00:02
<sayrer>
I love that no one has explained why font is bad
00:03
<sayrer>
"you might be able to create inaccessible websites"
00:03
<sayrer>
still true with html5
00:03
<sayrer>
don't let that stop you
03:57
<ezyang>
Gah, more changes :-)
09:51
jgraham
notes he did try to explain some problems with <font>
14:05
<gsnedders>
archtech: The actual parsing of entities we can't really change (because of need for compat. with existing content), what we can change is what is conforming, and making & legal on its own gets fairly complex.
14:05
<archtech>
gsnedders, the format of entities is that above today
14:06
<annevk42>
e.g. &copy= does turn into ©=
14:06
<gsnedders>
archtech: It isn't. Things like &copy do parse like that, and need to for compatibility, even though they are non-conforming.
14:07
<archtech>
gsnedders, not every non-conforming change should be supported
14:07
<gsnedders>
archtech: Browser vendors will not make changes to parsing that breaks existing content.
14:08
<gsnedders>
(And that is absolutely non-negotiable: if the spec changes to something incompatible, we'll just break the spec)
14:08
<archtech>
gsnedders, does that mean <b> foo <i> bar </i> baz </b> is part of HTML5
14:08
<archtech>
ops
14:08
<archtech>
I mean b i b i
14:08
<gsnedders>
archtech: It has a defined parse tree. It is non-conforming.
14:08
<gsnedders>
archtech: Any character stream has a defined parse tree.
14:09
<archtech>
So is that valid HTML5?
14:09
<gsnedders>
archtech: No, it is non-conforming.
14:09
<archtech>
gsnedders, but &copy= is?
14:09
<gsnedders>
archtech: No, that is non-conforming too.
14:09
<archtech>
I see.
14:10
<archtech>
So HTML5 defines the exact behavior of non-conforming code.
14:10
<gsnedders>
Yes.
14:10
<archtech>
And & as a literal is supported, but non-conforming.
14:10
<archtech>
For example: M & M
14:11
<gsnedders>
archtech: That is legal because a whitespace character follows it
14:12
<archtech>
I see.
14:12
<archtech>
Thanks for the info, guys
14:13
<gsnedders>
archtech: The only place you're allowed a & followed by anything within script and style
14:17
<archtech>
gsnedders, sure
14:17
<archtech>
those are defined as literal blocks terminated by strictly </script> and </style> I suppose
14:17
<archtech>
this is their de-facto behavior.
14:17
<gsnedders>
Yeah
14:17
<archtech>
nice
14:17
<gsnedders>
(case insensitively)
14:17
<archtech>
oh yea
14:20
<archtech>
I gotta check how html5 deals with textarea
14:20
<archtech>
;)
14:20
<archtech>
entities are supported.. but nothing else :)
14:21
<archtech>
curious one
14:21
gsnedders
is kinda running around, but can probably answer most of these questions :)
15:49
<Philip`>
archtech: Actually they're terminated by "</script" and "</style" (case-insensitively) followed by whitespace or "/" or ">" or EOF
15:53
<gsnedders>
Philip`: Not EOF, I thought, after the latest change.
15:54
<gsnedders>
Philip`: Well, I guess the contents to remain, we just drop the end tag
15:58
<Philip`>
gsnedders: I'm just going by what http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#close-tag-open-state says, for look-ahead recognition of the end tag
15:58
<gsnedders>
Philip`: You're right.
15:58
<gsnedders>
Philip`: Ignore me, I'm ignorant.
16:00
<Philip`>
annevk42: Having not thought about it very much, I dislike the idea of NaCl because it provides a significant drawback to non-x86 CPUs
16:00
<gsnedders>
NaCI?
16:01
<Philip`>
(and currently you can use non-x86 CPUs reasonably happy if you're running open source software and web stuff and not Flash, so if lots of people used NaCl then it'd have a significant effect on that)
16:01
<Philip`>
s/happy/happily/
16:02
<Philip`>
gsnedders: http://code.google.com/p/nativeclient/
16:02
Philip`
suggests gsnedders gets a font where l and I look different :-p
16:02
<gsnedders>
ah
16:03
<gsnedders>
They do, they have slightly different heights :P
16:06
<Philip`>
(You could emulate x86 on other ISAs but it's going to be pretty rubbish at performance, particularly if the code uses SSE, and it's marketed for computation-intensive processes)
16:06
<Philip`>
s/it's/NaCl is/
16:10
<annevk42>
can someone run http://xmlhttprequest.testsuite.org/dump/2009/status/001.htm in IE8?
