00:00
<TabAtkins>
I need those soundtracks. MGS music is awesome.
02:38
<Dashiva>
othermaciej: "We will not do any numerical counting all the votes." ... "To the extent we apply any numerical analysis to the results"
02:40
<othermaciej>
Dashiva: belt and suspenders!
02:40
<othermaciej>
or were you pointing out my typo?
02:40
<Dashiva>
No, just wondering which one is closer to reality :)
02:41
<othermaciej>
the former
02:42
<othermaciej>
the latter I stated in case anyone gets worked up about, say, all 10 of the opera reps responding
02:43
<Dashiva>
The vast Opera-wing conspiracy
02:44
<Dashiva>
But okay, answer accepted
04:42
<erlehmann>
tach
05:54
<bfulgham_>
Does anyone know if there is a proposal for animating transitions on visibility so that 'hidden' could accordion the surround blocks?
05:55
<bfulgham_>
Or rather, animating on "display" to allow accordion or other effects?
06:08
<smaug>
bfulgham_: you're talking about css transitions? That discussion should happen on www-style⊙wo
06:08
<bfulgham_>
smaug: thanks
08:55
<Philip`>
http://software.intel.com/en-us/articles/xml-parsing-accelerator-with-intel-streaming-simd-extensions-4-intel-sse4/ - intriguing
08:55
<Philip`>
html5lib would be much faster if it was rewritten to use SSE assembly instead of Python
08:57
<annevk2>
do it!
08:57
<annevk2>
the problem is of course that the code will only run on a proprietary processor
08:58
Philip`
wonders what a non-proprietary processor would be
08:59
<annevk2>
one with a standardized programming API?
08:59
<annevk2>
I suppose it might still be proprietary in that case
09:00
<Philip`>
Is there such a thing that exists today and is used?
09:01
<Philip`>
Everyone just uses x86 and copies Intel (except occasionally when Intel copies AMD) so it seems the closest thing to a standard :-)
09:15
<Hixie>
http://www.google.com/search?hl=en&safe=off&esrch=RTSearch&gl=us&tbo=1&tbs=mbl:1&q=html5&aq=f&oq=&aqi=g10
09:15
<Hixie>
woo
10:44
<annevk2>
Hixie, so you wanna combine the UI for streaming video/audio with geolocation, etc.?
10:44
<annevk2>
and most other things pages can ask for going forward, such as e.g. digital camera harddrive
10:44
<Hixie>
only if that is the best solution
10:45
<Hixie>
i don't know what the best solution is
10:45
<annevk2>
how would you combine that with the existing API?
10:45
<annevk2>
yeah I guess
10:45
<annevk2>
meh
10:45
<Hixie>
i wasn't really thinking of geolocation
10:45
<Hixie>
did tab update his proposal?
10:46
<annevk2>
for microdata? haven't seen anything
10:48
<nessy>
ohh! microdata! I wonder how that combines with video metadata
11:45
<Hixie>
what time is the htmlwg telecon these days?
11:47
<AryehGregor>
How is the telecon conducted? IRC?
11:47
<jgraham>
Telephone
11:47
<AryehGregor>
Amazing.
11:47
<jgraham>
Hence the "tele"
11:47
<AryehGregor>
"tele" means "distant".
11:47
<AryehGregor>
You know, like "telecommunications", and "teleport".
11:47
<jgraham>
Only in greek (or something)
11:48
<AryehGregor>
Okay, anyway. I guess Hixie will have to borrow someone's phone. :)
11:48
<gsnedders>
television is vision over telephone?
11:48
<jgraham>
Gah
11:48
<jgraham>
Hixie: 6pm CET on a Thursday
11:48
<AryehGregor>
Telepathy is detecting thoughts via telephone.
11:48
<Hixie>
jgraham: thanks
11:48
Hixie
changes to CET
11:49
jgraham
has never heard of a telecon that was not conducted by telephone
11:49
Hixie
changes back to PDT
11:50
<Philip`>
It's 1700Z according to two out of three repetitions of the time in the last reminder email
11:51
<Hixie>
figured i should at least be aware that i'm missing it this week since it's plausible i'll be awake at that time
11:51
<Hixie>
(though not near a phone unless i get some meeting room booked)
11:52
<AryehGregor>
Teleconferences can also be by video.
