hsivonen: i think i found a bug in the live html5 dom viewer (and in the spec)

[04:19:02.0000] <Hixie>
http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0A...%3Cdiv%3E%3Coption%3E%3C%2Fdiv%3Ex

[04:19:03.0000] <Philip`>
I don't think it can ever be 6

[04:19:04.0000] <Hixie>
<option> is not defined as phrasing or formatting

[04:19:05.0000] <Hixie>
so the </div> shouldn't see the <div>

[04:20:00.0000] <Philip`>
http://msdn.microsoft.com/en-us/library/cc196988(VS.85).aspx but don't believe where it says "In Internet Explorer 8 and later, the compatibility mode of the document is determined wholly by the X-UA-Compatible meta tag." because it's lying

[04:20:01.0000] <annevk>
it might help to rename the title of http://livedom.validator.nu/

[04:21:00.0000] <annevk>
it's now identical to the service hixie provides while the two are clearly distinct

[04:21:01.0000] <Hixie>
weird, html5lib does the same thing

[04:21:02.0000] <Hixie>
am i misreading the spec or something?

[04:21:03.0000] <Hixie>
http://parsetree.validator.nu/?parser=html5&content=%3C%21DOCTYPE+html%3E%0D%0A...%3Cdiv%3E%3Coption%3Ez%3C%2Fdiv%3Ex&submit=Print+Tree

[04:21:04.0000] <annevk>
wasn't <option> automatically closed?

[04:22:00.0000] <annevk>
see implied end tags

[04:22:01.0000] <Hixie>
why would an end tag be implied?

[04:22:02.0000] <Hixie>
it wouldn't be if you replaced the </div> with </foo>

[04:22:03.0000] <Hixie>
and you should bail before you find out that you know about a <div>

[04:22:04.0000] <Hixie>
because <option> isn't phrasing or formatting

[04:23:00.0000] <annevk>
huh?

[04:23:01.0000] <Hixie>
(which is the bug in the spec, and will be changed shortly)

[04:23:02.0000] <annevk>
when you see the end tag </div>

[04:23:03.0000] <annevk>
you generate implied end tags

[04:23:04.0000] <Hixie>
why?

[04:23:05.0000] <annevk>
because that's what the spec says

[04:23:06.0000] <Hixie>
ohhh

[04:23:07.0000] <Hixie>
my bad

[04:23:08.0000] <Hixie>
i'm being dumb

[04:23:09.0000] <hsivonen>
Hixie: so I'm implementing the spec, right? then it's not a live dom bug :-)

[04:24:00.0000] <hsivonen>
annevk: I'll change the title

[04:24:01.0000] <Hixie>
yeah

[04:24:02.0000] <Hixie>
my bad

[04:24:03.0000] <Hixie>
forgot </div> had its own entry, duh

[04:24:04.0000] <Hixie>
anyway

[04:24:05.0000] <hsivonen>
annevk: title fixed

[04:25:00.0000] <Hixie>
<optgroup> and <option> are becoming phrasing-level

[04:25:01.0000] <hsivonen>
Hixie: why?

[04:25:02.0000] <Hixie>
so that <datalist><option>x</datalist>y works

[04:26:00.0000] <Hixie>
(also <option> is getting its own entry in the "in body" state that implies </option>)

[04:26:01.0000] <Hixie>
(so that <datalist><option>x<option>y</datalist> works)

[04:26:02.0000] <hsivonen>
did the chairs decide that datalist is in?

[04:27:00.0000] <Hixie>
i've been told to go ahead and merge wf2 in, yes

[04:27:01.0000] <hsivonen>
ok

[04:27:02.0000] <Hixie>
but even if that hadn't happened, we'd still want to fix the parsing model

[04:27:03.0000] <Hixie>
so that we have the option later

[04:28:00.0000] <Philip`>
It's odd that IE8 devotes equal screen space to the "render with IE7 bug compatibility" button as it does to "refresh page"

[04:28:01.0000] <Philip`>
Prediction: People will click the compatibility button, see that it causes the page to reload, and deduce that that must be the refresh button

[04:29:00.0000] <Philip`>
and so they'll unknowingly toggle between IE7-bugs mode and IE8-bugs mode each time they refresh a page

[04:29:01.0000] <Hixie>
i guess <optgroup> should also imply </option>

[04:30:00.0000] <Hixie>
just in case we ever introduce <optgroup> to something

[04:30:01.0000] <Hixie>
like datalist

[04:34:00.0000] <Hixie>
the html5 live dom viewer rocks.

[04:40:00.0000] <annevk>
how the hell can the headers issue be resolved?

[04:41:00.0000] <hsivonen>
hmm. did the HTML WG make a decision on a telecon?

[04:41:01.0000] <hsivonen>
fwiw, I'd have resolved it by letting td[headers] to point to th and letting th[headers] to point to another th

[04:42:00.0000] <annevk>
for people who care so much about process they are doing a bad job at following the charter

[04:42:01.0000] <hsivonen>
pointing it to td seems like semantic abuse to me

[04:42:02.0000] <Philip`>
It sounded like a decision in the "agreement between some group of people" sense, not the "formal binding WG decision" sense

[04:42:03.0000] <hsivonen>
and to make a migration from bolt-on to built-in, authors should get in the habit of using <th> where appropriate

[04:44:00.0000] <annevk>
Philip`, not the way steven announced it

[04:44:01.0000] <Lachy>
Hixie, in case you missed this bug I noticed yesterday, see http://krijnhoetmer.nl/irc-logs/whatwg/20080828#l-569

[04:45:00.0000] <annevk>
fwiw: http://lists.w3.org/Archives/Public/www-archive/2008Aug/0129.html

[04:45:01.0000] <Hixie>
Lachy: i think it was just that the server side of the script failed to update the file with the latest version in it

[04:45:02.0000] <Lachy>
ok

[04:47:00.0000] <annevk>
Hixie, <optgroup> should close <optgroup>

[04:47:01.0000] <Hixie>
that would prevent us from ever nesting optgroups

[04:48:00.0000] <annevk>
<select> already prevents that

[04:48:01.0000] <Hixie>
this is for outside select

[04:48:02.0000] <annevk>
do we want them to be inconsistent? hmm

[04:48:03.0000] <Hixie>
<optgroup> is currently never allowed outside <select>

[04:48:04.0000] <Hixie>
<datalist> only accepts <option>s

[04:49:00.0000] <Hixie>
iirc

[04:49:01.0000] <Hixie>
i just want to leave the door open, just in case

[04:49:02.0000] <annevk>
okay

[04:50:00.0000] <Philip`>
annevk: Ah, I meant during the telecon - I agree his email sounds more like it was a Process-style decision

[04:56:00.0000] <annevk>
ah, Mark Birbeck replied to http://www.ibm.com/developerworks/blogs/page/JohnBoyer?entry=the_ubiquity_strategy_for_promoting

[04:57:00.0000] <annevk>
seems to me people actually want new functionality from browsers, eg Offline Web Applications, multi-core CPU usage, etc.

[05:05:00.0000] <zcorpan>
/me updates html5-elements with bb and hidden

[05:05:01.0000] <zcorpan>
is there something else that has changed or been added recently?

[05:06:00.0000] <zcorpan>
where recently is after event-source was renamed

[05:07:00.0000] <annevk>
hidden should be before id

[05:08:00.0000] <Hixie>
<action> sounds like <command>

[05:08:01.0000] <Hixie>
and <form action>

[05:08:02.0000] <annevk>
yeah, <browserbutton>

[05:08:03.0000] <zcorpan>
annevk: thanks

[05:11:00.0000] <Hixie>
hsivonen: <!DOCTYPE html PUBLIC ""> fails the "make it look worse" test to avoid cargo cultists thinking more is more

[05:11:01.0000] <annevk>
does anyone recall when I tried to avoid accepting what the HTML charter said?

[05:11:02.0000] <annevk>
was that ever the case?

[05:11:03.0000] <Hixie>
has anyone ever suggested ignoring the charter?

[05:12:00.0000] <annevk>
stevef just told me "btw It is good to see you recognising the HTML WG charter in this case."

[05:12:01.0000] <Hixie>
o_O

[05:12:02.0000] <annevk>
yeah, wtf

[05:16:00.0000] <Hixie>
i wonder when people will come to realise that when an e-mail just says "+1" or "I agree" i delete it without even seeing who wrote it

[05:16:01.0000] <Hixie>
and never actually take that e-mail into account when making the related edits

[05:16:02.0000] <annevk>
http://www.devx.com/webdev/Article/38983 cool

[05:18:00.0000] <annevk>
hmm, code doesn't do much

[05:21:00.0000] <annevk>
(moving is s-x and j-l)

[05:23:00.0000] <hsivonen>
Hixie: At least I mentioned testing browsers instead of just saying that I agree.

[05:24:00.0000] <hsivonen>
I wonder what it would take to make a compiler that takes Flex stuff and outputs SVG+JS instead of Flash

[05:25:00.0000] <Hixie>
hsivonen: your e-mail ended up in the pile of e-mails that i will consider

[05:25:01.0000] <Hixie>
because it had actual information in it

[05:25:02.0000] <Hixie>
like almost all your e-mails :-)

[05:25:03.0000] <Hixie>
btw, i don't understand what ui you would have for hierarchical headers

[05:26:00.0000] <Hixie>
just have the header cell itself read out its own headers? or propagate them to the data cells too?

[05:26:01.0000] <hsivonen>
both, I suppose

[05:26:02.0000] <Hixie>
the latter is pretty annoying

[05:27:00.0000] <Hixie>
especially for the kind of tables where you see it

[05:27:01.0000] <jgraham>
Hixie: I would like to see user tests to prove that

[05:27:02.0000] <annevk>
incidentally, the minutes for that particular discussion are a mess, with Lachy asking several questions not being answered and Chris Wilson stating that people should show up to telcons if they had wanted to advance counter arguments...

[05:27:03.0000] <jgraham>
But the a11y people hate me for saying that

[05:27:04.0000] <Hixie>
jgraham: yeah, user tests would help a lot here

[05:27:05.0000] <jgraham>
Shall I raise the process issue with SF's email?

[05:28:00.0000] <annevk>
jgraham, I raised something on www-archive

[05:28:01.0000] <annevk>
jgraham, not sure if more is required, but maybe

[05:28:02.0000] <jgraham>
annevk: OK, well I will do it then

[05:32:00.0000] <Hixie>
svn blame on the html5 spec takes an ungodly amount of time

[05:32:01.0000] <hsivonen>
even though you are to blame for each check-in

[05:32:02.0000] <Dashiva>
Doesn't every line say hixie?

[05:33:00.0000] <Hixie>
Dashiva: yeah, trying to pin down which rev i added <rt> and <rp> to the "optional tags" section since i have a note here saying i have to do something with them, but i can't tell if the something is to add them (and the note predates the add) or to fix them (and the note postdates the add)

[05:33:01.0000] <Dashiva>
How about a binary reach instead? :)

[05:33:02.0000] <Dashiva>
*search

[05:33:03.0000] <Lachy>
annevk, the minutes for telcons are always a mess and freqently omit information

[05:34:00.0000] <Hixie>
Dashiva: would prolly have been quicker

[05:35:00.0000] <Hixie>
Lachy: that's one reason i avoid telecons

[05:35:01.0000] <Lachy>
I would have avoided it, but annevk asked me to call in for him, so I did

[05:35:02.0000] <Hixie>
heh

[05:35:03.0000] <Hixie>
why would he ask you something like that

[05:35:04.0000] <Hixie>
that sounds like work!

[05:36:00.0000] <Dashiva>
Telcons sound kinda like the spec equivalent of 5-second soundbites

[05:37:00.0000] <Lachy>
Hixie, it's fair enough. Since I don't do any other work, 30 minutes on a telcon isn't too bad :-)

[05:37:01.0000] <Hixie>
hah

[05:38:00.0000] <Lachy>
Hixie, what can I say to convince you that you're wrong about not allowing <area alt="..."> without href=""? http://lists.w3.org/Archives/Public/public-html/2008Aug/0860.html

[05:40:00.0000] <Hixie>
you can show how that text would be used

[05:40:01.0000] <Lachy>
since you rejected the valid use case I presented, finding more evidence means more work.

[05:40:02.0000] <Hixie>
right now the spec says it must be ignored

[05:40:03.0000] <annevk>
Hixie, web-apps-tracker takes a limit parameter which you can set to -1 or 1000 or something

[05:40:04.0000] <Hixie>
and we generally want conformance checkers to report things that aren't useful

[05:40:05.0000] <annevk>
Hixie, that's probably easier than using svn blame

[05:40:06.0000] <Hixie>
annevk: quicker, certainly

[05:40:07.0000] <Hixie>
annevk: easier, no :-)

[05:40:08.0000] <annevk>
ok

[05:40:09.0000] <Hixie>
annevk: if blame was quick, i could just look up the number and be done with it :-)

[05:41:00.0000] <hsivonen>
Hixie: I hope you missed a "don't" in the statement about conformance checkers above

[05:42:00.0000] <Hixie>
i mean, report things in the markup that aren't going to do anything

[05:42:01.0000] <Hixie>
since they are likely an indication of an error

[05:42:02.0000] <Hixie>
since authors don't generally go out of their way to do something that isn't goign to do anything

[05:42:03.0000] <hsivonen>
actually, they do it all the time

[05:42:04.0000] <hsivonen>
<style type=text/css>

[05:42:05.0000] <Hixie>
let me rephrase

[05:43:00.0000] <hsivonen>
type=text/css is a no-op

[05:43:01.0000] <Hixie>
authors don't intentionally and knowingly do it

[05:43:02.0000] <Lachy>
Hixie, using <area> to annotate regions of images with additional info, even if they aren't links, seems like a perfectly reasonable thing to do. I guess I will have to look at some more real world examples of image maps

[05:44:00.0000] <Hixie>
Lachy: an image map is a list of links. using it for something else is an abuse of the semantics.

[05:44:01.0000] <Hixie>
Lachy: now if we want to handle those other use cases, maybe we should repurpose <area> in some way, e.g. with some element other than <map>

[05:44:02.0000] <Hixie>
Lachy: or with some new ua-level requirements

[05:44:03.0000] <Hixie>
Lachy: but we'd have to justify doing so

[05:55:00.0000] <Lachy>
Hixie, HTML4 defined an image map as: "Image maps allow authors to specify regions of an image or object and assign a specific action to each region"

[05:55:01.0000] <Lachy>
The action could be an event handler like onclick instead of just a link

[05:56:00.0000] <Hixie>
then alt="" would be allowed with onclick or href=""

[05:56:01.0000] <Hixie>
we could do that i guess

[05:56:02.0000] <Hixie>
seems a bit weird

[05:57:00.0000] <Hixie>
we don't make <a onclick=""> into a link

[05:58:00.0000] <Lachy>
the event handler could be attached dynamically, so there wouldn't need to be onclick in the markup

[05:58:01.0000] <hsivonen>
'null' is null or not an object

[05:59:00.0000] <hsivonen>
says IE8

[05:59:01.0000] <hsivonen>
if I had a stringified 'null', it would be an object, wouldn't it?

[06:00:00.0000] <Lachy>
hsivonen, for what code does it say that?

[06:00:01.0000] <Dashiva>
It would act as one, at least

[06:00:02.0000] <hsivonen>
Lachy: the script on validator.nu

[06:01:00.0000] <Lachy>
grrr. Why does IE8 make the error dialog modal?!

[06:01:01.0000] <Dashiva>
hsivonen: You're living pretty dangerously with all those implied semicolons

[06:02:00.0000] <hsivonen>
Dashiva: what's a language feature good for if living with it is dangerous?

[06:03:00.0000] <annevk>
living on the edge can give you a kick?

