00:02
<jgraham_>
Hmm, it looks like the test data isn't being packaged up properly
00:15
<MikeSmith>
jgraham_: I can test html5lib if you still need testing
00:15
<jgraham_>
MikeSmith: I think I've just found the problem I know about
00:15
<MikeSmith>
OK
00:16
<jgraham_>
I'll upload a new build in a minute then go to bed :)
00:16
<Philip`>
I can probably test on Python 2.3 if you care :-)
00:20
<jgraham_>
MikeSmith: btw, Bugzilla sounds like a good idea. It would be cool if we could get the UI to encourage people to write RFE-type proposals in the use-case/solutions style that Hixie has tried to encourage
00:20
<Hixie>
good luck
00:20
<Hixie>
it would be nice to have a single place i can go to where i can just go down the list of all feedback the spec has received and just deal with them one by one
00:20
<Philip`>
Sounds like an attempt at a technical solution to a social problem :-)
00:21
<Hixie>
public-html isn't working for this purpose, sadly (unlike whatwg, which has worked quite well)
00:24
<jgraham_>
Philip`: ? I think UI that encourages useful behaviour is not really a technical solution to a social problem. Maybe bugzilla coud be accused of being a technical solution to a social problem
00:25
<othermaciej_>
some social problems are well-solved by technical solutions
00:25
<Hixie>
Philip`: the page splitting seems to cut at poor points (e.g. look at where http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded.html#focus starts)
00:25
<Hixie>
voting is a technical problem to a social one
00:25
<Hixie>
so is money
00:26
<Dashiva>
And filtering email :)
00:26
<othermaciej>
voting is solving a social problem with a social problem
00:26
<Hixie>
my definition of "technical" may be a bit broad :-)
00:27
<Dashiva>
You have a social problem, and you think to yourself "I know, I'll use democracy." Then you have two problems.
00:28
<othermaciej>
money is a solution to the price calculation problem
00:28
<othermaciej>
so I'll buy that
00:28
<othermaciej>
(well, money + relatively free exchange)
00:29
<Philip`>
Hixie: Oops, that was probably meant to cut at #embedded0 instead of #embedded
00:29
<othermaciej>
Dashiva: yes, exactly
00:31
<Hixie>
Philip`: your script breaks the back button :-(
00:32
<Philip`>
jgraham_: I think my point was that the problem is people not providing detailed real use cases and balancedly presenting various possible solutions, and that seems far too complex and subtle a problem to solve by providing some labelled textareas
00:33
<Philip`>
Hixie: Is that only in the cases where you follow a broken link, and therefore it's better than just dumping you on the wrong page?
00:33
<Hixie>
any comments on the new text in this section?: http://www.whatwg.org/specs/web-apps/current-work/#print
00:33
<Philip`>
Also, is there any way to fix it, or are back buttons just fundamentally broken?
00:34
<Hixie>
Philip`: well i'm now using it to create multipage links (i find the right section in the big page, then stick "multipage/" in the uri and copy the result when giving links to people)
00:34
<Hixie>
Philip`: and then i can't get back to the big page easily :-)
00:34
<Hixie>
no idea how to fix it
00:34
<Hixie>
location.replace() maybe?
00:34
<Philip`>
Hixie: You need a slower web browser, so you can hit the back button twice
00:35
<jgraham_>
Philip`: Well obviously we can't force people to get things right but good UI can certainly make it more likely. Compare a blank textarea to a solution (better than the one that I was proposing) where you have a whole page with textareas labelled with things like "Use case: this should be a statement of a problem that you are trying to solve, without reference to the particular means of solving it (example)"
00:36
<Hixie>
Philip`: my slower browsers can't handle the big html5 page :-/
00:36
<Hixie>
Philip`: .replace() should do it, iirc
00:37
<Hixie>
i wish i knew why the annotation script didn't work in safari (editing seems to fail silently)
00:37
<Dashiva>
I wish I knew how to get an error console in safari
00:40
<Hixie>
the develop menu is supposedly able to do it
00:40
<MikeSmith>
yeah
00:41
<Hixie>
but the inspector thing can't handle the html5 spec
00:41
<Hixie>
so it doesn't help me much
00:41
<MikeSmith>
Develop -> Show error console
00:41
<MikeSmith>
Hixie: html5 spec in WI seems to work for me in Webkit nightly at least
00:41
<MikeSmith>
or did the last time I tried it
00:42
<Hixie>
for js debugging?
00:42
<MikeSmith>
ah
00:42
<MikeSmith>
no, have not tried that with html5 doc
00:42
<MikeSmith>
you mean for the js for the annotations?
00:43
<Hixie>
yeah
00:43
jgraham_
destroys his python installation by mistake
00:43
<Hixie>
editing annotations just seems to fail
00:43
<jgraham_>
this could be evidence that I should be asleep
00:44
<MikeSmith>
jgraham_: the sleep thing does help some
00:44
<jgraham_>
html5lib-0.11 will have to wait for the morning. Goodnight :)
00:44
<Hixie>
nn
00:44
<MikeSmith>
'night
00:45
<Dashiva>
Where's this develop bar thing? I can't seem to find it in the menus or the preferences.
00:46
<Dashiva>
Safari on mac only thing?
