02:35
<Hixie>
seems like css3 ruby can't describe html5 ruby (and IE ruby)
02:35
<Hixie>
:-(
02:38
<Hixie>
if anyone can come up with an ordering that makes sense for the subsections in 10.4 Self-contained features, let me know
08:38
<annevk>
hmm, CSS3 ruby can't describe HTML5 ruby? that'll give a rathole
08:38
<annevk>
apart from HTML5 ruby already being a potential rathole due to the existence of XHTML Ruby
08:44
<hsivonen>
Is it just me or do the Unicode folk have a pattern of first creating small and contained hacks that get accepted under the premise of not having a huge all-encompassing impact and then they point out interesting corner case interactions that would have huge impact if fully addressed?
08:47
<annevk>
it might be that since it is not their area they do not fully grasp the implications
08:47
<annevk>
if they do graps the implications however, the way they go around things is not really desirable
08:48
<hsivonen>
I mean, the Unicode was first sold as "just" expanding characters to 16 bits
08:48
<hsivonen>
and then later when you've expanded to 16 bits you are told that there are now surrogates and astral planes
08:49
<hsivonen>
and once you've dealt with astral planes, you are told that you need to perform normalization upon string compares
08:50
<hsivonen>
and the marginal benefit diminishes pretty significantly as the cost goes up
08:50
<annevk>
ooh, I thought you meant pointing out a small issue with Selectors that actually affects a whole lot more
08:50
<annevk>
to the extent that they even go asking consumers of the Selectors spec to point out the issue because Selectors has not changed yet
08:51
<annevk>
hsivonen, I'm sure it could've been worse :)
09:15
jgraham
is increasingly unsure that preserving LF in @title is the right thing to do
09:20
annevk
is increasingly more sure it's the right thing to do
09:23
<jgraham>
annevk: Contrarian
09:24
<annevk>
jgraham, looking that up, I'm not sure there's coventional wisdom or a consensus opinion here :p
09:26
<jgraham>
annevk: Well I guess, under the assumption that the spec represents the majority opinion, I am the contrarian.
09:26
<jgraham>
But still
09:27
<annevk>
rubys would slap you for that statement
09:27
<jgraham>
annevk: I know, that's why I said it :)
09:28
<annevk>
lastweekinhtml5 wannabe :p
09:30
<jgraham>
I donnu, if I was a lastweek wannabe, presumably I would have a poll for whether Hixie or MikeSmith smelt more like cheese or something.
09:30
<jgraham>
s/donnu/dunno/
09:47
<Hixie>
gecko has some really weird code
09:47
<Hixie>
like http://mxr.mozilla.org/mozilla-central/source/content/html/content/src/nsHTMLSharedElement.cpp
09:47
<Hixie>
wtf is with all the <spacer> stuff
09:54
<annevk>
Hixie, they tried to have NN4 compat for a while
09:56
<Hixie>
apparently they didn't try very _hard_...
10:06
<annevk>
at some point IE compat became more important
10:07
<roc>
fortunately we didn't try that very hard either
10:08
<annevk>
:)
10:15
<fearphage>
i have a test case that i'm trying to make sure is valid. could i bother anyone with that?
10:17
<annevk>
you can try
10:17
<annevk>
just paste a link
10:18
<fearphage>
http://files.myopera.com/fearphage/static/bugs.htm
10:18
<fearphage>
the one test that is yellow
10:18
<fearphage>
seemingly it doesn't apply unless it is served as xml... i want to make sure that is a true statement
10:19
<fearphage>
it'd take work for me to split them but i can if you want
10:20
<fearphage>
view the source and search for svg:script
10:20
<fearphage>
actually, here is the xml version failing http://files.myopera.com/fearphage/static/tc/svg-script.xhtml
10:21
<annevk>
that bug is marked fixed internally
10:22
<fearphage>
ok, i just want to know if i can test it without serving it as xml. http://files.myopera.com/fearphage/static/bugs.xhtml <-- xml mimetype
10:49
<annevk>
fearphage, how about generating everything through script?
10:50
<fearphage>
come again?