[06:04:00.0000] <Dashiva>
One could imagine JS was made for small scripts with small chances of breaking horribly :)

[06:04:01.0000] <Lachy>
I never remember when it's ok to omit semicolons, and it always seems to complain when I do. So I do it anyway

[06:04:02.0000] <Lachy>
I mean, I always use them

[06:05:00.0000] <Lachy>
it seems to be complaining about this line: newStyle.styleSheet.cssText = rule

[06:05:01.0000] <Dashiva>
Lachy: I'm pretty sure you're leaving some out, you just don't know it :)

[06:05:02.0000] <Lachy>
so it means either newStyle or newStyle.styleSheet is returning null

[06:07:00.0000] <Lachy>
Dashiva, if I do, it's only by accident; But I do always try to end every line with a semicolon;

[06:07:01.0000] <hsivonen>
boohoo. the data I recorded is corrupt. It's corrupt in such a crazy way that I suspect gzipping gone wrong at some point

[06:08:00.0000] <Philip`>
/me wonders if "some point" was 4GB in

[06:09:00.0000] <annevk>
heh, Opera shipped with APNG support before Firefox; I did not know

[06:09:01.0000] <hsivonen>
let's find out

[06:09:02.0000] <Philip`>
annevk: Only by a few days :-p

[06:10:00.0000] <Philip`>
and since there are approximately zero APNG images in the world, it doesn't matter much in practice

[06:12:00.0000] <Philip`>
Opera renders some perfectly valid APNGs incorrectly on some platforms anyway

[06:13:00.0000] <Philip`>
whereas Firefox seems to work fine, except in some obscure cases like 8-bit greyscale or 1-bit APNGs

[06:13:01.0000] <Hixie>
wow, IE8 finally adds the inline find toolbar, and they manage to make the one mistake that every other browser has agonised over finding a way to avoid

[06:13:02.0000] <Hixie>
they move the whole content down when you hit find

[06:13:03.0000] <Hixie>
ok with that, bed time

[06:13:04.0000] <Hixie>
nn

[06:16:00.0000] <hsivonen>
Philip`: is there a known problem with GZIPOutputStream with stream longer than 4 GB?

[06:18:00.0000] <Philip`>
hsivonen: None that I know of; I'm just assuming it's the most obvious point of failure when dealing with unusually large files

[06:18:01.0000] <hsivonen>
quick googling shows problems with .zip files over 4GB and java class lib

[06:18:02.0000] <hsivonen>
nothing about gzip, though

[06:20:00.0000] <Philip`>
It's also possible that the compression is fine but the decompression isn't working

[06:20:01.0000] <hsivonen>
good point

[06:22:00.0000] <hsivonen>
could be the python side http://bugs.python.org/issue1672853

[06:23:00.0000] <hsivonen>
one would think that these were solved problems by now

[06:23:01.0000] <hsivonen>
all handled in the infrastructure

[06:25:00.0000] <hsivonen>
interestingly, I can't use zless to traverse the whole file. it complains about mmap failing at some point

[06:25:01.0000] <Philip`>
Use 'gzip -cd | less' ?

[06:26:00.0000] <hsivonen>
isn't that what zless is?

[06:26:01.0000] <Philip`>
I would assume zless has to do effectively that anyway, since I didn't think you could decompress zlib without starting from the beginning

[06:27:00.0000] <Philip`>
I would assume zless is a program that does some stuff itself, so it'd have different bugs to gzip/less

[06:28:00.0000] <hsivonen>
Philip`: zless is a shell script that runs gzip and less

[06:28:01.0000] <Philip`>
Oh, okay

[06:29:00.0000] <hsivonen>
anyway, I made a copy of the first 4GB with dd. now trying if that part makes Python fail

[06:31:00.0000] <hsivonen>
how can this be so hard?

[06:33:00.0000] <Philip`>
Alternatively: How can it be so easy for you to manipulate a trillion bits of data in a few hours from the comfort of your chair? :-)

[06:34:00.0000] <Philip`>
It's not the kind of thing you could contemplate doing five or ten years ago

[06:34:01.0000] <hsivonen>
well, the failure happens before 4 GB of compressed data

[06:34:02.0000] <hsivonen>
dunno what the uncompressed counter was

[06:37:00.0000] <Philip`>
Easy to find out - just do 32 bisections :-)

[06:48:00.0000] <annevk>
hmm, I need to fix cookies for the various trackers

[06:55:00.0000] <annevk>
fixed

[06:59:00.0000] <hsivonen>
the decompressed data is at 5.9 GB when the failure happens

[06:59:01.0000] <hsivonen>
weird

[07:03:00.0000] <Philip`>
Does anyone happen to have a copy of IE8b2's mshtml.dll they could send me? (I want to look at some stuff in it, but left my copy at home)

[07:21:00.0000] <Philip`>
/me sees that http://blogs.msdn.com/ie/archive/2008/08/27/internet-explorer-8-beta-2-now-available.aspx has at least three requests for <canvas> support, and can't see any for any other HTML5 features

[07:30:00.0000] <Philip`>
gsnedders: Are you aware that http://google.com/search?q=cache:simplepie.org/ is still full of spam?

[07:32:00.0000] <Philip`>
(and now it has a "Buy Phentermine" link - when I last looked, I think all these hacked pages only had links to each other, so maybe that was just in preparation for inserting the actual target link)

[07:33:00.0000] <virtuelv>
speaking of spam, is anyone able to get hold of Blake Ross? Turn off JS and look at http://www.blakeross.com/

[07:39:00.0000] <Philip`>
shepazu: http://google.com/search?q=cache:http://schepers.cc still has the same spam too

[07:40:00.0000] <shepazu>
Philip`: what should I do about this?

[07:40:01.0000] <Philip`>
shepazu: I have no idea

[07:40:02.0000] <shepazu>
the real page seems fine...

[07:41:00.0000] <Philip`>
Live Search and Yahoo have the same spam in their cached versions

[07:41:01.0000] <shepazu>
/me decides not to care :)

[07:41:02.0000] <Philip`>
(I assume it's only sending spam to certain IPs or something)

[07:41:03.0000] <annevk>
check  the source code of the PHP pages?

[07:42:00.0000] <Philip`>
shepazu: I suppose that's the easiest solution :-p

[07:43:00.0000] <svl>
upgrade to the latest wordpress version? (which is 2.6.1 rather than 2.5.1)

[07:44:00.0000] <svl>
/me assumes there's a vulnerability in 2.5.1 which is being exploited to have the spam inserted.

[07:44:01.0000] <shepazu>
bbiab

[07:46:00.0000] <Philip`>
http://philip.html5.org/misc/spammy-sites.txt is my notes from when I last looked at this - it seemed to be specific to Dreamhost

[07:52:00.0000] <csarven>
Is krijnh human or bot?

[07:52:01.0000] <svl>
human

[07:53:00.0000] <svl>
Though not able to access IRC atm.

[07:54:00.0000] <csarven>
Do you know if I /msg him (or her), they'll be able to check their logs when they return?

[07:55:00.0000] <svl>
He will be, yes. And he's also reading log files, so even saying it here in the channel should reach him. *waves to krijn* ;)

[07:56:00.0000] <csarven>
Oh alright then. In that case, I've already done that yesterday on #microformats :) Thanks. Figured he was more of a regular here.

[07:58:00.0000] <svl>
Not quite certain if he'd have read it in there, but now that the pointer exists here, he should.

[07:59:00.0000] <csarven>
krijnh re: http://krijnhoetmer.nl/irc-logs/microformats/20080829#l-66 :)

[08:01:00.0000] <annevk>
_a should probably be data-author or something

[08:01:01.0000] <annevk>
and <abbr> should be <time>

[08:02:00.0000] <annevk>
though it seems like a lot of extra bandwidth

[08:02:01.0000] <csarven>
_a was already in place. I wanted to add minimal changes and add hCard and hAtom

[08:02:02.0000] <csarven>
I don't know why there is _a and if it is required.

[08:03:00.0000] <Philip`>
It's there so you can make your own lines get highlighted, I think

[08:03:01.0000] <csarven>
You mean flagging?

[08:03:02.0000] <Philip`>
(via the "who are you?" box in http://krijnhoetmer.nl/irc-logs/ )

[08:04:00.0000] <csarven>
That's what the original <span> </span> is supposed to do

[08:04:01.0000] <Philip`>
Not the flagging - you can do client-side nick-based styling

[09:00:00.0000] <Lachy>
http://www.theregister.co.uk/2008/08/29/hakon_lie_ie8_interoperability/

[09:03:00.0000] <Lachy>
since a large majority of intranet pages are built for IE only, it's understandable would make IE7 mode the default for intranets. Though, it is sad because companies that make intranet pages could still continue targetting IE7 mode

[09:06:00.0000] <annevk>
they would, unless they use some MS-specific workaround

[09:09:00.0000] <Philip`>
hsivonen: Is there a way I can use something like getElementById on the Document returned by your parser, that doesn't appear to just return null all the time?

[09:11:00.0000] <hsivonen>
Philip`: are you using the DOM in Java?

[09:12:00.0000] <Philip`>
hsivonen: Yes

[09:12:01.0000] <hsivonen>
Philip`: does DOM have a public API for setting IDness?

[09:14:00.0000] <Philip`>
hsivonen: Uh...

[09:15:00.0000] <hsivonen>
Philip`: meanwhile, you can grab this code http://pastebin.ca/1188548

[09:15:01.0000] <hsivonen>
ooh. Element.setIdAttribute()

[09:16:00.0000] <Philip`>
It says "The DOM implementation is expected to use the attribute Attr.isId to determine if an attribute is of type ID."

[09:17:00.0000] <Philip`>
hsivonen: Aha, that code is handy and seems to work - thanks!

[09:32:00.0000] <annevk>
hsivonen, we want to remove that crap

[09:32:01.0000] <annevk>
(from the DOM)

[09:33:00.0000] <annevk>
(though that shouldn't affect Java I suppose)

[09:34:00.0000] <hsivonen>
Philip`: fixed on trunk

[09:34:01.0000] <hsivonen>
Philip`: thanks

[09:37:00.0000] <Philip`>
/me sees 3 pages that have @headers pointing at td, and 18 pointing at th

[09:37:01.0000] <Philip`>
and 16 that have totally bogus @headers

[09:38:00.0000] <Philip`>
(headers="24", headers="Symbol information: ", headers="100%", etc)

[09:39:00.0000] <Philip`>
Only one has <th headers>, vs 35 with <td headers>

[09:39:01.0000] <Philip`>
(out of ~130K pages)

[09:58:00.0000] <gsnedders>
Philip`: I see no spam there

[09:59:00.0000] <Philip`>
gsnedders: Oh

[09:59:01.0000] <Philip`>
gsnedders: I do

[10:00:00.0000] <gsnedders>
I do in the text only version, actally

[10:00:01.0000] <Philip`>
e.g. it says "online pharmacy" just before the "Newer. Better. Faster." button

[10:00:02.0000] <Philip`>
and then it says "I must be only to be vizier. "Holla, you can't unclog me", winged the excise. "I central nervous system, spongy degeneration of the delicately nib", he recognized."

[10:00:03.0000] <gsnedders>
Philip`: How are you finding this? :P

[10:01:00.0000] <Philip`>
gsnedders: I'm looking at the Google cache page, and then reading it with my eyes

[10:01:01.0000] <gsnedders>
Philip`: Yes, but why are you looking at the cache page anyway? :P

[10:01:02.0000] <Philip`>
gsnedders: Because I was wondering whether it had been fixed in the several months since you first noticed the problem

[10:01:03.0000] <gsnedders>
Philip`: It was

[10:02:00.0000] <Philip`>
Oh, okay

[10:02:01.0000] <Philip`>
If so, it broke again :-p

[10:02:02.0000] <Philip`>
(How was it fixed?)

[10:02:03.0000] <gsnedders>
/me emails the guy from Yahoo! Search (who originally found it) and Ryan Parman (who deals with the website mainly) about it again

[10:03:00.0000] <gsnedders>
"I still don't know how they got in, but the guys at Dreamhost found an obscure file that was being added through a couple layers of includes. In our case, we were using the PHP4-compatible nusoap class for our search results (accessing Google's SOAP API that isn't even supported anymore, but I hadn't updated the code yet).

[10:03:01.0000] <gsnedders>
The offending code has been removed and the search box now searches Google directly.

[10:03:02.0000] <gsnedders>
"

[10:04:00.0000] <Philip`>
If the guys at Dreamhost care, they could try following all the spam links on that page, since I expect they're all pointing to similarly-hacked Dreamhost sites

[10:06:00.0000] <krijnh>
*waves back at svl*

[10:06:01.0000] <krijnh>
csarven: I'll take a look at it, gotta cook some diner now :) (As you see, I'm not a bot)

[10:07:00.0000] <gsnedders>
Philip`: Oh, Ryan's email went on to say that Dreamhost basically didn't care

[10:07:01.0000] <Philip`>
krijnh: You might just be a particularly clever bot

[10:08:00.0000] <svl>
Round 1 of the Turing Test: *ding*

[10:08:01.0000] <krijnh>
Philip`: Thank you! In a way :)

[10:10:00.0000] <Philip`>
krijnh: It wasn't necessarily a compliment, given how dumb most existing bots are ;-)

[10:11:00.0000] <krijnh>
Hence the "in a way"

[10:14:00.0000] <csarven>
krijnh Awesome :)

[10:21:00.0000] <gsnedders>
Hixie: I'd probably rather you didn't use a web service for the same reason you don't want to run it yourself: the load :P

[10:22:00.0000] <Philip`>
But web services don't cause any more load than non-web services, and the service has to be available via some mechanism, so why not via the web?

[10:22:01.0000] <gsnedders>
Oh, sure, it will be. But I'm not sure I want my website to have to cope with the load of Hixie generating HTML 5 :P

[10:25:00.0000] <krijnh>
csarven: any particular reason why I should double my bandwidth with those fluffy microformats? :)

[10:26:00.0000] <Philip`>
krijnh: You have compression, so it wouldn't double :-)

[10:27:00.0000] <krijnh>
http://www.csarven.ca/labs/microformats/krijnhoetme.nl_irc-logs/code-change.txt

[10:27:01.0000] <krijnh>
Anybody else think this is useful?

[10:27:02.0000] <krijnh>
(It would mean some extra overhead in parsing each line, to get a timestamp)

[10:28:00.0000] <Philip`>
You should at least s/_a/data-author/ to be conforming HTML5 :-)

[10:28:01.0000] <krijnh>
But I want it to be accessible! :)

[10:29:00.0000] <krijnh>
I've read on the internets that HTML5 doesn't care about accessibility

[10:29:01.0000] <krijnh>
*changes doctype :X*

[10:30:00.0000] <Philip`>
/me doesn't personally have a use for any of the proposed changes

[10:33:00.0000] <krijnh>
csarven: do you?

[10:34:00.0000] <gsnedders>
/me realizes it's basically the weekend now

[10:35:00.0000] <gsnedders>
/me is slow

[10:37:00.0000] <Philip`>
jgraham: http://james.html5.org/cgi-bin/tables/table_inspector.py?input_type=type_uri&uri=http%3A%2F%2Fquickfacts.census.gov%2Fqfd%2Fstates%2F34%2F34003.html&source=&algorithm=html5 :-(

[10:39:00.0000] <csarven>
krijnh From #laconica: [10:18:10] <csarven> So, what I was thinking last night with hAtom->Atom->XMPP was to have IRC logs available as notices

[10:39:01.0000] <csarven>
People can also subscribe to the logs (for whatever reason)

[10:39:02.0000] <krijnh>
Would it be okay if I only made the fluffy format available for #microformats ?