00:48
<MikeSmith>
maybe only mac-only, I dunno
00:48
<hdh>
there's a preference to display the Dev menu
00:48
<MikeSmith>
I have never tried Safari for windows
00:48
<hasather>
Dashiva: http://developer.apple.com/internet/safari/faq.html#anchor14
00:50
<Dashiva>
Nice. It tells me to add it to a file that doesn't exist. :)
00:50
<MikeSmith>
there's op
00:50
<MikeSmith>
an option in the UI also
00:50
<MikeSmith>
now
00:51
<MikeSmith>
or is in my Safari 3.1 on Mac at least
00:51
<Dashiva>
I got it now, I think
00:51
<MikeSmith>
Preferences/Advanced/Show Develop menu in menu bar
00:52
<Dashiva>
It was at the very bottom of the advanced preferences
00:53
<Dashiva>
It would've been a bit worrying if manual config file editing actually was required :)
00:53
<MikeSmith>
Dashiva: it used to be
00:54
MikeSmith
wishes Safari had a about:config or safari:config thing
01:07
Philip`
updates the multipage thing, but then realises it's probably a bad idea that http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded.html is now 404
01:07
<Hixie>
why?
01:08
<Hixie>
crap, forgot to change the subject line of my last message
01:08
<Philip`>
Because I copied your link (minus the fragment bit) and then it didn't work (and didn't redirect me)
01:09
<Philip`>
Does /(?:...)/ work in all JS implementations?
01:09
<Hixie>
it should work :-)
01:11
<MikeSmith>
Hixie: so I see that YouTube is adding a way for users to put text annotations in their videos
01:15
<Philip`>
Hixie: Should be split more sensibly now - http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded.html is no longer the top of a page
01:15
<Philip`>
Also it uses window.location.replace(), which avoids the problems with Safari's broken back button
01:18
<Philip`>
(The section splitting is partly manually defined now (because that's the only reasonable way I can find to make sure the page gets split sensibly into 100-200KB chunks), so someone should poke me if it seems to be doing something silly and splitting in the wrong places)
01:20
<Hixie>
the 404 fixing didn't work on that link
01:20
<Hixie>
but cool
01:20
<Hixie>
oh, it worked on reload
01:20
<Hixie>
weird
01:20
<Philip`>
Probably caching the script?
01:20
<Hixie>
caching issue yeah
01:20
<Philip`>
Opera seems particularly bad at that
01:22
<Hixie>
ok i need a break
01:22
<Hixie>
bbl
01:30
<MikeSmith>
dear lazyweb
01:31
<MikeSmith>
or rather Dear LazyWags
01:31
<MikeSmith>
Question -
01:32
<MikeSmith>
If if want to truncate some text down to some reasonable but arbitrary length that's appropriate for putting in an e-mail Subject header
01:32
<MikeSmith>
...what should length be?
01:33
<Philip`>
My mail client shows about 50 characters of subject
01:33
<Dashiva>
After a few levels of replies, the nesting hides the subject anyway :)
01:33
<MikeSmith>
do any RFCs or anything authoritative elsewhere advise about Subject headers over a certain length
01:34
<Philip`>
Gmail shows about 90
01:34
<MikeSmith>
Dashiva: this would be for automated public-html-diffs mail that nobody will be responding to anyway
01:34
<MikeSmith>
Philip`: thanks
01:34
<Philip`>
Both truncate longer messages, so it's reasonable to expect clients to truncate appropriately
01:35
<Philip`>
so it's probably reasonable to send significantly more and trust the client
01:35
<MikeSmith>
I think the RFCs say that content of headers must/should be wrapped if it longer than a certain number of chars
01:35
<Philip`>
where "significantly more" might be 128 to the nearest binary order of magnitude
01:35
<MikeSmith>
but I can't remember that the number is
01:37
<MikeSmith>
for this case, I don't need all that must of it preserved
01:37
<MikeSmith>
maybe I should just pick a number
01:37
<Philip`>
MikeSmith: Do you mean http://www.faqs.org/rfcs/rfc2822.html talking about 78-character lines?
01:37
<MikeSmith>
Philip`: aha
01:37
<MikeSmith>
yeah, prolly
01:37
<Philip`>
MikeSmith: I imagine it's hard to do this without picking a number :-)
01:38
<MikeSmith>
:)
01:38
<MikeSmith>
I meant pick a number without worrying about the header-wrapping thing or what clients do with longer subjects and trying to max out to what they can handle or whatever
01:39
MikeSmith
used to did some work on e-mail delivery systems when he was at Openwave but his RFC-memory sucks
01:39
<Philip`>
I suppose you could have some heuristic that assigns a Gaussian-ish weight distribution centered on some particular region and sum that with some spikes at nice cut points like the ends of sentences, and pick whatever point has the highest weight
01:40
<Philip`>
but then you'd still have to pick a center for that distribution, so it doesn't solve the indecisiveness problem
01:40
<MikeSmith>
heh
01:42
<MikeSmith>
well, I'll go with 78, because I have seen some mail systems and clients that muff up the unwrapping and end up inserting some non-visible or other funked-up character in long unwrapped subjects
01:43
<Dashiva>
Then you should make sure it's 78 bytes and not characters, in case the diff starts using fancy code points
01:44
<MikeSmith>
Dashiva: only way it would in this case is if Hixie starts putting such into his svn commit descriptions
01:49
<Hixie>
MikeSmith: i recommend 80 - length('Subject ')
01:50
Hixie
will not use non-ascii
01:51
<Hixie>
ok, xml:base vs <base>
01:53
<MikeSmith>
Hixie: thanks
07:38
<met_>
http://www.wait-till-i.com/2008/06/05/north280-bring-keynote-to-the-web/
07:39
<met_>
He told thay made abstract language for canvas, svg and flash
08:35
<Lachy>
MikeSmith, yt?
08:48
<annevk>
wow, that's impressive ( http://280slides.com/Editor/ )
09:31
<annevk>
Hixie, shouldn't the events be called beforeprint and afterprint?