10:50
<fearphage>
generating the entire test case from the script?
10:52
<annevk>
if you do not want to use XML but still need namespaces and all in text/html you need to use script
10:53
<fearphage>
interesting. i'll have to look intot that
10:55
<Hixie>
man this rendering stuff is tedious
10:55
<roc>
I can't resist a UTF8 vs UTF16 war
10:56
<fearphage>
thanks annevk
10:56
<annevk>
roc, are you going to bring up the gzip argument?
10:56
annevk
has seen one of them before
10:58
<roc>
what argument is that?
10:59
<annevk>
that the byte savings UTF-16 gives for some languages are done away by using compression
10:59
<roc>
I'm just mentioning that we did some analysis showing UTF-8 is way more compact than UTF-16 for the DOM text of the front pages of some "top 20" CJK sites
11:00
<roc>
and Wikipedia articles too
11:00
<annevk>
I see, interesting
11:08
<roc>
The promulgation of UTF-16 is one of the dumbest, most wasteful things that's happened in software
11:40
<gsnedders>
At least it wasn't UTF-32.
11:42
<hsivonen>
roc: do you have an opinion on doing Unicode normalization on non-rendering layers in browsers?
11:43
<roc>
I know very little about Unicode normalization
11:44
<roc>
so, no
11:45
<hsivonen>
roc: ok. (I'm unsympathetic towards the idea of doing normalization on the parsing, DOM, JS or style system layers)
11:45
<roc>
I can see that
11:45
<Lachy>
hsivonen, do you understand why CC-by is not GPL compatible?
11:46
<hsivonen>
roc: ah. I now see you posted to the thread
11:46
<gsnedders>
What thread?
11:46
<annevk>
www-style
11:47
<gsnedders>
Ah
11:47
<gsnedders>
I haven't been reading www-style very closely recently
11:47
<hsivonen>
Lachy: for one thing, it has requirements about attribution that aren't in either version of GPL
11:48
<hsivonen>
Lachy: also, an issue in the CC-by 2.x vs. GPLv2 analysis was the anti-TPM clause
11:49
<Lachy>
sure, but I can't imagine why the mere requirement of attribution affects compatibility, especially since attribution is considered a moral right in some countries (though, I don't think it is in the US), which makes it effectively required anyway
11:49
<annevk>
roc, the main problem with UTF-8 is that the DOM and such now rely on 16bit representation :/
11:49
<Lachy>
what anti-TPM clause? in CC-by or GPL?
11:49
<hsivonen>
Lachy: that's a gray area in non-software works
11:49
<hsivonen>
Lachy: in CC-by
11:50
<Lachy>
oh, I wasn't aware of that
11:50
<Lachy>
personally, I don't like using CC licences anyway and would prefer we didn't do so for the spec
11:50
<hsivonen>
Lachy: IANAL, but IIRC the expectations of attribution for software are close to null even in moral rights regimes
11:51
jgraham
is bothered by the attribution clause in CC-by
11:51
<roc>
annevk: yes, that is the great tragedy of UTF-16
11:51
<hsivonen>
Lachy: also, one might argue that attribution in CC-by may go beyond statutory requirements
11:51
<roc>
it's baked into the DOM and JS and Java and C# and Windows, not to mention the code of Gecko and Webkit
11:52
<hsivonen>
also Cocoa
11:52
<roc>
yeah
11:52
jgraham
tries to avoid thinking about Python in case he gets annoyed again
11:52
<hsivonen>
which UTF did Python 3 pick?
11:53
<jgraham>
hsivonen: "UCS guess which build option was used this time"
11:53
<roc>
actually with a good deal of cleverness we could probably reimplement the browser using UTF-8 internally
11:53
hsivonen
mumbles something about fast string bridging in Jython, IronPython and PyObj-C
11:53
<roc>
and still have a fast charAt
11:53
<gsnedders>
hsivonen: It didn't.