[10:40:00.0000] <krijnh>
As a test :)

[10:40:01.0000] <csarven>
Of course, it is your call at the end :)

[10:40:02.0000] <krijnh>
Yeah, I'm the evil IRC logs overlord here :)

[10:40:03.0000] <Philip`>
They're not evil IRC logs

[10:41:00.0000] <csarven>
SO, the creator is evil but no the product

[10:41:01.0000] <krijnh>
Okay, you lost me :)

[10:41:02.0000] <Philip`>
By "subscribe to logs", you mean "put greatly increased load on the logs server"? :-)

[10:41:03.0000] <csarven>
:S

[10:42:00.0000] <csarven>
This is experimental. I highly doubt it is be impacted by load

[10:42:01.0000] <krijnh>
Okay, I'll fix it for #microformats, as an experiment

[10:42:02.0000] <csarven>
s/it is be/it would be

[10:42:03.0000] <krijnh>
Just like the [off] thing in #webapps :) (which was a FAIL btw)

[10:43:00.0000] <csarven>
Thanks.

[10:44:00.0000] <krijnh>
When I've got some time for it (probably after September 13th)

[10:44:01.0000] <csarven>
Cool.

[10:44:02.0000] <csarven>
I'll bug you on September 14 ;)

[10:45:00.0000] <csarven>
/me goes to hunt lunch

[10:45:01.0000] <krijnh>
csarven: Oki :)

[10:48:00.0000] <annevk>
/me will appear on http://openwebpodcast.com/ later today / tomorrow

[10:49:00.0000] <gsnedders>
Lachy: The prophesy in /topic will likely be fulfilled tomorrow

[10:50:00.0000] <gsnedders>
Philip`: more info from guy at Yahoo! Search: "BTW: when Wustl.edu got hit with this, they said that it was a modrewrite in Apache that was doing all this."

[11:01:00.0000] <hsivonen>
sigh. it seems that my data is corrupt.

[11:02:00.0000] <hsivonen>
at least Python's gzip and Java's gzip see the same problem

[11:02:01.0000] <hsivonen>
and zlib itself is supposed to get large files right

[11:03:00.0000] <Philip`>
/me wonders what's the probability of getting a 1-bit error on a full hard disk

[11:04:00.0000] <annevk>
geez, this RDF thread sure creates a lot of e-mail

[11:05:00.0000] <annevk>
the last six months we had around 300 messages, now it's almost 600

[11:05:01.0000] <annevk>
(300 messages on average)

[11:14:00.0000] <annevk>
http://lists.w3.org/Archives/Public/www-archive/2008Aug/0145.html o_O

[11:27:00.0000] <Philip`>
http://factfinder.census.gov/servlet/QTTable?_bm=n&_lang=en&qr_name=DEC_2000_SF1_U_DP1&ds_name=DEC_2000_SF1_U&geo_id=05000US48487 has interestingly nested headers

[11:37:00.0000] <annevk>
http://juicystudio.com/wcag/tables/complexdatatable.html uses <td class=header> uh?

[11:39:00.0000] <Philip`>
The cells contain data, so they should be <td>

[11:40:00.0000] <Philip`>
where you could define "data" as being "any values that get returned by the SQL query I'm using to generate this table" (as opposed to the headers which are strings that are typed in somewhere else)

[11:41:00.0000] <sicking>
annevk, does the XHR spec say what to do if the supplied username/password fails?

[11:41:01.0000] <sicking>
annevk, appears we put up a dialog asking the user, other browsers don't

[11:48:00.0000] <annevk>
sicking, it does

[11:49:00.0000] <annevk>
sicking, http://dev.w3.org/2006/webapi/XMLHttpRequest/#send "If authentication fails, and stored user and stored password are not provided, user agents should prompt the users for credentials."

[11:51:00.0000] <sicking>
annevk, cool

[11:52:00.0000] <sicking>
annevk, though

[11:53:00.0000] <sicking>
annevk, that doesn't really say what to do if the script provides a username/password

[11:54:00.0000] <annevk>
sicking, I could make it more explicit, the intent is that you don't do any UI interaction

[11:55:00.0000] <annevk>
(the idea is that the spec states what you need to do and everything else is "forbidden")

[11:55:01.0000] <sicking>
annevk, "user agents should prompt the users for credentials" sounds like UI interaction, no?

[11:55:02.0000] <annevk>
well yes, in that case everyone does UI interaction so it made sense to keep that

[12:04:00.0000] <sicking>
annevk, ah, so the idea is if the script provides credentials, use those, if they fail just fail. If the script doesn't provide, ask the user if needed?

[12:06:00.0000] <annevk>
sicking, yes

[12:06:01.0000] <sicking>
annevk, makes sense

[12:11:00.0000] <annevk>
good :)

[13:58:00.0000] <Hixie>
gsnedders: the main reason i like using web services is it means i can kick off several jobs without having to worry about the load they put on the machine, since they're all on different machines

[13:59:00.0000] <gsnedders>
Hixie: On the whole I expect it'll be all right… We'll see how it goes :)

[14:14:00.0000] <annevk>
did anyone put up a web service yet?

[14:14:01.0000] <annevk>
otherwise I might give it a try

[14:14:02.0000] <gsnedders>
No

[14:18:00.0000] <gsnedders>
annevk: As I said before, I can put it up on my website

[14:40:00.0000] <gsnedders>
/me wonders whether to bother to support py2.3

[14:41:00.0000] <BenMillard>
gsnedders, shall I make the hotel booking then?

[14:41:01.0000] <gsnedders>
BenMillard: If you want to. :)

[14:41:02.0000] <BenMillard>
gsnedders, it's not something which excites me but I guess I should do some part of the logistics. :P

[14:42:00.0000] <gsnedders>
/me tries to understand that column beside the form

[14:42:01.0000] <BenMillard>
gsnedders, my parents are taking my trip as an excuse to visit Cannes, so much of my travel arrangements will be identical to theirs

[14:42:02.0000] <gsnedders>
Heh.

[14:43:00.0000] <gsnedders>
The booking has to be confirmed by phone or by fax

[14:43:01.0000] <gsnedders>
/me looks at the French, and instantly it makes more sense

[14:45:00.0000] <Hixie>
i sent an e-mail and they confirmed it no problem

[14:45:01.0000] <Hixie>
still have to get my flights though

[14:45:02.0000] <gsnedders>
/me goes downstairs to con a card out of his parents to buy train tickets

[14:46:00.0000] <gsnedders>
And I'm gonna have to find somewhere to stay in Cannes for two nights, unless I go straight from my uncle's to the TOAC

[14:46:01.0000] <gsnedders>
*TPAC

[14:46:02.0000] <gsnedders>
Which on the whole I wasn't going to do

[14:47:00.0000] <gsnedders>
Hmm…

[14:48:00.0000] <gsnedders>
Maybe I should stay in Antibes…

[14:50:00.0000] <gsnedders>
Trains going from Lyon south are remarkably expensive compared with those further north

[14:58:00.0000] <annevk>
gsnedders, if you can do it that's fine too

[14:59:00.0000] <Hixie>
opening a new tab in IE8 takes takes an ungodly amount of time

[15:00:00.0000] <Hixie>
hitting ctrl+f in IE8 on the html5 spec is a bad idea

[15:00:01.0000] <annevk>
fwiw: http://openwebpodcast.com/episode-3-anne-van-kesteren-on-css-xhr-and-other-web-standards

[15:02:00.0000] <Hixie>
wow this find in page sucks ass on the spec

[15:06:00.0000] <Hixie>
aah, you mention me

[15:06:01.0000] <Hixie>
/me hides

[15:06:02.0000] <Hixie>
aaah, you did it again!

[15:11:00.0000] <Hixie>
hahaha

[15:11:01.0000] <Hixie>
that was funny

[15:11:02.0000] <Hixie>
about acid3

[15:11:03.0000] <Hixie>
"so ian can add anything to acid3, is that a bad thing?" "well actually it was the bits that other people added that were the problem..."

[15:12:00.0000] <Hixie>
(personally i think the range/traversal stuff i added to acid3 was a bigger problem than the smil stuff)

[15:34:00.0000] <Lachy>
/me disobeyed Hixe's instructions and pressed Ctrl+F on the spec in IE8... It's *really* not a good idea!

[15:34:01.0000] <Hixie>
BenMillard: your e-mail looks fine to me (indeed i'm saving it because it's got so much useful data and logical arguments!)

[15:34:02.0000] <jruderman>
Lachy: what happens?

[15:35:00.0000] <BenMillard>
Hixie, yay. :)

[15:35:01.0000] <Lachy>
jruderman, it takes ages to find anything

[15:35:02.0000] <Lachy>
become unresponsive

[15:37:00.0000] <Hixie>
fwiw my plan with this tables stuff is to do exactly what i was going to do anyway, which is to get to it when i get to it, and when i get to it, read all the arguments and research, and come to a conclusion based on that

[15:37:01.0000] <BenMillard>
Lachy, I've just forwarded the message Hixie is talking about to you, since you were at that telecon

[15:38:00.0000] <Hixie>
and unless some browser or AT vendor contacts me to say that this is urgent for them, i don't plan on getting to it any time soon

[15:38:01.0000] <BenMillard>
Hixie, yeah that's what I expect. You said after the changes earlier this year it'd be a while before returning to it, which works out well for me since I get time to do more research.

[15:38:02.0000] <Hixie>
i'm amused that chris wilson's e-mail suggests that we're not making progress

[15:38:03.0000] <Hixie>
and that the table headers thing is important for making progress

[15:39:00.0000] <Hixie>
especially given that the whole point of the table headers proposal, as i understand it, is to make it possible to mark up pages that work in existing ATs

[15:39:01.0000] <Hixie>
in a way that HTML4 still allows

[15:39:02.0000] <Hixie>
and is thus pretty much the exact opposite of "required for progress"

[15:40:00.0000] <BenMillard>
for the records, we're talking about this Public-HTML thread: http://lists.w3.org/Archives/Public/public-html/2008Aug/thread.html#msg518

[15:43:00.0000] <annevk>
I'm confused by CWs e-mail. Not sure whether to reply

[15:43:01.0000] <BenMillard>
(I'm gathering some peer review of my message before adding it to that topic.)

[15:43:02.0000] <annevk>
ah, I see Ben will reply, good :)

[15:46:00.0000] <annevk>
BenMillard, skimmed/read through your e-mail, looks good to me

[15:49:00.0000] <BenMillard>
annevk, cheers

[15:50:00.0000] <BenMillard>
I don't anticipate major changes unless reviewers find mistakes (there are a sprinkling of typos I'll address, though)

[15:55:00.0000] <Hixie>
oh great, anne's told everyone that the real reason for websocket is my model railway :-P

[15:57:00.0000] <annevk>
anecdotical evidence to feed the flames :p

[15:58:00.0000] <Lachy>
BenMillard, ok. reading it now

[15:58:01.0000] <Hixie>
annevk: good interview

[15:59:00.0000] <annevk>
ta

[15:59:01.0000] <annevk>
gsnedders, so are you putting up a Web service tonight?

[15:59:02.0000] <BenMillard>
smedero has sent me a review of the message. it's enlightening but I think it doesn't change what my message says

[15:59:03.0000] <smedero>
no, it shouldn't

[16:00:00.0000] <smedero>
it is just backstory, since you probably haven't been following the telecons....

[16:00:01.0000] <BenMillard>
yes, thanks for providing me with that

[16:07:00.0000] <jacobolus>
are unescaped ampersands allowed in html?

[16:08:00.0000] <annevk>
sometimes

[16:10:00.0000] <jacobolus>
the validator seems happy to accept “<!DOCTYPE html><title></title>&”

[16:11:00.0000] <annevk>
that's correct (not sure if the syntax section accurately reflects it currently)

[16:12:00.0000] <jacobolus>
I didn't look super carefully, but I only noticed a mention w.r.t. xml

[16:12:01.0000] <annevk>
jacobolus, it does not need to be escaped when followed by a space, EOF, another ampersand, start tag, end tag, comment

[16:12:02.0000] <jacobolus>
ah, okay

[16:13:00.0000] <jacobolus>
“<!DOCTYPE html><title></title>AT&T” properly fails then

[16:13:01.0000] <annevk>
yup

[16:14:00.0000] <annevk>
hsivonen, "Probable cause: & should have been escaped as &amp;D." what is the "D" doing there?

[16:15:00.0000] <BenMillard>
annevk, nothing much. :P

[16:16:00.0000] <jacobolus>
ah, it is described in detail, but should maybe be more explicit what authors should do?

[16:16:01.0000] <jacobolus>
(i.e. what implementors should do is described in detail)

[16:17:00.0000] <annevk>
section 8.1 describes in detail what authors should do

[16:17:01.0000] <jacobolus>
oh, nevermind

[16:17:02.0000] <annevk>
though it seems it's not entirely clear on things

[16:17:03.0000] <jacobolus>
the text must not contain the character U+003C LESS-THAN SIGN (<) or an ambiguous ampersand. 

[16:18:00.0000] <jacobolus>
okay, that's reasonable :)

[16:18:01.0000] <jacobolus>
http://www.whatwg.org/specs/web-apps/current-work/#ambiguous doesn't mention EOF :)

[16:19:00.0000] <BenMillard>
Lachy, are you double-checking my numbers? if so, thanks!

[16:20:00.0000] <annevk>
jacobolus, yeah, whatwg⊙wo ;)

[16:21:00.0000] <Lachy>
BenMillard, I suppose I can go through it again and do that. But let me finish it at least once first... I'm in the middle of doing some other things too

[16:21:01.0000] <BenMillard>
Lachy, sure thing...I don't wish to monopolise your time

[16:25:00.0000] <Lachy>
BenMillard, btw, I found a rather interesting table might be interesting for you to study. But it requires membership to see it on this site http://www.newzbin.com/

[16:26:00.0000] <BenMillard>
Lachy, would it be permitted for you to extract the table and send it to me for analysis and publishing in my collection? my work is non-commercial research so I imagine that would qualify as "fair use"

[16:26:01.0000] <annevk>
/me summons zcorpan

[16:27:00.0000] <BenMillard>
/me hopes zcorpan will arrive with a witty /me line...

[16:27:01.0000] <Lachy>
BenMillard, sure. I'll do it after I'm done with your email

[16:28:00.0000] <BenMillard>
Lachy, cool. I'll go start some dinner :)

[16:31:00.0000] <gsnedders>
annevk: As I said, if anyone writes it :P

[16:32:00.0000] <gsnedders>
BenMillard: Peh! Silly logn emails you want me to read!

[16:32:01.0000] <gsnedders>
*long

[16:34:00.0000] <Lachy>
gsnedders, you could just pretend to read it, skim it and post random comments about things you see to make him think you're really reading it. ;-)

[16:34:01.0000] <gsnedders>
Lachy: True.

[16:34:02.0000] <gsnedders>
/me hopes all this crazy insane ECMAScript optimization work shows through in other interpreted languages

[16:37:00.0000] <annevk>
gsnedders, oh I see

[16:37:01.0000] <annevk>
well, now it's too late

[16:37:02.0000] <annevk>
I guess tomorrow I can download the relevant packages and try making it work, shouldn't be too hard

[16:37:03.0000] <gsnedders>
annevk: It's probably worth working from what I'll put in hg tomorrow

[16:38:00.0000] <gsnedders>
Actually, if you want to work on it in the morning, I could do that now

[16:40:00.0000] <jgraham>
/me wonders if there is any point in arguing process with CW

[16:40:01.0000] <BenMillard>
gsnedders, there's no obligation and no salesperson will call. :D

[16:41:00.0000] <jgraham>
I think I'll fix html5lib first

[16:41:01.0000] <gsnedders>
Hixie: PostScript is a turing-complete language

[16:41:02.0000] <annevk>
gsnedders, that doesn't mean it will cause a risk

[16:41:03.0000] <gsnedders>
/me doesn't know enough to know that

[16:41:04.0000] <jgraham>
/me saw a ~3 line postscript file that did a ray tracing image of a ball on a chessboard or something once

[16:41:05.0000] <gsnedders>
I'm just pointing out what I know :P

[16:41:06.0000] <Philip`>
Turing machines with no IO devices are a bit rubbish in practice

[16:42:00.0000] <gsnedders>
/me passes Philip` an IO device

[16:42:01.0000] <Philip`>
I suppose you could try to make them jump backwards and forwards really fast so their tape catches fire

[16:44:00.0000] <gsnedders>
Philip`: Problem with that is next to no-one actually implements a turing-machine using tape

[16:45:00.0000] <Philip`>
Partly because it'd be physically impossible to implement one at all

[16:45:01.0000] <Lachy>
BenMillard, your byte counts for the sizes of the tables seem to be a little off

[16:45:02.0000] <BenMillard>
Lachy, perhaps there was a bug during the copying...