09:32
<annevk>
Hixie, IE prefixes all their event names with "on" and we're not calling the focus event onfocus either...
09:32
Philip`
sees http://hackademix.net/2008/06/05/site-security-policy-aka-content-restrictions/
09:32
<annevk>
Hixie, of course, the Window object should probably define two event handler DOM attributes called onbeforeprint and onafterprint
09:35
Philip`
also sees http://www.techcrunch.com/2008/06/05/the-next-gen-web-html5-will-we-ever-see-a-real-standard/
09:36
<annevk>
wow, comparing HTML5 with HTML 3.0 misses the point entirely
09:36
<hsivonen>
uh oh. we're now on the TechCrunch radar
09:36
<MikeSmith>
Lachy: here now
09:37
<MikeSmith>
Lachy: btw, in hindsight, I think you should have dropped some Crockford into your @media presentatin
09:38
<hsivonen>
are we going into the deadpool yet?
09:38
<MikeSmith>
in one of his presentations, Thomas Roessler has this great clip of Crockford
09:38
<MikeSmith>
where Crockford says that from what he's seen 99% (or something) of the JS he's seen on the Web is crap
09:42
<Lachy>
MikeSmith, did Marcos ask you to add standardssuck.org to PlanetHTML5?
09:42
<MikeSmith>
Lachy: yeah
09:42
<Lachy>
ok
09:43
<Philip`>
"XHTML1.1/HTML4.01 do *not* need updating, they need support. The W3C should be working with and promoting their existing good specs (listed) and not inventing new specs (XForms, widgets) or creating new versions of perfectly good standards (HTML5)."
09:43
<MikeSmith>
but I forgot about it in the space of the one hour since he asked me
09:43
<MikeSmith>
I will do now
09:43
<MikeSmith>
Philip`: who say that?
09:45
<Philip`>
MikeSmith: Some TechCrunch commenter
09:46
<MikeSmith>
Whoever said it, I wonder if he knows about the requirement that uttering the phrase "HTML4.01 do *not* need updating" (or similar) requires him to take an extra big hit on his crack pipe when he says it.
09:48
<Lachy>
luckily for him, we're not updating HTML 4.01. We're just creating a new standard to replace it.
09:50
<annevk>
Lachy, I made some small changes to your post on standardssuck.org
09:50
<hsivonen>
:-( at comment #17 at techcrunch
09:50
<Philip`>
Peter Kasting says: "I’m beginning to wonder how much research was actually done for this article." - me too :-)
09:51
<Lachy>
annevk, ok
09:54
<MikeSmith>
standardscuk.org added to planet html5
09:54
<MikeSmith>
but only marcos post getting through the filter
09:55
<Lachy>
wow, "...and an under representation of independent web developers and users". Ironically, there are more web developers involved with the development of HTML5, than any previous version.
09:55
<MikeSmith>
because only marcos exercised the secret special magic
09:55
<Lachy>
what's the special magic? an "HTML" tag?
09:56
<Lachy>
*html5 tag
09:56
<hsivonen>
Peter Kasting++
10:00
<MikeSmith>
heh
10:01
<MikeSmith>
most useful comment on that blog entry is, by far, "Is it just me, or does the Silverlight logo look like a pair of panties?"
10:01
<MikeSmith>
Lachy: damn, who told you the secret?
10:01
<MikeSmith>
I think I'm gonna have to change it now
10:14
<MikeSmith>
btw, you guys are breaking some new ground in geek waggish-ness with that standardssuck stuff
10:14
<MikeSmith>
a friend of mine told me to tell you that he hopes you keep it up
10:14
<Lachy>
MikeSmith, what do you mean?
10:15
<Lachy>
cool
10:15
<MikeSmith>
and that you should insert more irreverance
10:16
<Lachy>
MikeSmith, we'll save that for when we talk about specs we don't like
10:16
<MikeSmith>
Lachy: I mean that some of this material is ripe to for picking as far as poking fun some of it goes
10:16
<MikeSmith>
I mean, that's what my friend means
10:16
<Lachy>
like RDF, GRDDL, XHTML2, Semantic Web nonsense
10:16
<MikeSmith>
the one who asked me to pass it on
10:16
<MikeSmith>
Lachy: I'm not naming any name here
10:17
<MikeSmith>
take your pick
10:18
<MikeSmith>
I think I also remember my friend suggested sprinkling the phrases like "architectural astronautics" and "a plan to build a rocket ship to the moon" in reference to particular specs or groups
10:18
<MikeSmith>
I personally would not condone that, though
10:19
<hsivonen>
hmm. will the next standards sucks episode be hosted by Steve Faulkner?
10:20
<MikeSmith>
my friend also mentioned something about he wondered if he could find a way to contribute to the standardssuck effort anonymously in some way
10:22
<MikeSmith>
Lachy: do you have favorite comedy film or TV comedy or animation?
10:22
<hsivonen>
MikeSmith: perhaps your friend could appear as a black silhouette with a heavily DSP-distorted voice
10:22
<MikeSmith>
heh
10:22
<Lachy>
MikeSmith, why?
