00:08
<Philip`>
http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml doesn't work in Opera :-(
00:08
<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_."
00:09
<Philip`>
(XSLT, yay)
02:31
<Hixie>
did hsivonen mention if he was going to send an analysis of his data?
06:43
<BenMillard>
hello, is anyone here?
06:56
<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
06:57
<BenMillard>
so that's what I've been doing during the small hours of this morning, typing my reasoning as I go
06:58
<BenMillard>
since nobody is here, I'll go to sleep and return some time later...just thought this should be recorded somewhere
08:43
<annevk>
Hixie, no
10:35
<annevk>
html4all mailing list is funny
10:35
<annevk>
especially this recent debate between RB and LHS about what Hixie might have meant with "defining error handling"
11:36
<annevk>
http://groups.google.com/group/alt.html/msg/9d28ed5dc4c9fa36 "HTML 5 is a bunch of crazy egotists breaking stuff to no benefit."
11:38
<Lachy>
hah
11:43
<annevk>
these people see right through us, it's scary
11:43
<annevk>
in other news, base2 from Dean Edwards will support element.classList
12:34
<Lachy>
The XSLT void element thread is getting annoying.
12:36
Philip`
doesn't see why
12:46
<jgraham>
Lachy: FWIW I think Julian has a reasonable point
12:47
<jgraham>
I don't see the harm in allowing <eventsource></eventsource> as long as the closing tag immediatley follows the opening tag
12:47
<jgraham>
I would basically expect the parser to drop the closing tag
12:48
<jgraham>
So <eventsource>foo</eventsource> would be parsed like <eventsource>foo
12:49
jgraham
has wasted enough time on the html-wg today
12:51
<jgraham>
It would be different to </br> but not too different from <script> which requires a closing tag even when it is effectively void
12:52
<webben_>
jgraham: I thought a point was made that IE treats void element starting tags as self-closing?
12:53
<jgraham>
webben_: Maybe I haven't been following too closely. But I don't quite understand what you mean
12:54
<webben_>
actually, reading it again, I'm not sure what Robert meant
12:54
<webben_>
http://lists.w3.org/Archives/Public/public-html/2008Aug/0935.html
12:54
<webben_>
"IE treats unknown elements as void elements (adding the next element
12:54
<webben_>
as a next sibling)"
12:55
<webben_>
i'm not sure what that means in terms of how IE treats tags
13:02
<robburns>
hi webben
13:02
<robburns>
an example
13:02
<robburns>
<header>
13:02
<robburns>
some contents
13:02
<robburns>
</header>
13:02
<robburns>
In DOM as:
13:02
<robburns>
html
13:02
<robburns>
13:02
<robburns>
header
13:02
<robburns>
#text: some contents
13:02
<robburns>
/header
13:02
<robburns>
Instead of
13:02
<robburns>
html
13:03
<robburns>
13:03
<robburns>
header
13:03
<robburns>
#text: some contents
13:03
<robburns>
that's how IE will process an unknown non-void (HTML5 in this case) element
13:03
<robburns>
webbeb_
13:06
<robburns>
but on the other hand in Gecko
13:07
<robburns>
<p>
13:07
<robburns>
<eventsource>
13:07
<robburns>
some contents
13:07
<robburns>
</p>
13:07
<robburns>
html
13:07
<robburns>
13:07
<robburns>
p
13:07
<robburns>
eventsource
13:07
<robburns>
#text: contents
13:07
<robburns>
instead of html
13:07
<robburns>
13:07
<robburns>
p
13:07
<robburns>
eventsource
13:07
<robburns>
#text: contents
13:09
<Philip`>
(Actually it'd be called HEADER and /HEADER, not header and /header)
13:09
<Philip`>
(Wait, am I confused?)
13:10
<Philip`>
(At least I think IE does something funny with the case of unrecognised tags)
13:10
<robburns>
though in Firefox 3, the P element does not get implicitly closed (I think earlier versions would)
13:10
<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
13:10
<Philip`>
(Ah, right, it's if you do document.createElement('header') then the element is called "header" instead of "HEADER")
13:12
<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)
13:12
<robburns>
in IE that is
13:13
<robburns>
s/ text/html / tex/thml
13:14
<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
13:14
<Philip`>
but that's not particularly relevant to the point anyway
13:35
gsnedders
says hi to all the bureaucracy trolls
13:47
jgraham
wonders if that is supposed to include him
13:49
jgraham
doesn't see himself as trolling
13:50
<jgraham>
Since AFAICT the charter is designed to prevent exactly this
13:50
<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.
13:51
<Lachy>
also, allowing </foo> for new void elements, but not for previously existing void elements makes things inconsistent.