11:53
<roc>
but it's not worth it
11:54
gsnedders
sighs
11:54
<gsnedders>
Python's Unicode support is sadly pretty good for native Unicode support
11:55
<gsnedders>
I mean, as we get more memory, the overhead of UTF-16 arguably becomes less important
11:56
<jgraham>
Do non-python 16bit string languages get things like the length of non-BMP strings right?
11:57
<hsivonen>
gsnedders: do you mean Python 3 retained the gross misfeature that is the compile time switch?
11:57
<gsnedders>
hsivonen: YEs
11:57
<jgraham>
hsivonen: Yes
11:57
<Philip`>
gsnedders: That would only be true if the amount of string data stored by programs didn't increase proportionally to the amount of memory we have
11:57
<gsnedders>
*Yes
11:57
<hsivonen>
gsnedders, jgraham: that's very sad
11:57
<gsnedders>
hsivonen: Very.
11:58
<gsnedders>
jgraham: 32-bit Python gets it right :P
11:58
<hsivonen>
jgraham: isn't the length worth caring about the number of code units anyway?
11:58
<jgraham>
http://mail.python.org/pipermail/python-dev/2008-July/080886.html
11:59
<hsivonen>
the number of characters is neither useful as a measure of storage nor as a measure of human-perceived units
11:59
<Philip`>
gsnedders: (e.g. web browsers used to have 16MB of space to use for one page, and now they have 2GB and everyone wants a hundred tabs open, so it's no good if memory usage per page increases)
11:59
<jgraham>
hsivonen: That's not my experience
12:00
<hsivonen>
jgraham: ok. perhaps I'm falling into the trap of strict Unicode correctness including corner cases myself here
12:00
<jgraham>
Like I have no use for doing some_string[4] and finding out that it's the first part of a surrogate pair
12:01
<Philip`>
What do you want some_string[4] to mean?
12:01
<hsivonen>
oh yeah, making Validator.nu astrally correct in Java was "fun"
12:02
<Philip`>
(and for what use case?)
12:02
<jgraham>
(I haven't read much of ww-style but I just tested some browsers and it seems that Webkit handles <span>o<span style="color:green;">̈</span> fine wh but other browsers generally do not)
12:05
<jgraham>
Philip`: I want it to never return half a surrogate pair. I don't mind it returning a combining character
12:06
<hsivonen>
jgraham: why?
12:06
<hsivonen>
jgraham: isn't splitting a grapheme annoying, too?
12:07
<Philip`>
Why do you need the ability to index the string like an array anyway?
12:08
<Hixie>
man i wish someone had volunteered to do this rendering section
12:08
<jgraham>
hsivonen: It's not clear to me how it is possible to design a sane api where graphemes are not split when they are in fact seperate in the underlying representation
12:09
<annevk>
Hixie, people volunteer to do the simple stuff (and then get bitten, see e.g. XMLHttpRequest)
12:09
<Hixie>
heh
12:09
<Hixie>
there's no simple stuff :-)
12:09
<Hixie>
not if you do it right
12:10
<jgraham>
hsivonen: Although maybe if you make everything very complicated and put lots of normalisation-related methods on each string you can do it
12:12
wilhelm
noticed.
12:21
<annevk>
bit unfortunate that :indeterminate was dropped from CSS just now HTML5 adds it
12:21
<Hixie>
heh
12:22
<hsivonen>
does CSS have a pseudo class that would be suitable for denoting an active caption per video clock?
12:24
<annevk>
you lost me in the end
12:25
<Lachy>
hsivonen, I don't understand what you mean
12:26
<jgraham>
I assume hsivonen means a :showing pseudo class that would match the caption currently being displayed from some list of captions or so
12:26
<hsivonen>
suppose you have a caption that you want to be visible from t_1 to t_2
12:27
<hsivonen>
and you'd do this by making the caption gain a pseudo class when the video is playing between t_1 and t_2
12:27
<hsivonen>
jgraham: that's what I mean
12:28
<annevk>
since the captions are likely to be part of the video resource, wouldn't something like video::caption be better?
12:29
<hsivonen>
annevk: as far as I can tell, security requires captions to live in a nested browsing context as far as the DOM cares
12:30
<annevk>
that sounds like an implementation detail
12:30
<hsivonen>
annevk: do you mean you'd style the captions by styling a pseudo element in the outer cascade?