[16:46:00.0000] <Lachy>
I saved the both files using wget, stripped out all markup before and after the <table>...</table> and resaved the files

[16:46:01.0000] <BenMillard>
Lachy, what numbers do you get and by which method? I expect the error is on my part

[16:46:02.0000] <Lachy>
For noscope.html, I get 1659, and for complexdatatable.html, I get 2591

[16:48:00.0000] <BenMillard>
hmm, what types of line endings do you see? I see 2 characters per newline, which might be my error

[16:48:01.0000] <Lachy>
what method did you use to get the sizes?

[16:49:00.0000] <BenMillard>
I viewed source in Firefox 2, copied and pasted into a plain text editor, then selected the text from the start of "<table" to the end of "</table>" and read off what the statusbar said

[16:49:01.0000] <BenMillard>
your method sounds better to me :)

[16:49:02.0000] <gsnedders>
/me reads the start of BenMillard's email and turns off

[16:49:03.0000] <Lachy>
line endings are CRLF

[16:50:00.0000] <BenMillard>
Lachy, same here

[16:50:01.0000] <Lachy>
the problem with copying and pasting from firefox's view source is that it doesn't copy properly

[16:51:00.0000] <Lachy>
it frequently adds in extra lines in random places

[16:51:01.0000] <BenMillard>
Lachy, I now measure 2,625 for the complexdatatable.html

[16:51:02.0000] <BenMillard>
Lachy, yes I removed the empty lines

[16:51:03.0000] <BenMillard>
I'm happy to steal your numbers if that's OK by you :)

[16:52:00.0000] <Lachy>
no, those numbers are copyrighted to me! :-)

[16:53:00.0000] <BenMillard>
Lachy, perhaps you could e-mail me the files you downloaded and the cropped versions, then I can see where the difference is

[16:53:01.0000] <BenMillard>
thanks for going to the trouble to do this, btw

[16:55:00.0000] <Lachy>
http://lachy.id.au/temp/tables.zip

[16:57:00.0000] <BenMillard>
Lachy, both the text editor's selection and Windows Explorer agree with your numbers

[16:58:00.0000] <Lachy>
did you find the diffs with your files?

[16:59:00.0000] <BenMillard>
Lachy, in noscope.html I see class="header" on the <td> for each of the 3 dates

[16:59:01.0000] <jgraham>
argh


2008-08-30
[17:00:00.0000] <gsnedders>
I probably ought to do the work I'm meant to do for Computing

[17:00:01.0000] <gsnedders>
I'm really get quite far behind

[17:00:02.0000] <gsnedders>
Though de-facto as long as I've done everything I'm meant to by December it doesn't really matter when I do it

[17:00:03.0000] <BenMillard>
gsnedders, do prioritise things above this e-mail. I might not send for a day or two yet

[17:01:00.0000] <gsnedders>
BenMillard: I've just been dealing with other low priority email

[17:01:01.0000] <Lachy>
why? I took my files directly from juicystudio, and didn't modify anything else

[17:01:02.0000] <gsnedders>
BenMillard: From several months ago :)

[17:01:03.0000] <BenMillard>
Lachy, you see the class="headers" here?  http://juicystudio.com/wcag/tables/noscope.html

[17:01:04.0000] <jgraham>
Why did someone think it was a good idea for lxml to add a random doctype when parsing html documents?

[17:01:05.0000] <BenMillard>
sorry, class="header": "<td class="header">12/12/2005</td"

[17:01:06.0000] <gsnedders>
If you send me something low priority and don't get a reply within an hour or two, it'll probably take a few weeks or months :)

[17:02:00.0000] <gsnedders>
jgraham: Because libxml2's HTML support is just a big hack

[17:02:01.0000] <jgraham>
gsnedders: The problem is that their XML support tries to enforce XML rules

[17:02:02.0000] <BenMillard>
Lachy, the 3 instances of class="header" are also missing from your complexdatatable.html

[17:02:03.0000] <jgraham>
Like no : in tag names

[17:03:00.0000] <jgraham>
s/tag/attribute/

[17:03:01.0000] <Lachy>
Looks like they've just been removed from those files

[17:03:02.0000] <Lachy>
press reload

[17:03:03.0000] <BenMillard>
Lachy: yes, you're right

[17:03:04.0000] <BenMillard>
that's annoying

[17:03:05.0000] <Lachy>
I did at first, but that must have been cached

[17:03:06.0000] <BenMillard>
they're changing the record whilst I'm replying to it :(

[17:03:07.0000] <Lachy>
yeah, he's destroying the evidence

[17:04:00.0000] <BenMillard>
Lachy, OK, so if you saw those attributes were present then those are the values I'll give since they are historically accurate

[17:05:00.0000] <jgraham>
Those attributes were present for sure. I think I mentioned it in an email

[17:05:01.0000] <gsnedders>
Why hasn't anyone creating a decent Flickr downloader yet?

[17:05:02.0000] <gsnedders>
That is like, easy to use.

[17:05:03.0000] <jgraham>
gsnedders: What do you mean decent?

[17:05:04.0000] <Philip`>
gsnedders: Like, a web browser?

[17:05:05.0000] <gsnedders>
Philip`: But to download an entire set?

[17:05:06.0000] <Philip`>
gsnedders: Oh

[17:05:07.0000] <BenMillard>
Lachy, the absence of class="header" makes our numbers match, so at least I haven't forgotten how to count. :)

[17:05:08.0000] <gsnedders>
jgraham: Not having a crazily complex UI. Copy and pasting a URL from a browser would work fine.

[17:06:00.0000] <Philip`>
gsnedders: Write a few (dozen) lines of script to use their API?

[17:06:01.0000] <Lachy>
ok, so my values are wrong cause they're the new values without the classes

[17:06:02.0000] <BenMillard>
Lachy, correct

[17:06:03.0000] <gsnedders>
Philip`: Because I need something that works on my uncle's computer, so I could just write a web API

[17:06:04.0000] <gsnedders>
*web interface

[17:06:05.0000] <gsnedders>
/me yawns

[17:07:00.0000] <Lachy>
BenMillard, there are only 18 header cells by my count, not 20

[17:07:01.0000] <BenMillard>
Lachy, 12 in the <thead>, agreed?

[17:07:02.0000] <Lachy>
yes

[17:07:03.0000] <BenMillard>
Lachy, 2 in the first column of the <tbody>?

[17:08:00.0000] <Lachy>
plus the 6 for budgeted, actual and forcasted in the column

[17:08:01.0000] <BenMillard>
Lachy, 6 in column 7 as you say

[17:08:02.0000] <BenMillard>
Lachy, yep, I've forgotten how to add up them :)

[17:08:03.0000] <Lachy>
ah, I didn't count those first 2 as headers

[17:08:04.0000] <BenMillard>
oh wait, 12 + 6 + 2 = 20

[17:08:05.0000] <Lachy>
the "Partner Portal" ones?

[17:08:06.0000] <BenMillard>
Lachy, yeah

[17:09:00.0000] <BenMillard>
they are associated as being row headers

[17:10:00.0000] <Lachy>
in complexdatatable.html, they are. But in noscope.html, there's nothing that indicates they are headers

[17:10:01.0000] <BenMillard>
"<td scope="row" id="row1" rowspan="3">Partner Portal</td>" in http://juicystudio.com/wcag/tables/complexdatatable.html

[17:10:02.0000] <BenMillard>
Lachy, yeah, later in my e-mail I mention that test 1 is unfair

[17:10:03.0000] <Lachy>
ok

[17:10:04.0000] <BenMillard>
and that scope="row" was used in test 3 instead of using headers+id for all the associations

[17:10:05.0000] <Lachy>
you mean test 2

[17:11:00.0000] <BenMillard>
Lachy, oh sorry you're right

[17:11:01.0000] <BenMillard>
scope="row" is used in addition to headers+id in test 3

[17:11:02.0000] <Lachy>
oh, what are the final byte counts you used? I should check the percentage given too

[17:12:00.0000] <BenMillard>
Lachy, 1,704 and 2,625. yes, I have made percentage errors before now :)

[17:14:00.0000] <Lachy>
I get 54.05%

[17:14:01.0000] <BenMillard>
Lachy, when talking about test file 3, I say "5 cells use <td scope> and participate in headers+id, duplicating the association." so my e-mail about that aspect correct, I just got muddled during this review

[17:15:00.0000] <BenMillard>
Lachy, what is your calculation for that? Maybe I've forgotten percentage increase math...

[17:16:00.0000] <Lachy>
(2625 - 1704) / 1704 * 100 = 921 / 1704 * 100 = 54.04%

[17:16:01.0000] <BenMillard>
hmm, that's more complicated than what I did :)

[17:16:02.0000] <Lachy>
what did you do?

[17:17:00.0000] <BenMillard>
just now I tried 2,625 / 1,704 = 1.5404 so yours looks right

[17:17:01.0000] <BenMillard>
maybe I typod my sum first time round

[17:17:02.0000] <Lachy>
I assumed I needed to find the difference, and then find out what percentage that difference was with the lower value

[17:18:00.0000] <BenMillard>
Lachy, so me saying 36% more markup was understating the code bloat by quite a bit! thanks for spotting that

[17:18:01.0000] <Lachy>
our numbers are consistent. Yours (1.5404) means that 2625 is 154% the size of 1704

[17:19:00.0000] <Lachy>
whereas mine says it's 54% larger

[17:19:01.0000] <BenMillard>
Lachy, yeah that's how I interpret it

[17:19:02.0000] <Philip`>
(Markup-size doesn't seem a very interesting measure when these tables are probably generated by programs from databases, and no human ever needs to look at the markup, and simplicity of implementing the table-generating code seems much more relevant)

[17:20:00.0000] <BenMillard>
Philip`, I've seen auto-generated headers+id, for sure

[17:20:01.0000] <Lachy>
Philip`, throwing lots of data at people, regardless of how relevant it is, is a useful techniqe for winning an argument :-)

[17:20:02.0000] <BenMillard>
I've also seen typoed headers+id

[17:21:00.0000] <Lachy>
fyorfty percent of all people know that, Kent

[17:21:01.0000] <Philip`>
Lachy: Winning an argument is not the aim; the aim is to design the best possible system :-p

[17:21:02.0000] <BenMillard>
Philip`, it's also worth considering that if the generating code can be radically simpler (such as just using <th> for all headers) that reduces the likelihood of bugs in the table

[17:22:00.0000] <Lachy>
s/fyorfty/forfty/ (I messed the simpsons quote :-))

[17:23:00.0000] <Philip`>
BenMillard: It's good to encourage people to do the simplest thing, but sometimes they just have complex tables, so I thought the issue was how to support the most complex tables (e.g. whether to force them to use <th> instead of <td>)

[17:23:01.0000] <BenMillard>
Philip`, that's right. So if a table can be supported by plain <th> using a sane association algorithm, that's preferable over the complexity and bloat of headers+id, in my judgement.

[17:24:00.0000] <BenMillard>
but I can well imagine irregular tables will sometimes be necessary and need headers+id, although even then all the headers could be done as <th>

[17:24:01.0000] <jgraham>
Hmm BenMillard keeps saying sensible things so I don't have to

[17:25:00.0000] <Philip`>
BenMillard: Would the headers attribute be supported only on <td>, not <th>?

[17:25:01.0000] <BenMillard>
Philip`, I haven't studied that in detail yet. would you like me to forward the message to you?

[17:26:00.0000] <Philip`>
How would it handle something like http://factfinder.census.gov/servlet/QTTable?_bm=n&_lang=en&qr_name=DEC_2000_SF1_U_DP1&ds_name=DEC_2000_SF1_U&geo_id=05000US48487 where the numbers need to be associated with the label in the first column, but the labels in the first column also need to be associated with some random set of other label cells?

[17:26:01.0000] <jgraham>
Philip`: I think there is likely a use case for @headers on th although no one has actually brought forward a table that needs it (at least recently)

[17:27:00.0000] <BenMillard>
Philip & jgraham, I call those "heirarchical row headers" although nobody else does :)

[17:27:01.0000] <Lachy>
"Test file 1 erroneously uses <td> for 10 of the 20 header cells" Which headers make up the 10? I only count 9

[17:27:02.0000] <BenMillard>
Lachy, I'll recount

[17:27:03.0000] <Lachy>
actually, 11

[17:28:00.0000] <Lachy>
3 dates, 2 x Partner Portal, 6 Budged/actual/forcast

[17:28:01.0000] <Philip`>
BenMillard: It might be best to not forward the email, since I have too many other things I ought to be working on instead :-)

[17:28:02.0000] <BenMillard>
Philip`, sure thing

[17:28:03.0000] <BenMillard>
Lachy, so we're talking about? http://juicystudio.com/wcag/tables/noscope.html

[17:28:04.0000] <Lachy>
yes

[17:28:05.0000] <jgraham>
Philip`: That table looks like it should actually be several smaller tables

[17:29:00.0000] <BenMillard>
Lachy, I agree with 11. thanks!

[17:29:01.0000] <Philip`>
jgraham: I don't think splitting it into smaller tables would help with the "One race -> Asian -> Asian Indian" label hierarchy, which is the main problem

[17:29:02.0000] <BenMillard>
(so this is another case where I understated the error)

[17:30:00.0000] <jgraham>
Philip`: I think that layout would need @headers on <th> 

[17:30:01.0000] <Hixie>
iirc you can actually do Philip`'s table with some careful use of rowspans, but i forget if i ended up making that work or not (and it's dubious whether that's desireable anyway)

[17:30:02.0000] <jgraham>
Philip`: Sure but it would have confused me  hell of lot less

[17:30:03.0000] <Philip`>
(Also splitting it into smaller tables would make the layout go all ugly, because you want them to all be exactly the same column sizes, and there's no way to enforce that when they're multiple tables)

[17:30:04.0000] <jgraham>
adn I can see it

[17:30:05.0000] <BenMillard>
Hixie, yes, rowspan works for "heirarchical row header" case...if you've got enough width to present it

[17:32:00.0000] <Hixie>
Philip`: i'm not sure what the best way to render that table is, but i'm pretty sure that "0. Subject, Race, One Race, Native Hawaiian and Other Pacific Islander, Other Pacific Islander 2; Number" is not the best way to read out that cell

[17:32:01.0000] <BenMillard>
Philip`, the table uses fixed-width, such as width="385", so you could split it and keep the fixed widths

[17:32:02.0000] <Philip`>
BenMillard: Then you're making assumptions about how many pixels the user's font uses

[17:32:03.0000] <Hixie>
Philip`: which is presumably what one would get if we encouraged people to chain headers

[17:33:00.0000] <Hixie>
Philip`: it should definitely be possible to link columns into having the same widths even in different tables, though css can't do that (and likely won't for some time) so i agree that in this case we shouldn't assume that it is possible

[17:33:01.0000] <BenMillard>
Hixie, when moving from cell to cell the more sophisticated ATs only announce the headers which have changed

[17:33:02.0000] <Lachy>
BenMillard, "3 of the cells using <td scope="row"> also use rowspan." - I only see 2 scope="row" in test 3

[17:34:00.0000] <Hixie>
BenMillard: well then it would sound exactly like if there weren't chained headers, assuming you're navigating the table linearly

[17:34:01.0000] <Lachy>
and this assertion of yours is debatable "For scope to work here under HTML4, scope=""rowgroup" must be used with the appropriate use of <tbody> around the rows which are being spanned: "

[17:34:02.0000] <BenMillard>
Lachy, yep, well spotted

[17:34:03.0000] <Lachy>
the spec is ambiguous though

[17:34:04.0000] <jgraham>
Hixie: FWIW Al suggested that the common AT setup is to have headers red out on demand

[17:34:05.0000] <Lachy>
it says row, but technically it's still in 3 rows

[17:34:06.0000] <BenMillard>
Lachy, does scope="row" apply to multiple rows in HTML4?