16:11
<annevk42>
same for http://xmlhttprequest.testsuite.org/dump/2009/status/statustext.htm
16:12
<Philip`>
annevk42: "1xundefined,1xundefined,2x200,3x200,4x200" on 001.htm
16:12
<itpastorn>
FFox: 1xundefined,1xundefined,2xOK,3xOK,4xOK
16:12
<itpastorn>
MSIE 8: 1xundefined,1xundefined,2xundefined,3x200,4x200
16:13
<Philip`>
annevk42: "1xundefined,1xundefined,2xOK,3xOK,4xOK" on statustext.htm
16:14
<itpastorn>
My MSIE result was from 001.htm
16:15
<annevk42>
your MSIE results are different :/
16:15
<itpastorn>
Let me try again
16:16
<Philip`>
I definitely get "2x200" on 001.htm
16:16
<annevk42>
I like your result better :)
16:16
<Philip`>
not in compatibility view
16:16
<Philip`>
(I think)
16:17
<annevk42>
oh wait
16:18
<itpastorn>
1xundefined,1xundefined,2x200,3x200,4x200 in non compat view. 1xundefined,1xundefined,2x200,3x200,4x200 in compat view
16:18
<annevk42>
cheers, can you run them again? I made a small change to see where that undefined comes from
16:19
<itpastorn>
1x[object Error],1x[object Error],2x200,3x200,4x200 Compat view
16:19
<annevk42>
cool
16:20
<itpastorn>
1xError: Odefinierat fel.,1xError: Odefinierat fel.,2x200,3x200,4x200 Non-compat view (Swedish for "undefined error")
16:20
<annevk42>
seems everyone is pretty much the same then apart from Opera :/
16:21
<annevk42>
and only WebKit throws the correct exception
17:48
<archtech>
Philip`, so this would work? </script/
17:48
<archtech>
Interesting.
17:49
<archtech>
I only now realise how painful it is to reverse engineer a standard from a bunch of shoddy implementations ;)
17:49
<Philip`>
archtech: I don't know whether there's something to throw away incomplete tags and throw away unclosed script elements, in which case you might need </script/> instead
17:50
<Philip`>
(I remember some discussion about those issues a while ago but don't know what the spec says now)
17:50
<archtech>
Philip`, I have a slight complaint about the content models.
17:50
<archtech>
HTML4 seems to define those more clearly, and they match better what display property in CSS exposes
17:51
<archtech>
Also textarea has content model "text" which is not covered anywhere in the content model section it seems.
17:51
<Philip`>
archtech: You should probably complain to Hixie :-)
17:51
<archtech>
Hehe :)
17:51
<Philip`>
archtech: Also, are you looking at the "HTML 5: The Markup Language" document, rather than the main spec?
17:51
<Philip`>
in which case "text" is probably a RELAX NG type or something
17:52
<archtech>
Philip`, I think I'm looking at the last published spec
17:53
Philip`
actually looks at the spec
17:53
<Philip`>
Oh, yes, it says "Content model: Text."
17:53
<archtech>
Actually I'm looking at "W3C Working Draft 23 April 2009"
17:53
<archtech>
Here: http://www.w3.org/TR/html5/
17:53
<archtech>
Check section 3.4 Content models
17:53
<archtech>
There's no "text" in there :P
17:53
<archtech>
I'm not sure also why phrasing/heading content are differentiated?
17:54
<archtech>
Headings seem to behave mostly like paragraphs.
17:54
<Philip`>
(It's probably better to look at http://whatwg.org/html5 or http://www.w3.org/html/wg/html5/ to get the latest version)
17:54
<archtech>
Ah ok./
17:55
Philip`
doesn't really know anything about content models
17:55
<annevk42>
is "#PCDATA" really that much clearer than "Text"?
17:56
<archtech>
annevk2, hehe. There's link and PCDATA is defined ;)
17:57
<annevk42>
there's no link for me under http://www.w3.org/TR/html4/interact/forms.html#h-17.7
17:57
<annevk42>
I agree though that "Text" needs a definition
17:58
<archtech>
:)
17:59
<gsnedders>
archtech: CDATA can end with the </ + tagname followed by tab, lf, ff, space, >, /, or EOF
18:01
<annevk42>
that's not an exhaustive definition
18:01
<annevk42>
e.g. it does not cover </script foobar>
18:02
<Philip`>
Why doesn't it cover that case? That's the tag name followed by space, and it ends CDATA
18:02
<Philip`>
(unless I'm hopelessly mistaken)
18:03
<annevk42>
archtech, oh, Text is defined in http://www.whatwg.org/specs/web-apps/current-work/#text-0
18:03
<gsnedders>
annevk2: it's followed by space, so it does cover that
18:03
<gsnedders>
annevk2: (The fact that the space is then followed by other stuff is beside the point :P)
18:03
<annevk42>
I see..
18:04
<annevk42>
archtech, but I'd still email if I were you because a) it isn't linked and b) it's defined in a section specific to HTML
18:57
<Philip`>
If I wrote something like "The smallest positive numbers are 1, 2, 3, <abbr title='4, 5, 6, ...'>...</abbr>", so you can (partially) expand the "..." text via the tooltip (or equivalent in your independent medium), would I be shot for violations against semantics or is it fine?
19:31
<jjlee>
is it a known issue that html5lib trunk's html5lib.serializer.htmlserializer is broken?