12:31
Hixie
mumbles something about styling captions
12:31
<WebG-Alt>
Hi anne. Was this directed to me? - WebG
12:31
<hsivonen>
annevk: I'd expect what you are saying to be more complex than making the nested browsing context have its own cascade
12:31
<hsivonen>
annevk: but I'm not a style system implementor
12:33
<WebG-Alt>
Fair enough!
12:33
<annevk>
WebG-Alt, prolly not :)
12:34
<annevk>
hsivonen, I have no idea what you're on about; <video> does not have a nested browsing context; captions are not represented by elements in the DOM; what is a pseudo-class good for?
12:34
<WebG-Alt>
D'oh sorry. Thought message was for me. Sorry unfamiliar with IRC!
12:35
<hsivonen>
annevk: suppose a model where captions are rendered by the browser's CSS formatter in a nested browsing context that overlays the video frame
12:36
<gsnedders>
Anyone got a sane style sheet for a letter?
12:36
<hsivonen>
gsnedders: hendry has a Web service that might contain one
12:37
<annevk>
hsivonen, ok, how would a pseudo-class help?
12:37
<annevk>
hsivonen, would you have selectors that bleed through the browsing context boundary (as seen in <iframe seamless>, but not in any browser)
12:38
<hsivonen>
annevk: suppose the nested browsing context contains a DOM with captions for different times as siblings
12:38
<hsivonen>
annevk: and then the :showing (or whatever) pseudo class travels from one DOM item to another as playback time progresses
12:39
<jgraham>
(the obvious HTML would presumably be an <ol>)
12:39
<hsivonen>
jgraham: <div> :-)
12:39
<gsnedders>
Heck, how would you markup a letter in HTML?
12:39
<gsnedders>
HTML so isn't made for this :)
12:40
gsnedders
concludes LaTeX is probably better, again
12:40
<Philip`>
gsnedders: Use Word - it's made for this
12:41
<gsnedders>
Philip`: But LaTeX has the document class "letter"!
12:41
<annevk>
hsivonen, I think one draft had :show and :hide at one point, but I'm not sure how that would help
12:41
<gsnedders>
Philip`: And LaTeX > Word.
12:41
<Philip`>
gsnedders: And Word has a document template "Letter"
12:42
<jgraham>
gsnedders: Yes. But only in certian circumstances and if you are prepared to play free and easy with your sanity.
12:42
<hsivonen>
annevk: is :show generic enough that a vocabulary can give it a meaning without a lot of CSS WG process?
12:42
<annevk>
hsivonen, oh, I guess you want to style the caption file itself
12:43
<WebG-Alt>
This may be a crap suggestion, but what about just, e.g. dragging a .doc into say DreamWeaver and letting it autocreate the styles? Theyll be moronically named but it may do the trick?
12:43
<WebG-Alt>
For jgraham
12:44
<annevk>
hsivonen, can't find :show anymore
12:44
<annevk>
hsivonen, might have been :open and :closed from http://www.damowmow.com/temp/csswg/old/ui/dui.html
12:44
<jgraham>
WebG-Alt: I'm not trying to do anything. gsnedders is.
12:44
gsnedders
is the idiot, not jgraham
12:44
<hsivonen>
annevk: thanks
12:44
<WebG-Alt>
Sorry
12:46
<annevk>
hsivonen, http://www.w3.org/TR/css3-preslev/ comes to mind as well
12:48
<hsivonen>
I wonder how much process in involved in getting the CSS WG mint a pseudo class
12:49
<annevk>
there's lots of time involved
12:49
<annevk>
e.g. :scope still has not been discussed by the WG
12:49
<hsivonen>
annevk: that's what I suspected
12:49
<annevk>
took a long time to get http://www.w3.org/TR/becss/ published as well
12:51
<Hixie>
i don't remember that taking _that_ long
12:51
<annevk>
after it took several weeks longer than necessary to decide to publish, it took another coupld of months before it actually got published
12:52
<annevk>
the becss sample might be good to use for getting :scope through
12:52
<annevk>
though
12:54
<hendry>
gsnedders: letterly.com
12:54
<gsnedders>
hendry: thx
12:55
<hsivonen>
so if this pseudo class idea were pursued, should Mozilla just mint a :-moz-foo pseudo class and risk content coming to depend on it or how would things work out best?