10:22
<MikeSmith>
OK, I'll pass that on to him
10:22
<annevk>
hsivonen, heh :)
10:23
<MikeSmith>
Lachy: just wondering
10:23
<annevk>
hsivonen, how it currently went is not at all intentional
10:23
<MikeSmith>
I mean what your perspective on comedy is
10:23
<Lachy>
MikeSmith, I like lots of different shows. At the moment, I'm watching Friends cause I bought the DVDs in London
10:24
<MikeSmith>
one of my colleagues on the team really like Ren & Stimpy but I never found it very funny myself
10:24
<annevk>
Eric Cartman <<
10:24
<MikeSmith>
cartman++
10:24
<Lachy>
Simpsons is still one of the best cartoons ever
10:24
<MikeSmith>
Homer Simpson should be on the US 1 dollar bill
10:25
<MikeSmith>
he is the perfect representation of America and its values and collective intelligence
10:25
<Lachy>
of course, he deserves it more than Washington does!
10:25
<Lachy>
(it is George Washington on the $1 note now, right?)
10:26
<MikeSmith>
yeah
10:30
<MikeSmith>
btw, the ss.org "Unbiased journalism on Web standards since May 2008" masthead slogan in particular is solid gold
10:30
<MikeSmith>
my friend said
10:30
<annevk>
thanks
10:32
<Philip`>
"Unbiased journalism on Web standards since <del>May</del><ins>June</ins> 2008"
10:32
<Lachy>
MikeSmith, who is this anonymous friend of yours?
10:33
<MikeSmith>
Lachy: he's a mouse in my pocket
10:33
<MikeSmith>
so to speak
10:33
<Lachy>
oh, Stuart Little? Cool!
10:34
<annevk>
Philip`?
10:37
hsivonen
wonders what the business model of Cooliris, Inc. is (the makers of PicLens)
10:42
<Philip`>
annevk: I think I was thinking of something along the lines of http://www.penny-arcade.com/comic/2008/03/19/
10:42
hsivonen
reads the comments at http://weblogs.mozillazine.org/roc/archives/2008/06/applying_svg_ef.html
10:43
<MikeSmith>
Philip`: that is classic
10:43
<MikeSmith>
it reminds me of course they have in the United State Marine Corps
10:44
<MikeSmith>
it's called "Your Ass from a Hole in the Ground: A Comparative Study"
10:51
<roc>
enjoy
10:51
<annevk>
roc, quite cool stuff btw
10:52
<roc>
let me know if you can figure Chris Lilley out
10:52
<roc>
also, please cast votes for where you think this should be specced out
10:53
Philip`
thinks the people on standardssuck.org need to work out what to look at when they're not talking
10:55
<annevk>
roc, maybe chris is saying that if you define the coordinate system for HTML (or anything other than SVG) you don't need new specs
10:55
<annevk>
(not sure about his other comments about things that you're doing wrong though...)
10:55
<roc>
defining the coordinate system for HTML *is* a new spec
10:56
<roc>
You also need to amend the parts of the SVG spec that say what 'filter', 'clip-path' and 'mask' apply to
10:56
<annevk>
true
10:56
<annevk>
the CSSWG is the best place, but probably only in theory
10:57
<roc>
ok
10:58
<hsivonen>
CSSWG and WHATWG seem like potential candidates
10:58
<roc>
however, it's the SVG specs that need to be amended
10:58
<hsivonen>
coordinates for HTML rendering seem more like a CSS thing
10:58
<roc>
it's SVG behaviour that needs to be "reinterpreted"
10:59
<roc>
what a pain
10:59
<roc>
it would be simpler if Ian was the chair of all the working groups
11:00
<annevk>
haha
11:00
<annevk>
if the WHATWG had a patent policy...
11:01
<annevk>
btw, I think I agree with dhyatt in that the simple stuff should probably be doable using a select set of CSS properties without requiring SVG fragments
11:02
<roc>
I agree
11:03
<roc>
although we may not agree on what the "simple stuff" is
11:04
<roc>
the other thing I'm thinking about is how to handle this as a vendor extension
11:04
<roc>
it feels stupid to introduce -moz versions of the properties that are exactly the same as the standard properties except they apply to more stuff
11:05
<annevk>
there's precedent for simply putting your foot down in such cases
11:05
<annevk>
WebKit implements the multiple background stuff simply in 'background'
11:05
<Lachy>
annevk, http://www.paciellogroup.com/blog/?p=73
11:06
<annevk>
(though i'd be nice if there was some document written up first)
11:06
<roc>
yeah
11:06
<roc>
I'll write it
11:07
<annevk>
Lachy, heh, nice, a transcript
11:08
<annevk>
Philip`, suggestions? it's troublesome :)
11:08
<annevk>
roc, cool
11:09
<annevk>
roc, btw, how do you handle script execution and such in the referenced documents?
11:09
<roc>
we don't handle external references yet
11:10
<annevk>
ah ok, I suppose that makes it easier
11:10
<roc>
and a lot less useful
11:10
<annevk>
yeah
11:10
<roc>
I hope we can add external references fairly soon
11:10
<roc>
I think at least initially we won't run script in referenced documents
11:11
<roc>
but I would like to provide a DOM API that provides access to the DOM of referenced documents
11:12
<annevk>
our "SVG as image" implementation doesn't do scripts either, but that's probably slightly different
11:12
<roc>
it's related
11:12
Philip`
assumes it would be restricted to same-origin referenced documents
11:12
<annevk>
roc, at that point you get same-origin mess
11:12
<annevk>
hah
11:12
<roc>
sure
11:12
<roc>
doesn't Opera support general external references?