[17:34:07.0000] <jgraham>
s/red/read/

[17:35:00.0000] <Lachy>
in fact, it doesn't say one way or the other

[17:35:01.0000] <Lachy>
it just says "row: The current cell provides header information for the rest of the row that contains it"

[17:35:02.0000] <Hixie>
jgraham: that would suggest it would render as: "Zero." zero what? crap, what are the headers? "Subject, Race, One Race, Native Hawaiian and Other Pacific Islander, Other Pacific Islander 2; Number" say what now?

[17:35:03.0000] <Lachy>
so does that mean the rest of the <tr> that contains it, or the rest of the row(s) that it's actually in?

[17:36:00.0000] <jgraham>
Hixie: I agree in this case it's pretty hard to understand. But I find that table pretty hard to understand so maybe it's just a badly designed table

[17:36:01.0000] <BenMillard>
Lachy, it seems to think a "row" is different from a "row group" so my reading is that scope="row" applies to exactly one line of cells across the table

[17:36:02.0000] <Hixie>
jgraham: quite possible

[17:37:00.0000] <Lachy>
You say "This further exemplifies how difficult the headers+id system is to get right", after you mention errors with scope=""

[17:37:01.0000] <Hixie>
jgraham: but i think "Zero." zero what? crap, what are the headers? "Other Pacific Islander 2; Number" would be easier to understand.

[17:37:02.0000] <BenMillard>
Lachy, can we nail down the scope="row" thing first? :)

[17:37:03.0000] <jgraham>
Lachy: Trying to understand the HTML4 headers spec algorithm is a lost cause

[17:37:04.0000] <Hixie>
there's an algorithm?

[17:37:05.0000] <Lachy>
BenMillard, HTML4 is not clear enough to be certain one way or another

[17:37:06.0000] <Hixie>
i thought there was just some vague handwaving

[17:38:00.0000] <jgraham>
Hixie: Algorithm is a bit of a strong term

[17:38:01.0000] <jgraham>
vauge handwaving is indeed much closer

[17:38:02.0000] <BenMillard>
Lachy, it seems to make as much different between a row and a row group as it does between a column and a column group, though...

[17:39:00.0000] <BenMillard>
Lachy, indeed, why have a "rowgroup" value if "row" was intended to cover that case?

[17:39:01.0000] <Lachy>
hmm, perhaps.

[17:39:02.0000] <jgraham>
Hixie: re: what AT should read out; as I've said before this seems like exactly the sort of question that user testing could help answer

[17:39:03.0000] <jgraham>
BenMillard: If you care the Table Inspector has a HTML4 mode

[17:39:04.0000] <BenMillard>
Lachy, I agree that it's debateable, so I guess either interpretation is right. :)

[17:39:05.0000] <Lachy>
but I don't think it's a particularly strong argument

[17:40:00.0000] <jgraham>
BenMillard: I wouldn't expect miracles from it though

[17:40:01.0000] <Lachy>
anyway, with regards to that assertion I quoted above, the evidence you presented immediately before it doesn't support it

[17:42:00.0000] <BenMillard>
Lachy, I see what you mean

[17:43:00.0000] <BenMillard>
Lachy, my thinking was that headers+id "missed out" 8 associations in favour of using scope, while headers+id also duplicates the 6 associations which are made by scope

[17:44:00.0000] <BenMillard>
Lachy, I interpret the gaps and overlapping as authoring mistakes...

[17:44:01.0000] <Lachy>
if they're consistent, it's not really a mistake. Just redundant

[17:44:02.0000] <BenMillard>
Lachy, they are consistent, that's true

[17:45:00.0000] <BenMillard>
Lachy, what sentence would you suggest in place of that one?

[17:47:00.0000] <Lachy>
I don't know

[17:50:00.0000] <BenMillard>
Lachy, how about I strike that sentence and change the 1st one in that paragraph to "So, test file 3 uses a weird patchwork of techniques, with mistakes in the use of scope and colspan."

[17:51:00.0000] <Lachy>
yeah

[17:52:00.0000] <BenMillard>
Lachy, done. did you find anything else?

[17:53:00.0000] <BenMillard>
jgraham, thanks for your review, btw. Short but sweet. :)

[17:54:00.0000] <Hixie>
Lachy: yeah, but to do that we'd have to make a number of variants of that table, and then give each variant to three or four different users, and ask each user to answer questions about the table

[17:54:01.0000] <Hixie>
Lachy: so if we tried, say, three variants, and had three users, that's nine users to get under a usability study video camera

[17:55:00.0000] <BenMillard>
Hixie, is that towards jgraham?

[17:55:01.0000] <Hixie>
um

[17:55:02.0000] <Hixie>
yes

[17:55:03.0000] <Hixie>
my bad

[17:57:00.0000] <BenMillard>
I'll leave sending the mail about tables until tomorrow. I got a snapshot of all 3 tests.

[17:57:01.0000] <BenMillard>
Philip`, that table is going into my collection under "To Do".

[17:57:02.0000] <jgraham>
Hixie: Well I'm not sure how many people 9 is cmpared to the number that, say, Josh works with in a day. Plus given those 9 people they could each look at several different tables so once you had enough people to get data on one type of table, you'd have enough to get data on several 

[17:58:00.0000] <Hixie>
certainly would be great if we could do it

[17:59:00.0000] <jgraham>
Even without a full test like that one could try a single user with several similar tables and different amounts of verbosity, for example

[17:59:01.0000] <jgraham>
(one user obviously isn't a very good sample)

[18:10:00.0000] <BenMillard>
Philip`, I've actually put some notes with it, so it ended up as "USA FactFinder: Demographic Characteristics, 2000" here: http://projectcerbera.com/web/study/2008/collection#tables-government

[18:34:00.0000] <Dashiva>
/me equips vast-browser-wing-conspiracy hat

[18:52:00.0000] <takkaria>
ah, it's nice when you can mark 81 messages as read safely

[19:48:00.0000] <takkaria>
http://www.squarefree.com/burningedge/2008/08/29/2008-08-29-trunk-builds/ --  looks like yesterday was a pretty productive day for gecko

[19:51:00.0000] <jruderman>
that covers changes in the last two weeks, not just yesterday

[19:51:01.0000] <jruderman>
we only land that much in one day on crazy code freeze days

[19:52:00.0000] <takkaria>
ah, I thought it had rather a lot on it for a day

[19:52:01.0000] <takkaria>
still, pretty good going. :)

[19:53:00.0000] <alyosha>
hi ppl

[19:53:01.0000] <alyosha>
what do u guys think of IE 8 beta 2's HTML 5 support?

[19:54:00.0000] <alyosha>
I noticed (and I'm 100% sure I'm not the only one) a regression with unrecognized elements (eg. html 5 sectioning elements and inline elements such a mark)

[19:55:00.0000] <alyosha>
hopefully they'll fix it b4 final release

[20:05:00.0000] <takkaria>
have they removed the document.createElement() hack?

[20:05:01.0000] <alyosha>
yeah, pretty much

[20:06:00.0000] <alyosha>
but the elements to seem to show up correctly in the DOM tree in IE 8's developer tools

[20:07:00.0000] <alyosha>
*do

[20:13:00.0000] <alyosha>
I think it's probably an unintentional bug and they should fix it before final release, but I don't know for sure

[20:16:00.0000] <alyosha>
and the interesting thing is that the IE7 mode button is disabled on the html 5 doctype

[20:17:00.0000] <alyosha>
even though IE 7 rendering mode can be hacked to display new elements

[20:18:00.0000] <alyosha>
hmmm, what do u get with html 5 doctype and <meta http-equiv="X-UA-Compatible" content="IE=7">?

[20:22:00.0000] <alyosha>
html 5 doctype overrides the meta thingy

[20:23:00.0000] <alyosha>
IE 7 mode not available for html 5

[20:30:00.0000] <alyosha>
actually, they didn't remove the document.createElement() hack. It's just in the CSS, it makes unrecognized elements "UNKNOWN"

[20:30:01.0000] <alyosha>
just tried disabling script with the hack, and it still works

[20:30:02.0000] <alyosha>
but the styles just aren't applied

[20:34:00.0000] <alyosha>
style attributes are applied after the hack, but external stylesheets are not applied

[21:17:00.0000] <Hixie>
you gotta wonder what a mess their codebase is to get this kind of behaviour

[21:18:00.0000] <alyosha>
yeah, guess so

[21:19:00.0000] <alyosha>
IE 7 mode renders fine and shows the stylesheets fine too, but it can only be activated through developer tools or by adding the website to compatibility mode

[21:19:01.0000] <alyosha>
the meta thing is overridden by the doctype and the button is gone too

[21:20:00.0000] <alyosha>
gotta love M$, they make sure web designers won't lose their jobs (constantly gotta fix all their problems)

[21:20:01.0000] <alyosha>
lol

[21:22:00.0000] <alyosha>
nvm, adding it to compatibility view doesn't work either

[21:23:00.0000] <alyosha>
does MS have a bug tracker somewhere?

[21:24:00.0000] <Hixie>
https://connect.microsoft.com/feedback/AdvancedSearch.aspx?SiteID=136&Status=1&FeedbackType=1 i think?

[21:25:00.0000] <alyosha>
ooh, cool, Microsoft isn't completely submerged in the last decade after all.

[21:26:00.0000] <Hixie>
if you can get it to work, let me know

[21:27:00.0000] <alyosha>
sure. but I think most likely we'll have to wait for a fix from MS or do something like <header id="header"> ... #header { /*style here*/ } if they don't fix it

[21:31:00.0000] <alyosha>
according to this report IE8b1 didn't have this problem, so it's most likely a regression in IE8b2: https://connect.microsoft.com/IE/feedback/ViewFeedback.aspx?FeedbackID=364356

[21:34:00.0000] <alyosha>
well, g2g, l8rz

[23:20:00.0000] <Hixie>
"Ian's approach completely removes HTML conformance checking as a  

[23:20:01.0000] <Hixie>
mechanism to introduce authors to accessibility issues."

[23:20:02.0000] <Hixie>
-- http://html4all.org/pipermail/list_html4all.org/2008-August/000977.html

[23:20:03.0000] <Hixie>
well at least they admit that they are trying to use conformance checking for their own purposes

[23:21:00.0000] <Hixie>
and good to see others on that thread disagreeing with it :-)

[00:13:00.0000] <hsivonen>
wow. when I fixed bugs in my validation harness, it ran in 4 hours and the output was only 83.4 MB.

[00:16:00.0000] <hsivonen>
annevk: typo. thanks

[00:29:00.0000] <Hixie>
hsivonen: heh

[00:46:00.0000] <hsivonen>
whoa! there are many more 0-error docs than I would have thought

[00:49:00.0000] <Hixie>
hsivonen: 2?

[00:50:00.0000] <hsivonen>
Hixie: 4514

[00:50:01.0000] <Hixie>
out of a million?

[00:50:02.0000] <hsivonen>
out of 516875

[00:50:03.0000] <hsivonen>
and manual verification shows that it's really so

[00:50:04.0000] <hsivonen>
however, this ignores the document mode

[00:50:05.0000] <Hixie>
0.87%

[00:51:00.0000] <hsivonen>
so doctypeless files count

[00:51:01.0000] <Hixie>
does bgcolor in a transitional doc count as pass or fail?

[00:51:02.0000] <hsivonen>
fail

[00:51:03.0000] <hsivonen>
this is HTML5 rules

[00:51:04.0000] <hsivonen>
except for doctype

[00:51:05.0000] <Hixie>
wow that's not bad then

[00:52:00.0000] <hsivonen>
note that omitted alt doesn't count as an error

[00:52:01.0000] <Hixie>
what are we saying, that's horrific. but still. higher than i expected.

[00:52:02.0000] <hsivonen>
and IRIs on non-UTF-8 pages pass

[00:52:03.0000] <hsivonen>
no parse errors (doctype errors ignored) is 29%

[00:53:00.0000] <hsivonen>
which is rather high compared to your old numbers

[00:54:00.0000] <hsivonen>
but now the results look pretty consistent with what I've seen before in terms of the relative frequencies

[00:55:00.0000] <Hixie>
i had two numbers, one that counted /> and doctypes as errors and one that didn't

[00:55:01.0000] <hsivonen>
ah

[00:56:00.0000] <Hixie>
i forget what my exact numbers were

[00:56:01.0000] <Hixie>
but one was about 70% and one was about 90%

[01:25:00.0000] <annevk>
hsivonen, MB or GB?

[01:26:00.0000] <annevk>
gsnedders, hmm, you didn't do your checkin

[01:34:00.0000] <hsivonen>
annevk: MB

[01:35:00.0000] <hsivonen>
annevk: the harness used to have a simple but serious bug

[01:35:01.0000] <annevk>
but you expected 80GB initially?!

[01:35:02.0000] <hsivonen>
annevk: 60 GB actually, but that expectation was based on the bug, too

[01:35:03.0000] <annevk>
ok

[01:46:00.0000] <Hixie>
hsivonen: very interesting results

[01:47:00.0000] <Hixie>
hsivonen: these results really argue for consolidating all "attribute [known presentational attribute] not allowed" messages into a single message "This page contains presentational markup. More details... Help on removing presentational markup..."

[01:48:00.0000] <Hixie>
wow, 7% of pages had an </embed> ?

[01:49:00.0000] <hsivonen>
so it seems

[01:49:01.0000] <hsivonen>
crazy

[01:49:02.0000] <annevk>
lots of people think <embed> needs a closing tag, I once did so too

[01:49:03.0000] <annevk>
it's not like there was good documentation out there on how it works...

[01:51:00.0000] <Hixie>
wow, malformed byte sequences aren't that common either

[01:51:01.0000] <annevk>
"No “p” element in scope but a “p” end tag seen." 9%!

[01:52:00.0000] <annevk>
madness

[01:52:01.0000] <Hixie>
that's probably a lot of <p><table></table></p>-type stuff

[01:53:00.0000] <annevk>
and 5% had "Element “frameset” not allowed in this context. (The parent was element “html”.) Suppressing further errors from this subtree." so many frames still around?