12:56
<gsnedders>
hendry: Heh. Just uses XeTeX :)
12:56
<hsivonen>
the relevant discussion is on xiph's accessibility list, btw
12:56
<Hixie>
ok bed time
12:56
<Hixie>
nn
12:56
<gsnedders>
sleep tight, little boy :P
13:05
<annevk>
hmm, hsivonen, are you following the www-tag discussion on making rel take URLs?
13:06
annevk
is not sure how to say that making rel= significantly more complex without direct benefit is not such a great idea
13:06
<hsivonen>
annevk: not properly. I'm trying to escape the tyranny of email and get actual code written
13:06
annevk
can think of more important features to implement than resolving rel values as URLs before comparison
13:07
<hsivonen>
annevk: I did notice that I got CCs on the thread
13:07
<annevk>
hsivonen, more power to you then :)
13:10
<Philip`>
http://philip.html5.org/data/link-rel-rev.txt - some of those don't look like they're going to work too happily if you resolve them as URLs
13:15
<annevk>
the idea is also rather incompatible with relList at the moment
13:15
<hsivonen>
sigh. I can't even program simple BOM sniffing right
13:28
<jgraham>
Philip`: Apparentely the TAG don't have to worry about the real world or something
13:29
<annevk>
replied
13:29
<annevk>
we'll see what happens
13:29
<annevk>
I'm not really inclined to even ask for resources to implement such features; seems so far away from what everyday authors are asking for...
14:07
<annevk>
hsivonen, I'm curious, how far are you in making the HTML5 parser work in Gecko?
14:09
<hsivonen>
right now I'm trying to complete all the encoding stuff that I needed to reimplement with Gecko's APIs
14:09
<hsivonen>
and layout is still notified too often
14:46
<zcorpan>
gsnedders: anolis doesn't handle <dl><dt><dt> correctly
15:17
<jgraham>
Does anyone know of a simple js plotting library where I can give it a html table and say "plot column 1 against column 3"
15:20
<Lachy>
jgraham, plotkit
15:24
<jgraham>
Lachy: Oh, right, I missed that part of the docs. Thanks :)
15:43
gsnedders
guesses the parser doesn't, and Anolis does fine
15:43
<gsnedders>
Because Anolis doesn't do anything with dl/dt AFAIK
15:56
<yorick>
with the application cache, is there any way to access online data when online instead of updating the cache and then load it?
16:00
Philip`
wonders if anyone has made a blog commenting system where 'trusted' users (e.g. those who have posted non-spam comments in the past) can view the moderation queue of posts from untrusted users and approve them
16:08
bzed
thinks that should not be too hard to add that to ikiwiki
16:11
<annevk>
http://blogs.msdn.com/ie/archive/2009/02/02/birth-of-a-security-feature-clickjacking-defense.aspx I do not really agree with their notion of "open forum". They could have discussed this e.g. on various public mailing lists, but didn't.
16:16
<Lachy>
Philip`, that sounds like a good idea
16:16
<Lachy>
I don't know of any that do that
16:17
<Lachy>
I think the closest there is to it is probably the slashdot system where people are granted temporary rights to moderate comments
16:48
<dglazkov>
good morning, WebKit!
16:50
<Philip`>
dglazkov: Wrong channel? :-)
17:04
<dglazkov>
indeed
17:37
<gsnedders>
hsivonen: validator.nu often includes <http://c.validator.nu/all/>; which doesn'texist
18:10
<Philip`>
gsnedders: Hasn't XML Namespaces taught you that you shouldn't expect URIs to be dereferencable?