11:12
<roc>
I thought it did
11:12
<hsivonen>
IIRC, yes
11:12
<annevk>
true
11:13
<annevk>
i suppose those are different anyway
11:13
<roc>
the whole area is under-specified ... maybe you should write up what Opera does before we reverse-engineer it
11:14
<annevk>
e-mail ed⊙oc for that
11:14
<annevk>
he's lead of the SVG team and also does standards
11:16
<roc>
ok
11:16
<roc>
the whole SVG WG has not been all that responsive to my questions
11:17
<annevk>
:(
11:18
<annevk>
they don't really have editor driven specs
11:18
<annevk>
so everything goes through the WG telcon it seems
11:18
Dashiva
waits for a <Hixie> I told you so
11:18
<roc>
yeah
11:19
<annevk>
the CSS WG does have editor driven specs, but the process to agree upon doing a spec is lengthy
11:19
<hsivonen>
http://www.itscj.ipsj.or.jp/sc34/open/1037c.htm
11:19
<hsivonen>
what format is that?
11:21
hsivonen
wonders if MS Office has anything to do with EOT
11:22
<annevk>
the Open Font Format
11:23
<annevk>
"ISO/IEC 9541 is a standard of font information interchange and designed to be independent with concrete font
11:23
<annevk>
file format. The Open Font Format (ISO/IEC 14496-22) is a font file format specification that is based on
11:23
<annevk>
TrueType font file format. About the handling and utilization of the typographic properties stored in OFF font
11:23
<annevk>
file, ISO/IEC 14496-22 describes the implementations on Microsoft Windows or IBM OS/2 only. Therefore,
11:23
<annevk>
ISO/IEC 14496-22 does not define the method how to define a font resource in ISO/IEC 9541 architectures
11:23
<annevk>
from given OFF font file, because it is out of the scope. ISO/IEC 9541-4 is a standard to fill the gap between
11:23
<annevk>
OFF font file and font resource in ISO/IEC 9541."
11:23
<annevk>
-- http://www.itscj.ipsj.or.jp/sc34/open/1037.pdf
11:23
<hsivonen>
what's the point compared to .otf?
11:24
<annevk>
(it's funny that the link to the pdf file has the correct data but the incorrect metadata)
12:05
<annevk>
oh, seems Hixie already fixed the event names, duh
12:14
<Hixie>
just did it now while reading scrollback
12:15
<Hixie>
roc_: while i am flattered by your proposal, i don't think i have the time :-P
12:17
<roc_>
well, it's like that old joke
12:17
<roc_>
first prize: chair of the WG
12:17
<roc_>
second prize: chair of two WGs
12:19
<hsivonen>
Hixie: did you really mean to say 'id' (as opposed to 'name') under 'valid hashed reference'?
12:19
<Hixie>
roc: indeed
12:19
<Hixie>
hsivonen: i don't think i changed anything there, other than the term's name; should i have?
12:20
<hsivonen>
Hixie: isn't it supposed to point primarily to the name attribute now?
12:20
<hsivonen>
Hixie: since name is now the required attribute on map
12:21
Hixie
goes to actually read the part of hte spec hsivonen is talking about
12:21
<Hixie>
hm, yeah, that should be changed
12:21
Hixie
goes to fix
12:21
<hsivonen>
Hixie: thanks
12:22
<Hixie>
i'll change the term to "hash-name reference" while i'm at it
12:23
<Hixie>
lest i accidentally use it to refer to something expecting an id reference
13:56
hsivonen
reads http://www.w3.org/2001/tag/2008/05/19-minutes.html
13:59
<hsivonen>
"Henri says they don't think that namespaces are important"
14:00
<hsivonen>
I don't recall saying exactly that. I think I've said that in retrospect, Namespaces wouldn't have been necessary.
14:03
<annevk>
i suppose you can e-mail www-tag to clarify, but it prolly doesn't matter much
14:03
<MikeSmith>
yeah
14:03
<MikeSmith>
that battle been won
14:06
<hsivonen>
I'll clarify my opinion if it comes up in the context of MathML and SVG (which is what I thought the context was initially until I realized how old the minutes were)
14:06
<hsivonen>
also, "I asked Henri and Anne if a new version of XmL that was as HTML friendly as possible would be acceptable to HTML WG, and they demurred."
14:06
<hsivonen>
demurred: showed a contradiction in legacy HTML and legacy XML requirements
14:07
<annevk>
yeah :/
14:26
<heycam>
is 'this == evt.currentTarget' always true when an EventListener is invoked?
14:27
<zcorpan_>
Hixie: what about print preview and onbeforeprint?
14:28
<hsivonen>
zcorpan_: </noscript> fixed. thanks
14:29
<annevk>
heycam, should be
14:31
<annevk>
hmm, anyone with tips on how to implement dataset in JavaScript?
14:34
<annevk>
getters and setters don't seem to cover the case where someone gets an arbitrary property on an object or sets one
14:35
<heycam>
writing a whole custom [[Get]] and [[Put]] sounds like the way to go
14:51
<annevk>
how do you go about that?
15:04
<Philip`>
hsivonen: http://validator.nu/?doc=http%3A%2F%2Fwww.atkinsfamily.ukonline.co.uk%2Fmayweek_2008.html&parser=xmldtd&laxtype=yes&showsource=yes says "Error: Attribute shape not allowed on XHTML element a at this point." when actually there's no shape attribute
15:05
<hsivonen>
Philip`: wow. That's weird. thanks
15:05
<hsivonen>
Philip`: hey! you are loading external entities
15:05
<hsivonen>
Philip`: so sure, there's an attribute shape
15:06
<Philip`>
hsivonen: Uh
15:06
<hsivonen>
DTDs <3
15:06
<Philip`>
Do you mean there is a shape attribute on that a element?
15:06
<hsivonen>
Philip`: yeah, defaulted from the DTD
15:07
<Philip`>
Oh - that seems extremely stupid
15:07
<hsivonen>
there's a reason why that mode isn't the default :-)
15:08
<Philip`>
It works fine if I select the "XHTML 1.0 Strict, SVG 1.1, MathML 2.0 + IRI" preset
15:08
<Philip`>
just not with XHTML5 + ...