[01:53:01.0000] <Hixie>
this sample didn't bias for date of creation

[01:53:02.0000] <Hixie>
so it includes stuff going back many years

[01:53:03.0000] <Hixie>
there's a lot of old content out there still

[01:55:00.0000] <Hixie>
sigh i really don't want to reintroduce <script language="">, people typo it so much

[01:55:01.0000] <Hixie>
and the & issue is a sad one

[01:55:02.0000] <annevk>
>2% uses <head profile>

[01:56:00.0000] <Hixie>
iirc there's a lot of pages that have <head profile=""> (blank)

[01:56:01.0000] <hsivonen>
annevk: wordpress.com gives distinct host names to users

[01:56:02.0000] <hsivonen>
annevk: livejournal, too

[01:56:03.0000] <Hixie>
like there are a lot of <a> elements with shape="rect"

[01:56:04.0000] <hsivonen>
annevk: I was too lazy to deal with those

[01:56:05.0000] <hsivonen>
annevk: although I did collapse MySpace profiles

[01:57:00.0000] <Hixie>
hsivonen: i'll give you a domain-separated set of urls next time instead of site-separated

[01:58:00.0000] <Hixie>
maybe we should make & followed by alphanumerics, followed by =, a non-ambiguous ampersand

[01:58:01.0000] <Hixie>
that might deal with a bunch of these & errors

[01:58:02.0000] <annevk>
"Bad value (consolidated) for attribute “lang” from namespace “http://www.w3.org/XML/1998/namespace” on element “html”: Bad language tag: Bad variant subtag." XML sites were included?

[01:59:00.0000] <hsivonen>
annevk: no

[01:59:01.0000] <hsivonen>
annevk: the validator sees HTML lang as XML lang internally

[01:59:02.0000] <takkaria>
Hixie: I think that could be a big win for authoring

[01:59:03.0000] <hsivonen>
annevk: and these messages weren't fully sanitized for UI consumption

[01:59:04.0000] <annevk>
Hixie, maybe also allow anything but [a-Z#]

[02:00:00.0000] <annevk>
to follow it

[02:00:01.0000] <Hixie>
annevk: ?

[02:00:02.0000] <annevk>
&" would be conforming

[02:00:03.0000] <annevk>
and so would 2&2

[02:01:00.0000] <Hixie>
the character encoding thing -- we could make <meta charset> allowed if not preceeded by any non-ASCII

[02:01:01.0000] <annevk>
or (&)

[02:01:02.0000] <Hixie>
annevk: i posit that the problem is just urls in attributes

[02:03:00.0000] <takkaria>
fwiw I'd prefer the "get a character reference" algorithm not to depend on whether you're in an attribute value state or not

[02:03:01.0000] <annevk>
I don't see what's wrong loosening them up both, given that you keep several extension points

[02:03:02.0000] <annevk>
takkaria, it already does

[02:04:00.0000] <annevk>
takkaria, and if we are to keep compat with IE, it has to be that way

[02:05:00.0000] <takkaria>
I mean in this particular case.  i.e. if you can paste an unescaped URL into an attribute value you should also be able to conformingly paste it outside an attribute value

[02:07:00.0000] <annevk>
that wouldn't work well

[02:08:00.0000] <annevk>
eg, it would go wrong with &AMP= which does different things

[02:09:00.0000] <takkaria>
mm, that's a point

[02:10:00.0000] <takkaria>
ah well.  it would be nice, though

[02:13:00.0000] <annevk>
at this point chaals would ask for a pony

[02:51:00.0000] <annevk>
grmbl, how do you properly configure lxml?

[02:52:00.0000] <annevk>
unzipped it's 25MB

[03:34:00.0000] <hsivonen>
Unsupported character encoding name: “iso-utf-8”. Will continue sniffing.

[03:34:01.0000] <hsivonen>
Unsupported character encoding name: “44-iso-8859-1”. Will continue sniffing.

[03:35:00.0000] <hsivonen>
crazy ebcdic charset in HTTP: http://web-sniffer.net/?url=http%3A%2F%2Fwww.antalis.fr%2Fsitesweb%2FFO%2Fpages%2Finterne-2-66-2122-rich_text-73228.html&submit=Submit&http=1.1&type=GET&uak=0

[03:36:00.0000] <hsivonen>
Unsupported character encoding name: “gb2312,big5,euc-kr”. Will sniff.

[03:37:00.0000] <hsivonen>
Unsupported character encoding name: “zh-tw”. Will sniff.

[03:37:01.0000] <hsivonen>
you can't make this stuff up

[03:54:00.0000] <jgraham>
hsivonen: btw, I'm not sure that such a thing as an unbiased sample of webpages exists

[03:55:00.0000] <hsivonen>
jgraham: sure. I said it was biased. :-)

[03:55:01.0000] <Philip`>
You can't even know how it's biased, because you can't know what the population is

[03:56:00.0000] <jgraham>
hsivonen: I know. I just think it's a tautology

[03:56:01.0000] <hsivonen>
yeah

[03:56:02.0000] <hsivonen>
and, yet, with different page sets, the same common errors come to the top

[03:59:00.0000] <jgraham>
In some sense approximately all the pages on the web are autogenerated pages which use the url to determine the content e.g. calendar.example.com/year/month/day with only implementation limits on the value of year

[04:00:00.0000] <jgraham>
So an unbiased sample of the whole population of http URLs that return 200 would be very misleading

[04:01:00.0000] <Philip`>
"approximately all" is not a concept that makes sense, where there's an infinite number of pages

[04:03:00.0000] <hsivonen>
more to the point, the number of pages in countably infinite which should make counting proportions a bit more tractable

[04:04:00.0000] <hsivonen>
Unsupported character encoding name: “big6”. Will sniff.

[04:04:01.0000] <gsnedders>
annevk: I asked if you wanted me to do it last night so you could work on it this morning. I got no answer :P

[04:04:02.0000] <Philip`>
Positive integers are countably infinite too, but it doesn't make sense to ask for an unbiased random sampling of positive integers

[04:05:00.0000] <gsnedders>
annevk: I took the default-lazy solution

[04:05:01.0000] <hsivonen>
Philip`: true, but you can say that half of the integers are positive

[04:05:02.0000] <Philip`>
hsivonen: No you can't :-p

[04:06:00.0000] <annevk>
gsnedders, I thought the default was yes!

[04:06:01.0000] <Philip`>
For every positive integer you give me, I'll give you back two negative integers, so there's twice as many :-)

[04:06:02.0000] <hsivonen>
Philip`: hmm. right.

[04:06:03.0000] <hsivonen>
now I appear silly and badly educated

[04:06:04.0000] <gsnedders>
/me attempts to cd Documents/Stuff\ I\'m\ Working\ On/spec-gen

[04:06:05.0000] <annevk>
gsnedders, I would appreciate a bundle of lxml+anolis+html5lib so I can just write the frontend script and don't have to worry about the bundling as I'm really bad at that

[04:07:00.0000] <annevk>
/me tried it this morning and couldn't get the lxml dependency to work

[04:07:01.0000] <gsnedders>
annevk: I've never tried bundling :)

[04:07:02.0000] <gsnedders>
annevk: lxml is written in C, which may make it harder

[04:07:03.0000] <Philip`>
(It does make sense to ask for an unbiased random real number between 0 and 1, even though that's an uncountable set)

[04:07:04.0000] <annevk>
gsnedders, I think that's the problem, yes

[04:07:05.0000] <Philip`>
(or at least I think it makes sense)

[04:08:00.0000] <gsnedders>
But it really does need to be for the sake of being reasonably quick

[04:09:00.0000] <annevk>
what's a difference between a pleonasm and tautology?

[04:09:01.0000] <Hixie>
hsivonen, Philip`: in this particular case the population was itself a (biased, non-random) subset of google's index

[04:10:00.0000] <annevk>
ah I see, tautology is also used in logic

[04:11:00.0000] <Hixie>
a tautology is specifically being overly specific in a redundant manner. a pleonasm is just using too many words. as i understand it.

[04:12:00.0000] <Philip`>
I think the logical meaning of tautology is a statement that's true regardless of the values of any variables in it

[04:12:01.0000] <annevk>
maybe the Dutch and English pleonasm are different then (in Dutch "round circle" is considered a "pleonasme")

[04:12:02.0000] <annevk>
Philip`, yeah

[04:13:00.0000] <Philip`>
/me guesses that must include all true statements that don't have any variables

[04:14:00.0000] <annevk>
"2. Logic. An empty or vacuous statement composed of simpler statements in a fashion that makes it logically true whether the simpler statements are factually true or false; for example, the statement Either it will rain tomorrow or it will not rain tomorrow."

[04:16:00.0000] <GregHouston>
Logical "proofs" of the existence of God generally falls into the category of a tautology.

[04:18:00.0000] <annevk>
gsnedders, anyway, for you the stuff is running right? can't you just zip that dir? :)

[04:19:00.0000] <gsnedders>
annevk: Only if you're running OS X/x86 :)

[04:19:01.0000] <gsnedders>
As of course the compiled C stuff…

[04:21:00.0000] <annevk>
grmbl

[04:24:00.0000] <annevk>
so how do I install lxml?

[04:24:01.0000] <annevk>
running setup.py install fails

[04:25:00.0000] <gsnedders>
annevk: http://codespeak.net/lxml/installation.html :P

[04:26:00.0000] <virtuelv>
annevk: sudo apt-get install python-lxml :P

[04:27:00.0000] <annevk>
hmm

[04:27:01.0000] <annevk>
/me wonders if dreamhost supports that

[04:27:02.0000] <virtuelv>
they don't

[04:28:00.0000] <gsnedders>
You need to install it in a custom path

[04:28:01.0000] <virtuelv>
on slicehost, that stuff is a bit easier, given that you have root

[04:28:02.0000] <annevk>
"annevk is not in the sudoers file.  This incident will be reported."

[04:29:00.0000] <annevk>
gsnedders, DreamHost doesn't have easy_install

[04:32:00.0000] <Philip`>
Do they have hard_install?

[04:33:00.0000] <Hixie>
there appears to be an inverse corrolation between how much actual useful research someone has done, and how much they ask people who are doing research to do more

[04:33:01.0000] <annevk>
-_-

[04:34:00.0000] <gsnedders>
/me is gonna have to install it on (mt)

[04:35:00.0000] <Philip`>
Hixie: That would be because the people who can do research themselves do it themselves instead of having to ask others :-)

[04:35:01.0000] <annevk>
grmbl, even if I do apt-get on my local machine it complains about lxml.html not being there :/

[04:35:02.0000] <Hixie>
that and they know how much work it is, i imagine

[04:35:03.0000] <Philip`>
It would be nicer if they said *why* they wanted that research, and what useful information it would be likely to reveal

[04:42:00.0000] <annevk>
gsnedders, I guess the lxml dependency is pretty big?

[04:42:01.0000] <gsnedders>
annevk: Yeah.

[04:42:02.0000] <annevk>
sigh

[04:43:00.0000] <gsnedders>
annevk: It's the structure used for the tree everywhere

[04:43:01.0000] <jgraham>
Philip`: It's not clear to me that there are an infinite number of web pages given likely limits on URL length supported by servers

[04:44:00.0000] <jgraham>
annevk: If you want python to work sensibly on Dreamhost you have to install it youself under your home directory

[04:44:01.0000] <jgraham>
Then you install easy_install

[04:44:02.0000] <jgraham>
Then you do easy_install lxml

[04:45:00.0000] <jgraham>
Then you just have to rember to change anything like #!/usr/bin/env python to #!/home/annevk/bin/python

[04:46:00.0000] <jgraham>
Otherwise using any external dependencies seems to be really hard

[04:47:00.0000] <gsnedders>
Not really

[04:48:00.0000] <jgraham>
gsnedders: It's getting the paths right so you can import stuff that seemed to be hard

[04:48:01.0000] <gsnedders>
export PYTHONPATH=${HOME}/packages/lib/python

[04:48:02.0000] <gsnedders>
export PATH=${HOME}/packages/bin:$PATH

[04:48:03.0000] <gsnedders>
in .bash_profile

[04:48:04.0000] <gsnedders>
That's what used on sp.org

[04:48:05.0000] <jgraham>
Hmm, I thought I tried that and it didn't work

[04:49:00.0000] <jgraham>
Anyway setting PYTHONPATH is a bad idea in general

[04:49:01.0000] <gsnedders>
That's true, but it works ;P

[04:50:00.0000] <gsnedders>
annevk: See what I just pushed

[04:50:01.0000] <gsnedders>
i.e., http://hg.gsnedders.com/hgwebdir.cgi/anolis/rev/cf4770338aa0

[04:53:00.0000] <virtuelv>
annevk: there is some tutorial for rolling your own python on DH

[04:53:01.0000] <virtuelv>
http://wiki.dreamhost.com/Python#Building_a_custom_version_of_Python

[08:27:00.0000] <gsnedders>
Time to go out into town to do something about the /topic

[08:28:00.0000] <jcranmer>
gsnedders: you're leaving your sense of logic behind?

[09:07:00.0000] <virtuelv>
gsnedders: I presume you'll put URL in /topic

[09:19:00.0000] <gsnedders>
/me is too impatient to wait in a queue of the length there was

[09:20:00.0000] <gsnedders>
(i.e., my hair is still the same old colour)

[10:42:00.0000] <hsivonen>
weird. my Mac had bluescreened (literally) while unattended

[10:45:00.0000] <gsnedders>
/me still has never got a pinkscreen

[10:45:01.0000] <Lachy>
hsivonen, do you mean a kernel panic?

[10:46:00.0000] <Lachy>
AFAIK, macs can't get BSODs

[10:46:01.0000] <gsnedders>
Lachy: They can however get stuck on a blank blue screen

[10:46:02.0000] <gsnedders>
Lachy: For no apparent reason

[10:47:00.0000] <Lachy>
I've never seen that

[10:48:00.0000] <gsnedders>
/me tries to decide in what order to post his blog posts

[10:48:01.0000] <Lachy>
I've had my machines have kernel panics a couple of times, and just freeze with the spinning beachball cursor.

[10:48:02.0000] <Lachy>
gsnedders, I'd recommend starting with number 1 followed by number 2

[10:49:00.0000] <gsnedders>
Lachy: It would make more sense to do them in chronological order, but the earlier one is far more time-consuming to write

[10:49:01.0000] <Lachy>
ok

[10:49:02.0000] <Lachy>
I have a number of blog posts I have to finish writing

[10:50:00.0000] <Lachy>
I suppose I should just post something about IE8 tonight, and then post my other, significantly longer, potentially 3-part series later

[10:52:00.0000] <gsnedders>
I have eight drafts currently

[10:53:00.0000] <gsnedders>
One gives a useful answer to <http://krijnhoetmer.nl/irc-logs/whatwg/20080605#l-450>

[10:54:00.0000] <gsnedders>
The other follows on from that

[10:55:00.0000] <Lachy>
I'd forgotten I'd even asked that question. I suppose it'll be good to get a better answer than "Stuff"

[10:55:01.0000] <gsnedders>
It was the first place I could think of that has a public record of me avoiding that question.

[10:59:00.0000] <gsnedders>
Writing about May last year is rather time-consuming.

[11:03:00.0000] <gsnedders>
/me smacks his old writing

[11:04:00.0000] <gsnedders>
It uses -ise :(

[11:04:01.0000] <Lachy>
VMWare ThinApp is absolutely brilliant! Now I can seamlessly run IE6, IE7, and IE8b1 and IE8b2 all within the same copy of Windows XP, which is itself running in VMWare Fusion on OS X.

[11:05:00.0000] <Lachy>
it basically runs each version of IE, or any other application I like, within its own sandbox

[11:06:00.0000] <gsnedders>
As long as no sand falls over the edge, I guess that's all right

[11:07:00.0000] <Lachy>
gsnedders, what is wrong with using -ise?

[11:07:01.0000] <gsnedders>
Lachy: en-gb-oed prefers -ize :P

[11:07:02.0000] <Lachy>
what?!

[11:07:03.0000] <Lachy>
nooO!

[11:08:00.0000] <Lachy>
-ize is wrong. Stupid American misspelling

[11:08:01.0000] <gsnedders>
No, it isn't.

[11:08:02.0000] <Lachy>
yes, it is

[11:08:03.0000] <Lachy>
I thought en-GB used -ise, just like en-AU

[11:08:04.0000] <gsnedders>
-ize comes from Greek, and should be used on Greek-derived words

[11:09:00.0000] <GregHouston>
Am I looking at the right thing. It looks like Thin App starts around $6000. I have Workstation and it was a little under $200.