19:32
<jjlee>
seems somebody changed string constants into numeric ones
19:34
<jjlee>
I should say, branch "default", I think (I'm new to hg)
19:35
<jjlee>
hmm, runtests.sh fails with SyntaxError. Presumably that *is* a known issue.
19:36
<jjlee>
in which case, I guess there are two reasons why http://code.google.com/p/html5lib/ shouldn't say "The SVN trunk is currently rather stable in anticipation of 0.12"
19:36
<jjlee>
first, it's not stable
19:36
<jjlee>
second, it's not in SVN any more (I assume)
19:40
<annevk2>
hmm, jgraham would know more
21:55
<Lachy_>
gsnedders, re twitter, what exactly are you expecting Audi to win or lose? is it a race or something? http://twitter.com/gsnedders/status/2157634810
21:55
<gsnedders>
Lachy_: 24 Heures du Mans
21:56
<gsnedders>
Lachy_: They have won every year from 2000 to 2008, except for 2003 which they did not enter.
21:57
<Lachy_>
is that the french name for 24 Hours of Le Mans
21:57
<gsnedders>
Lachy_: Oui
21:58
<Lachy_>
I thought Le Mans was already French. I wonder why they say "du Mans" instead
21:58
<gsnedders>
"le" is the masculine form of the definite article
21:59
<Lachy_>
ok
21:59
<gsnedders>
For masculine "of", "du" is used, which means either "of" or "of the"
21:59
<gsnedders>
So the "le" vanishes
22:00
<Lachy_>
anyway, I've not seen that race before. What kind of track do they race on?
22:01
<Lachy_>
found it http://en.wikipedia.org/wiki/24_Hours_of_Le_Mans#The_circuit
22:02
<gsnedders>
Lachy_: A high speed one.
22:02
<gsnedders>
Lachy_: 24 Heures du Mans
22:02
<gsnedders>
Lachy_: Oh, that on my clip board?
22:02
<gsnedders>
Lachy_: http://en.wikipedia.org/wiki/Circuit_de_la_Sarthe
22:02
<gsnedders>
Lachy_: There we go, that's what I meant to paste :P
22:05
<Lachy>
ok, so it seems similar to the Bathurst 24 Hour, except it
22:05
<Lachy>
it's been running for a lot longer
22:05
<gsnedders>
A lot quicker cars :)
22:06
<Lachy>
Are you saying we drive slow race cars in Aus?
22:06
<gsnedders>
I think it's true that the quickest cars at Bathurst are the slowest at Le Mans
22:07
<Lachy>
do you know about the Bathurst race?
22:07
<gsnedders>
a bit
22:07
<Lachy>
wow, you're the first non-Australian I know who's heard of it
22:08
<gsnedders>
hehe
22:14
<annevk2>
per wikipedia Bathurst 24 Hour only ran for years
22:14
<annevk2>
two years*
22:15
<gsnedders>
Lachy: (We're talking that the top cars have over 700 hp/tonne)
22:15
<gsnedders>
annevk2: That's untrue
22:15
<gsnedders>
No, I'm being dumb. It's 12 hour and 1000km races it's known for.
22:17
<gsnedders>
(The race that became the 1000km race goes back to 1963)
22:20
jgraham
is disturbed that there seems to be a useful productive discussion about <video> going on
22:21
<jgraham>
That violates one of the unshakable pillars of my existance
22:22
<jgraham>
jjlee: It is known that the html5lib trunk is somewhat broken
22:22
<annevk2>
that sounds a bit dramatic jgraham
22:23
<jgraham>
I merged in support for namespaced content in the main treebuilder and it has broken all the bits that I didn't fix on the branch e.g. all the stuff that relies on treewalkers
22:23
<jgraham>
annevk2: Really?
22:25
<annevk2>
your existance is in part based and in fact relies on unproductive discussions about <video>?
22:25
<jgraham>
jjlee: So I should fix it soon I guess. Sorry for the inconvenience
22:27
<jgraham>
annevk2: It's more that it invalidated my theory that it was no longer possible to have a sensible conversation about any remotely controversial HTML-related topic without one or more of the participants acting like a complete arsehat
22:28
<jgraham>
So I don't know if this says good things about Mike Shaver and Chris DiBona or bad things about everyone else
22:29
<annevk2>
lets go for somewhere inbetween
22:30
<jgraham>
I was going for "both" which was a bit depressing becuase I'm in the set of everyone else
22:53
<jjlee>
jgraham: np, but if development works that way, maybe it's easier to change that sentence on the web page than to change your style of development :-)
22:59
<jgraham>
jjlee: Will do; when the sentnce was written it was true. The idea of merging the branch was to fix all the issues rather than just let it sit but that hasn't really happened yet
23:02
<jjlee>
sure. I live in a glass house, shouldn't throw stones
23:16
<jgraham>
jjlee: Thanks for complaining, I am much more likely to work on something if it seems like someone actually needs it o work
23:17
<jjlee>
jgraham: any chance of a patch to expose character indices of tokens being accepted?