18:10
<gsnedders>
Philip`: :)
20:19
zcorpan
looks at http://html4all.org/HTML_Workgroup/tHTML_Parsing.html
20:22
<zcorpan>
it seems a script would only execute if it is in the form <script src=... />
20:24
<zcorpan>
<!doctype html><head><div> would insert the div in head
20:27
<zcorpan>
oh wait i was wrong about the script thing
20:31
<zcorpan>
<html relaxngdefinition=...>?
20:34
<zcorpan>
there's <handler> and <separator> and <h> and <l> but not <nl>
22:36
<Hixie>
gsnedders: my opinion is that i micromanage my biblio so much that any kind of automated system would actually be a net loss of productivity to me. :-)
22:43
<gsnedders>
Hixie: :)
22:53
<Lachy>
gsnedders, for now, I think you could just focus on getting support for biblio using an internal database or file containing the common RFC and W3C TR entries. Then you can look into providing custom entries after that's done
22:53
<gsnedders>
Lachy: I want opinions now! I've basically hidden this Anolis module as my computing project for school :)
22:54
<Lachy>
but when you do get around to doing custom biblio entries, it would be easiest if I could just specifiy a command line parameter with a file path, pointing to a file using the same format as whatever is used for the internal data file
23:27
<Hixie>
so long as pimpmyspec.net supports it...
23:27
<Hixie>
(assuming i'm going to use it)
23:28
<Dashiva>
pimpmyspec needs some an actually pimpin' options
23:28
<Dashiva>
-an
23:32
<Hixie>
anyone got IE handy?
23:32
<Hixie>
any version
23:32
<Hixie>
what does it do with http://www.hixie.ch/tests/adhoc/http/content-type/nph-030 ?
23:33
<Dashiva>
blue
23:33
<Dashiva>
IE6 win
23:33
<Dashiva>
*7
23:34
<Hixie>
thanks
23:37
<Hixie>
heycam: i've written every word of the html5 spec
23:37
<heycam>
Hixie, ok. so there has been at least one occasion where i remember you saying that you used text that i suggested. was it reworded instead?
23:37
<Hixie>
almost certainly
23:37
<heycam>
ok
23:38
<Hixie>
there may have been minor exceptions, like one or two lines or some such
23:38
<Hixie>
but good luck to anyone who wants to try their luck in court over that :-)
23:38
<heycam>
too small to be copyrightable, perhaps
23:38
<Hixie>
right
23:38
<Dashiva>
Like when someone tells you how to change the "relationship to x" parts :)
23:39
<Hixie>
if such lines were to be a problem, the examples would be a much bigger problem first
23:39
<Hixie>
(many of them quote things like monty python)
23:39
<Hixie>
(one even quotes part of the svg spec's copyright notice...)
23:51
<Lachy>
Hixie, first example in http://www.whatwg.org/specs/web-apps/current-work/#the-section-element copied verbatim from http://lists.whatwg.org/pipermail/help-whatwg.org/2008-December/000157.html
23:51
<Lachy>
I'll see you in court! ;-)
23:52
Dashiva
wonders how much HTML5 is in Lachy's text...
23:52
<Lachy>
Dashiva, shh!
23:52
<Dashiva>
Smells like settlement
23:52
<Hixie>
Lachy: opera owns your work, and opera is one of the copyright owners of html5, so good luck with that :-P
23:53
<Lachy>
ah, damnit
23:53
<Hixie>
this is not my first barbequeue :-P
23:55
<heycam>
barbecue?
23:55
<heycam>
bbq.enqueue(aSausage)
23:57
<Lachy>
heycam, the correct spelling is Barbeque. Barbecue is the American spelling
23:57
<heycam>
Lachy, ok thanks. i was tricked by gtk's spell checking text box. =)
23:58
<Hixie>
oops, i had a banana problem with my barbeque
23:58
<Hixie>
how embarassing
23:58
<sicking>
haha
23:58
<Lachy>
heycam, set your spell checker to en-AU instead of en-US
23:58
<heycam>
let's just call it a barbie and be done with it :)
23:59
heycam
wonders where he might do that
23:59
<Lachy>
heycam, careful, the non-Aussies here might think you're talking about a toy doll