[11:10:00.0000] <gsnedders>
en-gb only uses -ise, en-gb-oed uses -ize for words of Greek origin and -ise for those of French, en-us uses -ize

[11:10:01.0000] <gsnedders>
"[T]he suffix…, whatever the element to which it is added, is in its origin the Gr[eek] -ιζειν, L[atin] -izāre; and, as the pronunciation is also with z, there is no reason why in English the special French spelling in -iser should be followed, in opposition to that which is at once etymological and phonetic." — the OED

[11:11:00.0000] <gsnedders>
en-us also over does the entire z thing. Analyze is wrong.

[11:11:01.0000] <Lachy>
hmm, interesting

[11:12:00.0000] <gsnedders>
en-gb uses -ise too much, en-us uses -ize too much

[11:12:01.0000] <Lachy>
I still think -ise should be used for *everything*

[11:13:00.0000] <Lachy>
except for words like prize which are supposed to end in -ize

[11:14:00.0000] <Lachy>
wiktionary says that it's supposed to be -ise for french-origin words and -ize for greek-origin words. But to do that, I would have to know the origin of each word before I tried to spell it

[11:14:01.0000] <gsnedders>
me wonders whether he really should add a certain girl on Facebook…

[11:17:00.0000] <gsnedders>
(She is all ready convinced that I'm secretly in love with her, which is totally untrue)

[11:24:00.0000] <GregHouston>
It appears Thin App really is $6k. Application virtualization must be pretty tricky to cost 20 times that of a virtual machine.

[11:24:01.0000] <GregHouston>
I can't multipy. Make that 30 times.

[11:25:00.0000] <GregHouston>
Or spell. * multiply

[12:08:00.0000] <Philip`>
jgraham: You only need a single custom HTTP server that supports arbitrary-length URLs, and then the web can have an infinite number of pages, and I would have thought at least one person would have made such a server

[12:08:01.0000] <Philip`>
If nobody has, I'll make one, just to prove my point :-p

[13:10:00.0000] <gsnedders>
Philip`: You have calendars that can be navigated endlessly. There's no need for custom HTTP servers.

[13:17:00.0000] <Philip`>
gsnedders: But those calendars might have finite URL limitations

[13:23:00.0000] <Philip`>
(even if it's only limited by the amount of RAM available)

[14:58:00.0000] <gsnedders>
Philip`: Your webserver that supports arbitrary-length URLs will have the same RAM limitations

[15:00:00.0000] <Philip`>
gsnedders: No it won't - it won't store the URL in memory

[15:00:01.0000] <gsnedders>
Philip`: It just returns something for any request?

[15:03:00.0000] <Philip`>
gsnedders: It could ignore the URL entirely, or it could do some streaming processing of it to calculate a finite output

[15:03:01.0000] <Philip`>
(I assume HTTP doesn't particularly like you sending the response before you've received the request, so you can't do anything like echo the URL back to the client)

[15:04:00.0000] <gsnedders>
I don't think RFC2616 actually forbids you from doing so…

[15:48:00.0000] <Philip`>
gsnedders: Does it never require you receive the whole header so you can detect invalid requests and send an appropriate response?

[16:07:00.0000] <gsnedders>
Philip`: I don't think so. But it's not the best of specs.

[16:07:01.0000] <gsnedders>
Philip`: It doesn't require anything in specific in the case of invalid requests

[16:19:00.0000] <annevk>
Hmm, installing Python on DreamHost might be ok, but I can't even get lxml running in Ubuntu...

[16:34:00.0000] <Hixie>
/me battles iPod and Time Machine woes

[16:34:01.0000] <Hixie>
looks like the USB ports on my cinema display are busted

[16:34:02.0000] <Hixie>
no idea how THAT happened

[16:35:00.0000] <Philip`>
Maybe they're jammed full of popcorn and coke

[16:43:00.0000] <annevk>
nn


2008-08-31
[17:08:00.0000] <Philip`>
http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml doesn't work in Opera :-(

[17:08:01.0000] <Philip`>
e.g. it says "The format of the symbol name should be ___H_." instead of "The format of the symbol name should be <PROJECT>_<PATH>_<FILE>_H_."

[17:09:00.0000] <Philip`>
(XSLT, yay)

[19:31:00.0000] <Hixie>
did hsivonen mention if he was going to send an analysis of his data?

[23:43:00.0000] <BenMillard>
hello, is anyone here?

[23:56:00.0000] <BenMillard>
well, to anyone who reads the logs: I realised that redesigning the complexdatatable.html (test file 3) was suggested by a couple of people (including me), nobody has yet produced a redesigned variant for further review

[23:57:00.0000] <BenMillard>
so that's what I've been doing during the small hours of this morning, typing my reasoning as I go

[23:58:00.0000] <BenMillard>
since nobody is here, I'll go to sleep and return some time later...just thought this should be recorded somewhere

[01:43:00.0000] <annevk>
Hixie, no

[03:35:00.0000] <annevk>
html4all mailing list is funny

[03:35:01.0000] <annevk>
especially this recent debate between RB and LHS about what Hixie might have meant with "defining error handling"

[04:36:00.0000] <annevk>
http://groups.google.com/group/alt.html/msg/9d28ed5dc4c9fa36 "HTML 5 is a bunch of crazy egotists breaking stuff to no benefit."

[04:38:00.0000] <Lachy>
hah

[04:43:00.0000] <annevk>
these people see right through us, it's scary

[04:43:01.0000] <annevk>
in other news, base2 from Dean Edwards will support element.classList

[05:34:00.0000] <Lachy>
The XSLT void element thread is getting annoying.

[05:36:00.0000] <Philip`>
/me doesn't see why

[05:46:00.0000] <jgraham>
Lachy: FWIW I think Julian has a reasonable point

[05:47:00.0000] <jgraham>
I don't see the harm in allowing <eventsource></eventsource> as long as the closing tag immediatley follows the opening tag

[05:47:01.0000] <jgraham>
I would basically expect the parser to drop the closing tag

[05:48:00.0000] <jgraham>
So <eventsource>foo</eventsource> would be parsed like <eventsource>foo

[05:49:00.0000] <jgraham>
/me has wasted enough time on the html-wg today

[05:51:00.0000] <jgraham>
It would be different to </br> but not too different from <script> which requires a closing tag even when it is effectively void

[05:52:00.0000] <webben_>
jgraham: I thought a point was made that IE treats void element starting tags as self-closing?

[05:53:00.0000] <jgraham>
webben_: Maybe I haven't been following too closely. But I don't quite understand what you mean

[05:54:00.0000] <webben_>
actually, reading it again, I'm not sure what Robert meant

[05:54:01.0000] <webben_>
http://lists.w3.org/Archives/Public/public-html/2008Aug/0935.html

[05:54:02.0000] <webben_>
"IE treats unknown elements as void elements (adding the next element  

[05:54:03.0000] <webben_>
as a next sibling)"

[05:55:00.0000] <webben_>
i'm not sure what that means in terms of how IE treats tags

[06:02:00.0000] <robburns>
hi webben

[06:02:01.0000] <robburns>
an example

[06:02:02.0000] <robburns>
<header>

[06:02:03.0000] <robburns>
some contents

[06:02:04.0000] <robburns>
</header>

[06:02:05.0000] <robburns>
In DOM as:

[06:02:06.0000] <robburns>
html

[06:02:07.0000] <robburns>
   ⋮

[06:02:08.0000] <robburns>
        header

[06:02:09.0000] <robburns>
        #text: some contents

[06:02:10.0000] <robburns>
        /header

[06:02:11.0000] <robburns>
Instead of

[06:02:12.0000] <robburns>
html

[06:03:00.0000] <robburns>
   ⋮

[06:03:01.0000] <robburns>
        header

[06:03:02.0000] <robburns>
             #text: some contents

[06:03:03.0000] <robburns>
that's how IE will process an unknown non-void (HTML5 in this case) element

[06:03:04.0000] <robburns>
webbeb_

[06:06:00.0000] <robburns>
but on the other hand in Gecko

[06:07:00.0000] <robburns>
<p>

[06:07:01.0000] <robburns>
<eventsource>

[06:07:02.0000] <robburns>
some contents

[06:07:03.0000] <robburns>
</p>

[06:07:04.0000] <robburns>
html

[06:07:05.0000] <robburns>
   ⋮

[06:07:06.0000] <robburns>
        p

[06:07:07.0000] <robburns>
        eventsource

[06:07:08.0000] <robburns>
            #text: contents

[06:07:09.0000] <robburns>
instead of html

[06:07:10.0000] <robburns>
   ⋮

[06:07:11.0000] <robburns>
        p

[06:07:12.0000] <robburns>
            eventsource

[06:07:13.0000] <robburns>
            #text: contents

[06:09:00.0000] <Philip`>
(Actually it'd be called HEADER and /HEADER, not header and /header)

[06:09:01.0000] <Philip`>
(Wait, am I confused?)

[06:10:00.0000] <Philip`>
(At least I think IE does something funny with the case of unrecognised tags)

[06:10:01.0000] <robburns>
though in Firefox 3, the P element does not get implicitly closed (I think earlier versions would)

[06:10:02.0000] <robburns>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0A%3Cp%3E%0A%3Ceventsource%3E%0Acontent%0A%3C%2Fp%3E

[06:10:03.0000] <Philip`>
(Ah, right, it's if you do document.createElement('header') then the element is called "header" instead of "HEADER")

[06:12:00.0000] <robburns>
That's with the DOM, but with deserializing tex/thml 'header' will be 'HEADER', but it will be a void element followed by '/HEADER' (another void element)

[06:12:01.0000] <robburns>
in IE that is

[06:13:00.0000] <robburns>
s/ text/html / tex/thml

[06:14:00.0000] <Philip`>
It's not just for elements created by the DOM APIs: if you write <script>document.createElement('header')</script> then any subsequent <header>...</header> in the HTML markup will be treated as an element named 'header' with content, rather than as empty 'HEADER' + content + empty '/HEADER', because IE is insane

[06:14:01.0000] <Philip`>
but that's not particularly relevant to the point anyway

[06:35:00.0000] <gsnedders>
/me says hi to all the bureaucracy trolls

[06:47:00.0000] <jgraham>
/me wonders if that is supposed to include him

[06:49:00.0000] <jgraham>
/me doesn't see himself as trolling

[06:50:00.0000] <jgraham>
Since AFAICT the charter is designed to prevent exactly this

[06:50:01.0000] <Lachy>
jgraham, I disagree. We shouldn't let the limitations of tools like XSLT that clearly weren't designed to handle HTML well guide the development of HTML5.

[06:51:00.0000] <Lachy>
also, allowing </foo> for new void elements, but not for previously existing void elements makes things inconsistent.

[06:52:00.0000] <jgraham>
Lachy: I agree about the inconsistent part but that seems to have been a weak argument elsewhere (I think <a> enclosing blocks is a bad idea because it introduces inconsistencies)

[06:52:01.0000] <jgraham>
If the problem is XSLT-only it's not worth worrying about

[06:53:00.0000] <jgraham>
But if it is more general, it seems bad to couple the serializer to the version of the language being output

[06:54:00.0000] <jgraham>
because it makes migration hard 

[06:54:01.0000] <jgraham>
anyway I don't care very much

[06:57:00.0000] <Lachy>
so far, the problem is XSLT only.  No other tool has been mentioned with the problem

[06:58:00.0000] <Lachy>
and I just don't want XSLT to have any impact on the development of HTML5

[06:59:00.0000] <webben_>
I think the key thing is to ensure that XML output by XSLT can be translated with FOSS tools to text/html HTML5.

[07:00:00.0000] <webben_>
what I'm not sure about is what a tool is supposed to do if it gets XML that can't be translated to text/html HTML5.

[07:01:00.0000] <webben_>
e.g. <p>foo<blockquote><p>bar</p></blockquote></p>

[07:07:00.0000] <gsnedders>
webben_: It should do a backflip?

[07:07:01.0000] <gsnedders>
webben_: I mean, we already require UAs to take a deep breath in a certain case when parsing HTML

[07:07:02.0000] <Lachy>
webben_, you just have to accept that not all possible XML trees can be losslessly converted to HTML

[07:08:00.0000] <Lachy>
webben_, especially in cases like that which are invalid in XHTML too

[07:10:00.0000] <Lachy>
webben, incase you missed my messages just then due to connection issues, http://krijnhoetmer.nl/irc-logs/whatwg/20080831#l-149

[07:10:01.0000] <webben>
ta Lachy :)

[07:11:00.0000] <webben>
well, in this case, it's lossy because no feature has been provided to make it unlossy, not because of some sort of fundamental lack in text/html.

[07:12:00.0000] <webben>
but the worry is not just that it's lossy, but that such a tool would properly want to handle it somehow

[07:12:01.0000] <webben>
at which point you end up with differing implementations.

[07:12:02.0000] <webben>
that's not invalid in XML serialization of HTML5.

[07:13:00.0000] <Lachy>
it would have to make the blockquote a sibling of the p instead, and possibly create a new p after it

[07:13:01.0000] <webben>
it could do that. it wouldn't _have_ to.

[07:13:02.0000] <Lachy>
so <p>foo<blockquote>bar</blockquote>baz</p> would become <p>foo</p><blockquote>bar</blockquote><p>baz</p>

[07:13:03.0000] <webben>
another approach, for example, would be to go lossy in a different manner

[07:13:04.0000] <webben>
by genericising the p to a div.

[07:14:00.0000] <webben>
or processing agents could attach a class to the second p in an attempt to indicate what's happened.

[07:14:01.0000] <Lachy>
so it would become this instead? <div>foo<blockquote>bar</blockquote>baz</div>

[07:14:02.0000] <webben>
could do.

[07:15:00.0000] <Lachy>
perhaps the right solution is for the tool to check with the user what he wants to do

[07:15:01.0000] <webben>
that simply passes the problem on

[07:16:00.0000] <Lachy>
so do you want the spec to define how to serialise it?

[07:16:01.0000] <webben>
I'd prefer that, yes.

[07:16:02.0000] <webben>
i'd prefer the spec to define how XML HTML5 should be converted to text/html HTML5.

[07:16:03.0000] <gsnedders>
and the other way too

[07:16:04.0000] <webben>
yep

[07:17:00.0000] <Philip`>
webben: Like http://www.whatwg.org/specs/web-apps/current-work/multipage/serializing.html#serializing ?

[07:17:01.0000] <gsnedders>
Henri sent something about it a while back

[07:17:02.0000] <gsnedders>
Like, what he does in the validator

[07:17:03.0000] <Philip`>
/me doesn't know whether that algorithm is inappropriate for this

[07:19:00.0000] <webben>
that appears to do it.

[07:19:01.0000] <webben>
I don't like how it does it, but at least it does it.

[07:19:02.0000] <Lachy>
Philip`, that algorithm in the spec doesn't seem to have any special handling for the case being discussed. AFAICT, it would just output an invalid fragment

[07:21:00.0000] <webben>
the note says: "having a p element that contains a ul element (as the ul element's start tag would imply the end tag for the p"

[07:22:00.0000] <Lachy>
but that seems to be what Firefox does anyway

[07:22:01.0000] <webben>
does the algorithm not match the note?