13:52
<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)
13:52
<jgraham>
If the problem is XSLT-only it's not worth worrying about
13:53
<jgraham>
But if it is more general, it seems bad to couple the serializer to the version of the language being output
13:54
<jgraham>
because it makes migration hard
13:54
<jgraham>
anyway I don't care very much
13:57
<Lachy>
so far, the problem is XSLT only. No other tool has been mentioned with the problem
13:58
<Lachy>
and I just don't want XSLT to have any impact on the development of HTML5
13:59
<webben_>
I think the key thing is to ensure that XML output by XSLT can be translated with FOSS tools to text/html HTML5.
14:00
<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.
14:01
<webben_>
e.g. <p>foo<blockquote><p>bar</p></blockquote></p>
14:07
<gsnedders>
webben_: It should do a backflip?
14:07
<gsnedders>
webben_: I mean, we already require UAs to take a deep breath in a certain case when parsing HTML
14:07
<Lachy>
webben_, you just have to accept that not all possible XML trees can be losslessly converted to HTML
14:08
<Lachy>
webben_, especially in cases like that which are invalid in XHTML too
14:10
<Lachy>
webben, incase you missed my messages just then due to connection issues, http://krijnhoetmer.nl/irc-logs/whatwg/20080831#l-149
14:10
<webben>
ta Lachy :)
14:11
<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.
14:12
<webben>
but the worry is not just that it's lossy, but that such a tool would properly want to handle it somehow
14:12
<webben>
at which point you end up with differing implementations.
14:12
<webben>
that's not invalid in XML serialization of HTML5.
14:13
<Lachy>
it would have to make the blockquote a sibling of the p instead, and possibly create a new p after it
14:13
<webben>
it could do that. it wouldn't _have_ to.
14:13
<Lachy>
so <p>foo<blockquote>bar</blockquote>baz</p> would become <p>foo</p><blockquote>bar</blockquote><p>baz</p>
14:13
<webben>
another approach, for example, would be to go lossy in a different manner
14:13
<webben>
by genericising the p to a div.
14:14
<webben>
or processing agents could attach a class to the second p in an attempt to indicate what's happened.
14:14
<Lachy>
so it would become this instead? <div>foo<blockquote>bar</blockquote>baz</div>
14:14
<webben>
could do.
14:15
<Lachy>
perhaps the right solution is for the tool to check with the user what he wants to do
14:15
<webben>
that simply passes the problem on
14:16
<Lachy>
so do you want the spec to define how to serialise it?
14:16
<webben>
I'd prefer that, yes.
14:16
<webben>
i'd prefer the spec to define how XML HTML5 should be converted to text/html HTML5.
14:16
<gsnedders>
and the other way too
14:16
<webben>
yep
14:17
<Philip`>
webben: Like http://www.whatwg.org/specs/web-apps/current-work/multipage/serializing.html#serializing ?
14:17
<gsnedders>
Henri sent something about it a while back
14:17
<gsnedders>
Like, what he does in the validator
14:17
Philip`
doesn't know whether that algorithm is inappropriate for this
14:19
<webben>
that appears to do it.
14:19
<webben>
I don't like how it does it, but at least it does it.
14:19
<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
14:21
<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"
14:22
<Lachy>
but that seems to be what Firefox does anyway
14:22
<webben>
does the algorithm not match the note?
14:22
<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
15:39
gsnedders
wonders what he's meant to know for physics test tomorrow
16:33
<Philip`>
gsnedders: Remember that gravity goes downwards
16:33
<Philip`>
That's probably enough for a few marks
17:20
<hsivonen>
Hixie: yes, I intend to follow up with more opinion and analysis regarding the validation results
18:22
<gsnedders>
Philip`: :)
18:22
<gsnedders>
Philip`: Actually, with what will be in the test, gravity may as well not exist
18:37
<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
18:38
<hsivonen>
Lachy: Macs draw a light blue screen when there's nothing drawing the desktop background
18:42
<hsivonen>
gsnedders: I don't have a mapping from non-conforming application/xhtml+xml onto text/html
18:42
<gsnedders>
hsivonen: I thought I saw an email from you about that a while ago
18:42
gsnedders
shrugs
18:43
<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
18:43
<hsivonen>
gsnedders: are you sure my email wasn't about going in the other direction than what webben asked about?
18:43
<gsnedders>
hsivonen: Myself and webben were saying both directions
18:43
<hsivonen>
ah ok.