15:09
<hsivonen>
that's because shape is allowed in XHTML 1.0 but not in XHTML5
15:09
<Philip`>
Oh, right
15:09
<Philip`>
I think that almost makes sense
15:11
<Philip`>
But it is kind of confusing :-)
15:11
<hsivonen>
well, that's XML
15:12
<Philip`>
It's an attempt at validating a reasonably typical web page, and getting something crazy and seemingly inexplicable from it
15:13
<hsivonen>
should I add warnings whenever a DTD actually has an effect?
15:15
<Philip`>
I want the DTD to have the effect of stopping the complaints about &ndash; etc since I know those are legimitate and not practical errors
15:15
<Dashiva>
Oh look, another group wanting XHR without window defined
15:15
<hsivonen>
Philip`: it seems to me you want the XML parser to default to Gecko emulation
15:16
<hsivonen>
I'd feel better about implementing Gecko emulation if it was a WHATWG spec
15:16
<hsivonen>
But I suppose I should do it anyway.
15:18
<Philip`>
I'm not sure what I want, except for it to say "this page (which happens to claim to be XHTML and I'm not going to bother telling the author to change to something else since that's a waste of time) is broken for these reasons (of which all the reasons are legitimate problems, and not stupid things like entities not quite being defined even though it'll work fine in practice)"
15:20
<annevk>
Dashiva, yeah, not sure what they're talking about
15:20
<annevk>
i'll deal with xhr feedback a bit later once it has stacked up some more
15:27
<annevk>
thanks to people doing copy and paste incorrectly HTML5 elements are now used in the wild: http://www.topxml.com/rbnews/XML/re-93299_HTML-5-and-a-different-kind-of-ruby-support.aspx
15:27
<Philip`>
<a rel="nofollow"udio>
15:29
<annevk>
nobody any idea on how to implement your custom [[Get]] and [[Set]] in JavaScript?
15:31
<annevk>
watch seems to have the same limitation as getters and setters
15:34
<annevk>
__noSuchMethod__ is closer
15:37
<annevk>
mu
15:38
<Dashiva>
annevk: As far as I know [[Get]] would require catchalls (as in es4) or doing it in the implementation itself
15:40
<annevk>
:(
15:40
<annevk>
i had this idea that maybe making a dataset impl would be feasible
15:40
<Dashiva>
If you're willing to be really really really hackish...
15:40
<annevk>
yes? :)
15:41
<Dashiva>
Create the object on every element with data attributes on dom complete, then use mutation events on the dom to catch changes to those attributes to update the dataset, and use getters and setters to bridge the two
15:42
<Dashiva>
I'm not sure if the middle step is possible, though
15:45
<Philip`>
Element.prototype.__defineGetter__('dataset', function () { alert('Just use getAttribute("data-..."), because that\'s trivial and perfectly compatible with everything') });
15:46
<Dashiva>
Now I'm curious
15:47
Dashiva
goes to read dataset spec
15:47
<annevk>
that doesn't work for .dataset.x = "test"
15:48
<annevk>
which would create a data-x attribute with the value 'test'
15:50
<Dashiva>
Yeah, [[Get]] only
15:51
<annevk>
:/
15:51
<Dashiva>
of course, you can take it one step further and run a manual check on the object with a setInterval, but that's just evil
15:52
<annevk>
I thought of that but that's a) not performing well and b) doesn't work if you set using dataset and then get using getAttribute() directly after
15:53
<Dashiva>
The b) case I don't really want to care about. They should make up their mind :)
15:54
<Philip`>
They should just use [sg]etAttribute all the time :-)
15:54
<Dashiva>
Catchalls would be so nice
15:54
<annevk>
at least for this
15:55
<annevk>
Philip`, stop being so pragmatic
15:55
<Dashiva>
How many browsers support mutation events anyhow?
15:55
<annevk>
most?
15:56
<Philip`>
Does IE6 support mutation events?
15:57
<Philip`>
If not, then it's not very useful for a legacy-UA compatibility script :-p
15:57
<Dashiva>
Eh, none of the IEs support getters and setters anyhow :)
15:57
<annevk>
the script was more about seeing if this could actually work
15:57
<annevk>
but it seems that it can't
15:58
<Dashiva>
Just make setters for all legal strings up to 16 characters
15:58
<Dashiva>
... on all nodes
16:00
<annevk>
that was another idea i had, but that didn't scale
16:00
<annevk>
guess i'm glad i didn't miss anything obvious :)
17:33
<gsnedders>
annevk: What should XHR.status return for an HTTP/0.9 response?
17:33
<gsnedders>
(that has no status code)
17:33
<annevk>
throw?
17:34
<gsnedders>
annevk: Throw INVALID_STATE_ERR?
17:34
<annevk>
that's what the spec currently says
17:36
<gsnedders>
It just seems odd
17:36
<annevk>
agreed
17:37
<gsnedders>
null is what I'd expect if it didn't exist and the object had a suitable state
17:37
<annevk>
or maybe just 0
17:38
<annevk>
or maybe HTTP should define what the status code is in that case :)
17:38
<annevk>
as in, if no status code is provided the status code is 200
17:38
<annevk>
that's prolly best
17:39
annevk
goes back to his semi-sleepy state
17:39
<gsnedders>
annevk: Less drugs! :P
17:42
<gsnedders>
othermaciej: Happy B'day!