[07:22:02.0000] <Lachy>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0A%3Cscript%3E%0Avar%20div%20%3D%20document.createElement(%22div%22)%3B%0Avar%20p%20%3D%20document.createElement(%22p%22)%3B%0Avar%20bq%20%3D%20document.createElement(%22blockquote%22)%3B%0Adiv.appendChild(p)%3B%0Ap.appendChild(bq)%3B%0Aw(div.innerHTML)%3B%0A%3C%2Fscript%3E%0A

[08:39:00.0000] <gsnedders>
/me wonders what he's meant to know for physics test tomorrow

[09:33:00.0000] <Philip`>
gsnedders: Remember that gravity goes downwards

[09:33:01.0000] <Philip`>
That's probably enough for a few marks

[10:20:00.0000] <hsivonen>
Hixie: yes, I intend to follow up with more opinion and analysis regarding the validation results

[11:22:00.0000] <gsnedders>
Philip`: :)

[11:22:01.0000] <gsnedders>
Philip`: Actually, with what will be in the test, gravity may as well not exist

[11:37:00.0000] <hsivonen>
Lachy: no, I didn't mean a kernel panic. I meant the screen turning blue with an occasional large instance of the spinning petals at the location where they appear a moment before shutdown

[11:38:00.0000] <hsivonen>
Lachy: Macs draw a light blue screen when there's nothing drawing the desktop background

[11:42:00.0000] <hsivonen>
gsnedders: I don't have a mapping from non-conforming application/xhtml+xml onto text/html

[11:42:01.0000] <gsnedders>
hsivonen: I thought I saw an email from you about that a while ago

[11:42:02.0000] <gsnedders>
/me shrugs

[11:43:00.0000] <hsivonen>
gsnedders: I have mappings from text/html onto XML, text/html onto Jing compatible XML-violating SAX and conforming application/xhtml+xml onto text/html

[11:43:01.0000] <hsivonen>
gsnedders: are you sure my email wasn't about going in the other direction than what webben asked about?

[11:43:02.0000] <gsnedders>
hsivonen: Myself and webben were saying both directions

[11:43:03.0000] <hsivonen>
ah ok.

[11:43:04.0000] <gsnedders>
hsivonen: So you have part of that

[11:44:00.0000] <hsivonen>
I have text/html to XML (covered by spec) and *conforming* application/xhtml+xml to text/html (not covered by the spec)

[11:45:00.0000] <hsivonen>
I'd love to see a sane and uncontroversial mapping from non-conforming application/xhtml+xml onto text/html

[11:52:00.0000] <gsnedders>
hsivonen: Where is text/html to XML covered in the spec?

[12:03:00.0000] <hsivonen>
gsnedders: http://www.whatwg.org/specs/web-apps/current-work/#coercing

[12:04:00.0000] <gsnedders>
Hixie: typo: mutatiosn

[12:10:00.0000] <hsivonen>
so now Google is saying Android supports H.264

[12:10:01.0000] <hsivonen>
what's the licensing story?

[12:51:00.0000] <Lachy>
hsivonen, where did you find that Android supports h.264?

[12:52:00.0000] <hsivonen>
Lachy: http://code.google.com/android/what-is-android.html

[12:53:00.0000] <Lachy>
Android uses the Apache2 licence

[12:53:01.0000] <hsivonen>
Lachy: it has patent language

[12:53:02.0000] <hsivonen>
how does that work with MPEG-LA?

[12:54:00.0000] <Lachy>
I don't know. What does the apache licence say about patents/

[12:54:01.0000] <Lachy>
?

[12:54:02.0000] <Dashiva>
It says "based on PacketVideo's OpenCORE" whatever that is

[12:55:00.0000] <hsivonen>
Lachy: hmm. actually, it only talks about patents that are licensable by a Contributor

[12:58:00.0000] <Lachy>
it's possible that Andriod could include closed source codecs for h264, and other patent encumbered stuff

[13:03:00.0000] <gsnedders>
Probably can be disabled in the build, and can be used only by those with licenses

[13:15:00.0000] <hsivonen>
hmm. the Microsoft EOT marketing partner Ascender made the Droid fonts

[13:22:00.0000] <hsivonen>
boohoo. Droid isn't Free as in Free Softwarle

[13:22:01.0000] <hsivonen>
Software even

[13:25:00.0000] <gsnedders>
/me is feeling that Sunday evening feeling: he ought to do homework for tomorrow.

[14:13:00.0000] <hsivonen>
http://twitter.com/stevefaulkner/statuses/902493395

[14:22:00.0000] <othermaciej>
hsivonen: self-fulfilling prophecy!

[14:54:00.0000] <gsnedders>
Hixie: You have any thoughts on making the spec-gen able to split a spec?

[15:00:00.0000] <gsnedders>
/me blogs http://gsnedders.com/anolis

[15:10:00.0000] <Philip`>
gsnedders: Why should that be part of the spec-gen, rather than a separate tool?

[15:12:00.0000] <Philip`>
gsnedders: s/it's/its/ :-p

[15:14:00.0000] <Philip`>
gsnedders: The filenames ought to say what the file is, rather than just being "1.0RC1.tar.bz2", else I'll forget what ~/download/1.0RC1.tar.bz2 is and it will annoy me

[15:16:00.0000] <gsnedders>
Philip`: The alternative is just to call it the revision hash :P

[15:17:00.0000] <Philip`>
gsnedders: The alternative is to call it anolis-1.0RC1.tar.bz2, and if Mercurial makes that alternative hard then I don't care and it's your problem :-p

[15:17:01.0000] <gsnedders>
Philip`: :P

[15:17:02.0000] <Philip`>
You could set up a script that downloads the archive files from Mercurial and renames them and copies them to a proper web server or something

[15:18:00.0000] <gsnedders>
/me is too lazy :P

[15:20:00.0000] <Philip`>
gsnedders: But you're not too lazy to go to all the effort of having Release Candidates and documentation and everything? :-)

[15:21:00.0000] <gsnedders>
Philip`: RCs are no effort. `hg tag 1.0RC1` and wait for people to bitch about a bug

[15:21:01.0000] <gsnedders>
Philip`: It's more a spec of what it does than a doc, covering all minor details, so if you want to copy it you can do it without looking at the code :P

[15:22:00.0000] <Philip`>
/me wonders why anyone would want to do that

[15:22:01.0000] <gsnedders>
Philip`: Well I did it to the CSS3 Module Postprocessor, didn't I?

[15:23:00.0000] <Philip`>
gsnedders: As far as I'm aware, you copied it without looking at the code because you couldn't copy its code, which is not a problem anyone would have when wanting to copy your code

[15:23:01.0000] <gsnedders>
Philip`: I could look at some of the code.

[15:24:00.0000] <Philip`>
Did you?

[15:24:01.0000] <gsnedders>
Yes, then couldn't work out how it dealt with some edge cases how it did, so went back to reverse-engineering it

[15:24:02.0000] <gsnedders>
(the issue really being my knowledge of C)

[15:24:03.0000] <Philip`>
Oh, okay

[15:25:00.0000] <Philip`>
I suppose if you don't like reading the code then that'd be a reason to copy it without reading the code

[15:26:00.0000] <gsnedders>
I could've looked at the C spec, but I thought it'd be just as easy to create a test case :P

[15:30:00.0000] <Philip`>
I think a book on C would be somewhat more useful than the C spec

[15:31:00.0000] <Philip`>
(unless you were trying to implement a C compiler yourself)

[15:31:01.0000] <Philip`>
(which is not entirely recommended)

[15:32:00.0000] <Philip`>
(and if you were implementing a C compiler, you'd probably want to start with the "how to write a C compiler" book instead of the specification)

[15:43:00.0000] <jcranmer>
well, it is an ISO spec

[15:43:01.0000] <jcranmer>
but it's not unreadable

[15:44:00.0000] <jcranmer>
then again, I *am* the kind of person who finds reading the contractual agreement that comes with your bank account quite fun to read...

[15:45:00.0000] <hsivonen>
ISO drafting rules suck

[15:45:01.0000] <hsivonen>
is the C spec available online?

[15:45:02.0000] <hsivonen>
I'm interested in seeing an ISO spec that is not unreadable

[15:46:00.0000] <hsivonen>
/me googles

[15:46:01.0000] <jcranmer>
you have to pay for the specs, but any drafts you happen along are free :-)

[15:46:02.0000] <jcranmer>
/me points to the draft C++0x spec sitting on his hard drive

[15:47:00.0000] <Philip`>
I'd agree it's not unreadable - I've occasionally referred to it when attempting to prove a pedantic point to somebody, and it mostly made sense to me, but it's not so good for someone who doesn't already know C well

[15:48:00.0000] <gsnedders>
/me has a draft C99

[15:48:01.0000] <Philip`>
and I don't think I'd be happy with it if I was looking for some higher-level information, since it's just focussed on the details

[15:49:00.0000] <jcranmer>
you have to know how everything lays out at a high level before you dive into a spe

[15:49:01.0000] <jcranmer>
spec

[15:50:00.0000] <hsivonen>
wow. the ISO C spec actually looks more readable than a run-of-the-mill ISO spec

[15:51:00.0000] <jcranmer>
ISO C++ is fairly approachable, if you use <Ctrl>-F

[15:54:00.0000] <gsnedders>
Find makes most docs readable

[15:54:01.0000] <jcranmer>
not the OOXML spec!

[15:54:02.0000] <Philip`>
Has anybody noticed that while IE's x-ua-compatible at least attempts to support non-IE browsers by allowing strings like "FF=4", its document.documentMode doesn't at all (since it just returns the IE version number)?

[15:57:00.0000] <hsivonen>
the points made about hyatt's mythicalness are a bit repetitive

[15:58:00.0000] <Hixie>
hsivonen: the harm language="" brings is that people typo it all the time and it'd be better to have people type type="" instead.

[15:58:01.0000] <Philip`>
But they seem true

[15:58:02.0000] <Hixie>
regarding <foo></foo> vs <foo>, because of the parser model the </foo> would not have any effect, but it would mislead people into thinking that <foo>x</foo> was a <foo> containing an x

[15:58:03.0000] <Hixie>
so it's bad

[15:58:04.0000] <jcranmer>
you'd be surprised at how bad people are at spelling

[15:59:00.0000] <hsivonen>
Hixie: the validator helps people spell it right :-)

[15:59:01.0000] <jcranmer>
I have a nice, phonetic last name and 85+% of people manage to both misspell and mispronounce it...

[15:59:02.0000] <Hixie>
hsivonen: apparently not enough people :-)

[16:00:00.0000] <jcranmer>
I figure if people misspell my last name with such frequency, any word greater than about 7 characters will be too long to spell correctly

[16:02:00.0000] <hsivonen>
Hixie: so if type=text/javascript, language=JavaScript and langauge=JavaScript are no-ops, why is allowing language=JavaScript a problem?

[16:03:00.0000] <Philip`>
Hixie: If many people were misled that way, wouldn't <embed>x</embed> be much more common than it is?

[16:03:01.0000] <Philip`>
jcranmer: I've never had such problems with my last name, but I often do with my first name :-(

[16:03:02.0000] <Hixie>
hsivonen: because people will think language="" isn't a no-op, and attempt to include on other pages, and will mistype it, leading to validation errors and wasted time.

[16:04:00.0000] <Hixie>
Philip`: </embed> is pretty damn common. i haven't checked to see how many </embed>s aren't immediately behind the <embed> they supposedly match.

[16:05:00.0000] <Philip`>
More than the number of people who already use language correctly, and would get validation errors when switching to HTML5, and would waste time?

[16:05:01.0000] <Philip`>
Hixie: I've only seen ones containing whitespace, &nbsp;, <br>, etc, and <noembed>

[16:05:02.0000] <jcranmer>
/me notes that hsivonen himself (herself?) misspelled 'language'

[16:06:00.0000] <hsivonen>
jcranmer: I misspelled it once on purpose above

[16:06:01.0000] <Hixie>
Philip`: i think we need a better transition story

[16:06:02.0000] <hsivonen>
jcranmer: did I misspell it also accidentally?

[16:06:03.0000] <jcranmer>
if you did it on purpose, then no

[16:06:04.0000] <hsivonen>
Hixie: why isn't type=text/javascript non-conforming?

[16:07:00.0000] <Hixie>
hsivonen: because we need to have a type="" to handle other types of scripts, and if we don't have a type for javascript, people will ask what the type is for javascript.

[16:08:00.0000] <Hixie>
brb

[16:19:00.0000] <Philip`>
Hmm, I think IE8b2 never renders pages in IE7-mode unless explicitly told to, regardless of doctype

[16:20:00.0000] <Philip`>
i.e. pages that used to be quirks mode are IE5-mode, and pages that used to be standards mode are IE8-mode

[16:20:01.0000] <gsnedders>
That's what IE8b1 did :P

[16:21:00.0000] <Philip`>
I thought it made recognised standards-mode doctypes trigger IE7 mode

[16:21:01.0000] <Philip`>
Wait, I'm probably confused

[16:21:02.0000] <Philip`>
That's what it was going to do before they changed it

[16:21:03.0000] <Philip`>
I think

[16:22:00.0000] <gsnedders>
They changed it before b1

[16:24:00.0000] <Philip`>
Hmm... If I have a standards-mode page with an iframe with quirks-mode content, then document.querySelector === undefined from code within it; but iframe.contentWindow.document.querySelector == some function from the container document

[16:25:00.0000] <gsnedders>
heh

[16:30:00.0000] <roc>
yeah

[16:30:01.0000] <roc>
it's a very deep hole the IE team is digging

[16:31:00.0000] <gsnedders>
roc: What's the escape velocity?

[16:31:01.0000] <roc>
and the really bad news is that with every release it will get a lot worse

[16:32:00.0000] <roc>
there is no escape :-)

[16:32:01.0000] <gsnedders>
So above 3*10^8?

[16:32:02.0000] <gsnedders>
(ms^{-1})

[16:32:03.0000] <roc>
there's always Hawking raditation

[16:32:04.0000] <roc>
but that's not a fun way to travel

[16:33:00.0000] <gsnedders>
Hawking radiation?

[16:33:01.0000] <gsnedders>
/me notes his physics has limits

[16:33:02.0000] <roc>
so does mine, so I refer you to the Internet

[16:34:00.0000] <gsnedders>
What's that?

[16:34:01.0000] <Philip`>
It's fun how IE8's script debugger shows that element collection objects have properties named "length" and "ie8_length"

[16:34:02.0000] <Philip`>
and some other type of object has properties "item", "namedItem", "ie8_item" and "ie8_namedItem"

[16:35:00.0000] <roc>
heh

[16:35:01.0000] <Philip`>
Hawking radiation = matter/antimatter pair randomly springs into existence near a black hole, antimatter particle falls in, matter particle falls out, black hole radiates and shrinks, if I remember correctly

[16:36:00.0000] <Philip`>
Please don't ask me why it's the antimatter particle and not the matter particle that falls into the black hole

[16:39:00.0000] <gsnedders>
why is it the antimatter particle and not the matter particle that falls into the black hole?

[16:40:00.0000] <Dashiva>
If only Philip` had used RDFa in his IRC text, we could make a machine reason out the answer for us

[16:40:01.0000] <Philip`>
Hmm, Wikipedia effectively says it's to conserve energy

[16:41:00.0000] <Philip`>
(i.e. if the antiparticle (apparently not actually antimatter) with negative energy escapes from the hole, then it violates conservation of energy, for some reason)

[16:44:00.0000] <annevk2>
gsnedders, I commented on your blog, not sure it had effect

[16:49:00.0000] <jgraham>
/me has  notion that the particle/antiparticle thing is a ort of lie to children

[16:49:01.0000] <Philip`>
By "children" you mean "physics undergraduates"? ;-)

[16:49:02.0000] <BenMillard>
jgraham, I see the headers thread continued along the lines of Process instead of facts

[16:56:00.0000] <jgraham>
BenMillard: Yeah. It has become quite soul destroying. This may be partially my fault, I don't know. 

[16:57:00.0000] <jgraham>
Philip`: Well I actually meant "people who read popular science books". Physics undergrads don't learn about hawking radiation