18:43
<gsnedders>
hsivonen: So you have part of that
18:44
<hsivonen>
I have text/html to XML (covered by spec) and *conforming* application/xhtml+xml to text/html (not covered by the spec)
18:45
<hsivonen>
I'd love to see a sane and uncontroversial mapping from non-conforming application/xhtml+xml onto text/html
18:52
<gsnedders>
hsivonen: Where is text/html to XML covered in the spec?
19:03
<hsivonen>
gsnedders: http://www.whatwg.org/specs/web-apps/current-work/#coercing
19:04
<gsnedders>
Hixie: typo: mutatiosn
19:10
<hsivonen>
so now Google is saying Android supports H.264
19:10
<hsivonen>
what's the licensing story?
19:51
<Lachy>
hsivonen, where did you find that Android supports h.264?
19:52
<hsivonen>
Lachy: http://code.google.com/android/what-is-android.html
19:53
<Lachy>
Android uses the Apache2 licence
19:53
<hsivonen>
Lachy: it has patent language
19:53
<hsivonen>
how does that work with MPEG-LA?
19:54
<Lachy>
I don't know. What does the apache licence say about patents/
19:54
<Lachy>
?
19:54
<Dashiva>
It says "based on PacketVideo's OpenCORE" whatever that is
19:55
<hsivonen>
Lachy: hmm. actually, it only talks about patents that are licensable by a Contributor
19:58
<Lachy>
it's possible that Andriod could include closed source codecs for h264, and other patent encumbered stuff
20:03
<gsnedders>
Probably can be disabled in the build, and can be used only by those with licenses
20:15
<hsivonen>
hmm. the Microsoft EOT marketing partner Ascender made the Droid fonts
20:22
<hsivonen>
boohoo. Droid isn't Free as in Free Softwarle
20:22
<hsivonen>
Software even
20:25
gsnedders
is feeling that Sunday evening feeling: he ought to do homework for tomorrow.
21:13
<hsivonen>
http://twitter.com/stevefaulkner/statuses/902493395
21:22
<othermaciej>
hsivonen: self-fulfilling prophecy!
21:54
<gsnedders>
Hixie: You have any thoughts on making the spec-gen able to split a spec?
22:00
gsnedders
blogs http://gsnedders.com/anolis
22:10
<Philip`>
gsnedders: Why should that be part of the spec-gen, rather than a separate tool?
22:12
<Philip`>
gsnedders: s/it's/its/ :-p
22:14
<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
22:16
<gsnedders>
Philip`: The alternative is just to call it the revision hash :P
22:17
<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
22:17
<gsnedders>
Philip`: :P
22:17
<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
22:18
gsnedders
is too lazy :P
22:20
<Philip`>
gsnedders: But you're not too lazy to go to all the effort of having Release Candidates and documentation and everything? :-)
22:21
<gsnedders>
Philip`: RCs are no effort. `hg tag 1.0RC1` and wait for people to bitch about a bug
22:21
<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
22:22
Philip`
wonders why anyone would want to do that
22:22
<gsnedders>
Philip`: Well I did it to the CSS3 Module Postprocessor, didn't I?
22:23
<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
22:23
<gsnedders>
Philip`: I could look at some of the code.
22:24
<Philip`>
Did you?
22:24
<gsnedders>
Yes, then couldn't work out how it dealt with some edge cases how it did, so went back to reverse-engineering it
22:24
<gsnedders>
(the issue really being my knowledge of C)
22:24
<Philip`>
Oh, okay
22:25
<Philip`>
I suppose if you don't like reading the code then that'd be a reason to copy it without reading the code
22:26
<gsnedders>
I could've looked at the C spec, but I thought it'd be just as easy to create a test case :P
22:30
<Philip`>
I think a book on C would be somewhat more useful than the C spec
22:31
<Philip`>
(unless you were trying to implement a C compiler yourself)
22:31
<Philip`>
(which is not entirely recommended)
22:32
<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)
22:43
<jcranmer>
well, it is an ISO spec
22:43
<jcranmer>
but it's not unreadable
22:44
<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...
22:45
<hsivonen>
ISO drafting rules suck
22:45
<hsivonen>
is the C spec available online?
22:45
<hsivonen>
I'm interested in seeing an ISO spec that is not unreadable
22:46
hsivonen
googles
22:46
<jcranmer>
you have to pay for the specs, but any drafts you happen along are free :-)
22:46
jcranmer
points to the draft C++0x spec sitting on his hard drive
22:47
<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
22:48
gsnedders
has a draft C99
22:48
<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
22:49
<jcranmer>
you have to know how everything lays out at a high level before you dive into a spe
22:49
<jcranmer>
spec
22:50
<hsivonen>
wow. the ISO C spec actually looks more readable than a run-of-the-mill ISO spec
22:51
<jcranmer>
ISO C++ is fairly approachable, if you use <Ctrl>-F
22:54
<gsnedders>
Find makes most docs readable
22:54
<jcranmer>
not the OOXML spec!