17:43
<othermaciej>
thanks gsnedders
17:45
<MikeSmith>
standardssuck needs a logo
17:45
<MikeSmith>
maybe the guy who created the Squirrelfish logo can make one
17:45
<annevk>
congrats othermaciej
17:45
<annevk>
MikeSmith, hehe
17:45
<othermaciej>
congrats on what?
17:46
<annevk>
othermaciej, your birthday, it seems :)
17:46
<othermaciej>
oh
17:46
<othermaciej>
thanks
17:46
<othermaciej>
though I'm not sure it's all that impressive to survive to 0x20
17:46
<gsnedders>
annevk: I've just managed to get -9 returned by .status
17:47
<annevk>
othermaciej, :)
17:47
<annevk>
gsnedders, impressive
17:48
<gsnedders>
othermaciej: I hope it's your birthday, seeming it claims that on Facebook ;P
17:48
<gsnedders>
annevk: "HTTP /1.1 200 OK" does all kinds of odd things.
17:50
<gsnedders>
annevk: Including giving .status = -9 in IE
17:51
<annevk>
oh, IE has all kinds of weird stuff with .status
17:52
<MikeSmith>
gsnedders: please provide your assessment of the following
17:52
<MikeSmith>
http://www.flickr.com/photos/goopymart/2411605231/
17:52
<gsnedders>
I need to provide assessment? Meh.
17:52
<gsnedders>
MikeSmith: WTF?
17:52
<gsnedders>
(that's my assessment)
17:52
<othermaciej>
gsnedders: it is indeed
17:53
MikeSmith
finds fridge empty again, reduced to eating olives from jar
18:05
<gsnedders>
http://lists.w3.org/Archives/Public/ietf-http-wg/2008AprJun/0479.html — HTTP fun!
18:10
<annevk>
HTTP is just as broken as the rest of the Web, it's just noticed by less people than the brokenness of HTML/CSS
18:11
<annevk>
MikeSmith, true that, I'm going to do some shopping
18:14
<MikeSmith>
annevk: after I deplete this jar of olives, I think only thing left will be to turn to rock soup
18:15
<MikeSmith>
minus the rocks
18:33
<Hixie>
i'm amused at how dan and chris have just given up sending htmlwg status updates to the htmlcg list
18:45
<Dashiva>
Hixie: For lack of content, or lack of interest, or something else?
18:45
<Hixie>
no idea
18:45
<Hixie>
just haven't seen updates in a long time
19:00
<annevk>
maybe they don't attend the calls
19:00
<annevk>
or maybe HTML5 just doesn't need status updates, it's all over the place already anyway :)
19:01
<Dashiva>
Yeah, it's public all over the place in htmlwg
19:01
<Dashiva>
Except that cg :)
19:04
<annevk>
the HCG should really become a public group
19:05
<Hixie>
good luck wit hthat
19:06
<gsnedders>
HCG?
19:06
<annevk>
Hypertext Coordination Group also known as HTML CG
19:16
<Hixie>
so who is steven faulkner going to interview for standards suck 4?
19:20
<annevk>
there's no pattern
19:20
<annevk>
just a spoon
19:24
<gsnedders>
annevk: There is no spoon!
20:01
<Hixie>
i've been trying to avoid defining urls for so long
20:01
<Hixie>
but it's starting to end up at the top of all my automatically generated lists
20:01
<Hixie>
folder with oldest e-mail, folder with most e-mail, etc
20:04
<Philip`>
You just need to refine your automatic list generation - "folder with oldest e-mail that is not about defining URLs", etc
20:06
<Hixie>
i've already done that with any folder about WF2, rendering, video codecs...
20:06
<Hixie>
i don't really have a good reason to do it with urls!
20:06
<Hixie>
"i don't want to find out what browsers do" doesn't seem like a valid reason
20:06
<Hixie>
bbl
20:11
<annevk>
Hixie, adding HTML5 elements to the HTML parser has been suggested
20:11
<annevk>
Hixie, I think that would be a good thing too, fwiw
21:00
gsnedders
is far too indecisive to write a spec
21:01
<annevk>
if both options seem ok flip a coin
21:02
<annevk>
(and maybe document the choices in a comment)
21:02
<annevk>
then when people complain you can change it
21:03
<Philip`>
If both options seem okay, branch the spec and produce two versions
21:03
<Philip`>
and repeat the branching for each indecision point
21:04
<Philip`>
That way you won't be acting unfairly towards any of the options
21:06
<gsnedders>
In this case it's whether HTTP-Version should be case-sensitive (per HTTP.sys (in IIS) and CFNetwork (and through that Saf and other WebKit browsers on OS X) and RFC2616) or not (per Fx, IE, and Op)
21:06
<gsnedders>
It seems the case-insensitivity is no longer needed whatsoever
21:08
<gsnedders>
I'll send late LC comments on XHR this weekend, annevk, BTW
21:08
<gsnedders>
(I know I've been saying that for a while, but hopefully I finally will)
21:08
<annevk>
sure
21:12
Philip`
wonders how many invalid HTTP responses he encountered
21:12
<Philip`>
http://drmarin.galeon.com/ is one (returns just the body, no headers or status code or anything)
21:13
<Philip`>
and same on http://galeon.com/dilandau/mokushiroku.htm
21:16
<Philip`>
http://www.guadeloupe-fr.com/ too
21:18
<hsivonen>
Hixie: I second annevk's and zcorpan's suggestion to address new elements in parsing
21:18
<Philip`>
http://www.truemetal.org/powermetal/ too
21:19
<Philip`>
Two send "HTTP/1.x ..." lines twice (so the second time gets parsed as a header)
21:20
<Philip`>
I don't see any other interesting problems that HttpClient detected
21:23
<virtuelv>
did the svg wg come up with a proposal yet, btw?
21:23
<virtuelv>
for svg-in-html, that is
21:25
<Philip`>
virtuelv: Not yet, but I believe they have said they are working on it
21:46
<gsnedders>
Philip`: The ones without any status code or body are HTTP/0.9 servers
21:56
<Hixie>
right. new elements.
21:56
<annevk>
add some while you're at it :)
21:56
<annevk>
<kitchensink>
21:57
<gsnedders>
<bikeshed> too
22:05
<Hixie>
right, what are the actual elements that i should make sure i've added
22:05
<annevk>
event-source in <head>
22:05
<annevk>
command, source
22:05
<annevk>
datalist / dialog
22:05
<annevk>
various sectioning elements
22:05
<Hixie>
"event-source", "section", "nav", "article", "aside", "header", "footer", "datagrid", "command"
22:06
<Hixie>
"source", "datalist", "dialog"
22:06
<Hixie>
what does "datalist" need?
22:06
<annevk>
does <option> close without error?
22:07
<annevk>
and <optgroup> too maybe, forgot if that was allowed
22:07
<Hixie>
uh
22:07
<Hixie>
"option" gets ignored "in body"
22:07
<Hixie>
bummer
22:13
<Hixie>
ok the following get treated as "meta" does: "event-source", "command", "source"
22:13
<Hixie>
and the following get treated like "div": "section", "nav", "article", "aside", "header", "footer", "datagrid", "dialog"
22:14
<Hixie>
s/meta/link/
22:14
<annevk>
treating <source> like <param> seems better
22:15
<Hixie>
yes
22:16
<Hixie>
hm, <body><style scoped> gets parsed wrong
22:18
Hixie
makes <style> not allowed as a child of <body>
22:18
<annevk>
wtf?
22:18
<annevk>
i mean, why?
22:19
<Hixie>
because </head><style> has to be parsed as <head><style></head><body> not <head></head><body><style>
22:19
<Hixie>
and there's no real reason to have <style> be the first element of a <body> instead of having it in the head
22:20
<annevk>
</head><style> is already taken care of
22:20
<annevk>
see "after head"
22:21
<Hixie>
right, i'm not changing the parser
22:21
<annevk>
:/
22:22
<Hixie>
i just don't want it to be possible to write pages that are conforming but parse as something very different
22:22
<Hixie>
since that would be dumb
22:23
<annevk>
is this about </head><style> still?
22:23
<Hixie>
hm, i also need to make <script> not be allowed at the start of <body> for the same reason
22:23
<Hixie>
actually maybe i just need to make the syntax section clearer
22:23
<Hixie>
and say that the tag can't be omitted
22:24
<annevk>
oh, I see
22:24
annevk
thought Hixie was proposing changing the parser
22:24
<Hixie>
no no
22:25
<Hixie>
oh hey what do you know
22:25
<Hixie>
the writing section already covers this
22:27
<annevk>
heh, the IE people have discovered the DOM Core bug as well
22:29
<roc>
I think I pissed off someone at Adobe
22:31
<annevk>
jd hmm
22:31
<Dashiva>
roc: Your pdfs won't open anymore?
22:33
<othermaciej>
roc: why do you say that?
22:33
<othermaciej>
annevk: I proposed errataing that years ago, sigh
22:33
<roc>
"<em>"since they're not standards they probably won't ever be invited to this party."</em> uh-oh, that sounds a little closed and exclusionary there...! :("
22:34
<annevk>
so we need to actively invest time in making stuff work with proprietary platforms we don't believe in? classic
22:35
<Dashiva>
That's what they pay w3c for, isn't it?
22:35
<othermaciej>
as did you
22:35
<Hixie>
we don't, but it makes sense that they'd try to make us
22:35
<roc>
as a practical matter we do have to invest in plugins
22:36
<roc>
ooh
22:36
<roc>
what the Web needs: NPAPI5
22:36
<othermaciej>
well, AIR is a valiant effort on their part to mix Flash with real Web content in more interesting ways
22:36
<othermaciej>
but it's not actually available on the web
22:36
<othermaciej>
and they intentionally crippled the WebKit part of AIR and have not been upgrading it since
22:52
<annevk>
yet more MS delay
22:59
<Hixie>
oh jeez
22:59
<Hixie>
microsoft are writing a Paper
22:59
<Hixie>
so
22:59
<Hixie>
glad
23:00
<Hixie>
that you took xhr from me
23:00
<gsnedders>
We'll make annevk regret it!
23:05
<annevk>
makes you wonder what they're planning for html5
23:06
<Hixie>
oh they've tried hard to delay html5
23:06
<Hixie>
without success so far (except for delaying the start of the patent clock)
23:10
<Hixie>
crap, the parser reports an error for <!DOCTYPE HTML><div><li><rp><li></div>
23:12
<Hixie>
i guess <li> should generate implied end tags...?
23:12
<Hixie>
why doesn't it
23:12
<Hixie>
hm
23:13
<Hixie>
oh i see
23:13
<Hixie>
until i added <rp> and <rt> it wasn't necessary
23:13
<Hixie>
oops
23:15
<annevk>
wh is the parse error an issue?
23:21
<Hixie>
well, it's extraneous
23:21
<Hixie>
the error is the missing <ruby>
23:22
<Hixie>
not the missing </rp>
23:23
<annevk>
fair enough
23:55
<jgraham_>
http://james.html5.org/temp/html5lib-0.11.zip a second attempt at html5lib-0.11