22:54
<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)?
22:57
<hsivonen>
the points made about hyatt's mythicalness are a bit repetitive
22:58
<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.
22:58
<Philip`>
But they seem true
22:58
<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
22:58
<Hixie>
so it's bad
22:58
<jcranmer>
you'd be surprised at how bad people are at spelling
22:59
<hsivonen>
Hixie: the validator helps people spell it right :-)
22:59
<jcranmer>
I have a nice, phonetic last name and 85+% of people manage to both misspell and mispronounce it...
22:59
<Hixie>
hsivonen: apparently not enough people :-)
23:00
<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
23:02
<hsivonen>
Hixie: so if type=text/javascript, language=JavaScript and langauge=JavaScript are no-ops, why is allowing language=JavaScript a problem?
23:03
<Philip`>
Hixie: If many people were misled that way, wouldn't <embed>x</embed> be much more common than it is?
23:03
<Philip`>
jcranmer: I've never had such problems with my last name, but I often do with my first name :-(
23:03
<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.
23:04
<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.
23:05
<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?
23:05
<Philip`>
Hixie: I've only seen ones containing whitespace, &nbsp;, <br>, etc, and <noembed>
23:05
jcranmer
notes that hsivonen himself (herself?) misspelled 'language'
23:06
<hsivonen>
jcranmer: I misspelled it once on purpose above
23:06
<Hixie>
Philip`: i think we need a better transition story
23:06
<hsivonen>
jcranmer: did I misspell it also accidentally?
23:06
<jcranmer>
if you did it on purpose, then no
23:06
<hsivonen>
Hixie: why isn't type=text/javascript non-conforming?
23:07
<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.
23:08
<Hixie>
brb
23:19
<Philip`>
Hmm, I think IE8b2 never renders pages in IE7-mode unless explicitly told to, regardless of doctype
23:20
<Philip`>
i.e. pages that used to be quirks mode are IE5-mode, and pages that used to be standards mode are IE8-mode
23:20
<gsnedders>
That's what IE8b1 did :P
23:21
<Philip`>
I thought it made recognised standards-mode doctypes trigger IE7 mode
23:21
<Philip`>
Wait, I'm probably confused
23:21
<Philip`>
That's what it was going to do before they changed it
23:21
<Philip`>
I think
23:22
<gsnedders>
They changed it before b1
23:24
<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
23:25
<gsnedders>
heh
23:30
<roc>
yeah
23:30
<roc>
it's a very deep hole the IE team is digging
23:31
<gsnedders>
roc: What's the escape velocity?
23:31
<roc>
and the really bad news is that with every release it will get a lot worse
23:32
<roc>
there is no escape :-)
23:32
<gsnedders>
So above 3*10^8?
23:32
<gsnedders>
(ms^{-1})
23:32
<roc>
there's always Hawking raditation
23:32
<roc>
but that's not a fun way to travel
23:33
<gsnedders>
Hawking radiation?
23:33
gsnedders
notes his physics has limits
23:33
<roc>
so does mine, so I refer you to the Internet
23:34
<gsnedders>
What's that?
23:34
<Philip`>
It's fun how IE8's script debugger shows that element collection objects have properties named "length" and "ie8_length"
23:34
<Philip`>
and some other type of object has properties "item", "namedItem", "ie8_item" and "ie8_namedItem"
23:35
<roc>
heh
23:35
<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
23:36
<Philip`>
Please don't ask me why it's the antimatter particle and not the matter particle that falls into the black hole
23:39
<gsnedders>
why is it the antimatter particle and not the matter particle that falls into the black hole?
23:40
<Dashiva>
If only Philip` had used RDFa in his IRC text, we could make a machine reason out the answer for us
23:40
<Philip`>
Hmm, Wikipedia effectively says it's to conserve energy
23:41
<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)
23:44
<annevk2>
gsnedders, I commented on your blog, not sure it had effect
23:49
jgraham
has notion that the particle/antiparticle thing is a ort of lie to children
23:49
<Philip`>
By "children" you mean "physics undergraduates"? ;-)
23:49
<BenMillard>
jgraham, I see the headers thread continued along the lines of Process instead of facts
23:56
<jgraham>
BenMillard: Yeah. It has become quite soul destroying. This may be partially my fault, I don't know.
23:57
<jgraham>
Philip`: Well I actually meant "people who read popular science books". Physics undergrads don't learn about hawking radiation