11:53
<Hixie>
we call those video conferences
11:53
<Hixie>
and they're much better than telecons though still almost as much of a waste of time
11:56
<AryehGregor>
What's wrong with IRC?
11:57
<Hixie>
or e-mail?
11:57
<Hixie>
beats me
11:57
<AryehGregor>
E-mail is a lot slower, it has definite disadvantages.
11:57
<AryehGregor>
Although also advantages.
11:57
<Hixie>
e-mail is far faster at making progress than telecons
11:57
<Hixie>
than scheduled telecons, i should say
11:58
<Hixie>
scheduled telecons cause working groups to only work immediately before and during a telecon
11:58
<Hixie>
so it becomes impossible for the group to make decisions except during the meeting
11:58
<Hixie>
which is ridiculous
11:58
<Hixie>
you can see this e.g. from the way the chairs of the htmlwg are only able to make decisions on tuesdays
11:58
<Hixie>
scheduled irc meetings would have the same problem
11:59
<AryehGregor>
Isn't it an HTMLWG policy that decisions are only made by e-mail and such? There's something about allowing asynchronous communication.
11:59
<Hixie>
yes, but subgroups (like the chairs) can arrange for their own decisions to be synchronous if they want
12:00
<Hixie>
so long as membership in the group isn't a requirement to influence decisions
12:02
AryehGregor
is glad to see that there will be no actual votes, just decisions by a cabal of three chairs instead of one editor
12:03
<AryehGregor>
Although theoretically the decision will be what causes the least strong objections instead of the one they think is technically best, so it's somewhat different.
12:05
<Philip`>
I guess "least strong objections" means you should threaten to shout loudly and file FOs etc if you don't get your way, so that your objections carry more weight
12:06
<Hixie>
yeah i'm curious to know how they intend to weigh rationales if not in the same way that i did when examining the issue as the editor
12:06
<AryehGregor>
Yes, that sounds like the strategically appropriate response to such a policy.
12:06
<AryehGregor>
Presumably they're allowed to ignore you if they think you're faking, though.
12:07
<jgraham>
Hixie: It is not impossible that they will weight rationales in the same way that you did but come to a different conclusion
12:07
<AryehGregor>
Honestly, no one can really fault them too much no matter what decision they make on something like microdata, since obviously they can't please everyone.
12:07
<jgraham>
But I agree that is not what they have stated they will do
12:08
<Hixie>
surely if they get a different conclusion then by definition they've used a different method of weighing arguments :-)
12:08
<Hixie>
i'm not saying there's anything wrong with that
12:08
<Hixie>
i'm just wondering what their method will be
12:08
<AryehGregor>
Even if they secretly just pick whatever solution they like, people will probably still end up being more satisfied because they were told it's a compromise rather than "I think you're wrong, too bad".
12:09
<Hixie>
(in fact i'd say that they _should_ use a different method, since otherwise the whole escalation process is pointless since they'd always agree with me)
12:09
<AryehGregor>
It wouldn't be pointless politically even if they always agreed with you, only technically.
12:10
<Hixie>
technically is what i care about
12:10
<AryehGregor>
Then why are you having anything to do with the W3C? :)
12:10
<jgraham>
Hixie: Well they can use the same _method_ but get different results due to different inputs
12:10
<Hixie>
AryehGregor: patent policy and to get microsoft's feedback
12:10
<Hixie>
AryehGregor: in retrospect, as i've said before, i think we would have been better off just getting our own patent policy
12:11
<jgraham>
e.g. you might tend to weigh arguments that correspond with your own experience more highly
12:11
<AryehGregor>
Well, but that's basically political.
12:11
<AryehGregor>
I'd think the WHATWG with its own patent policy would be at a significant disadvantage without Microsoft on board.
12:11
<Philip`>
The three chairs should meet on a wind-swept heath and drop the arguments into a cauldron and then dance around it until a conclusion reveals itself to them
12:12
<Hixie>
jgraham: i guess if you don't consider the weighing function part of the weighing function... :-)
12:12
<Hixie>
AryehGregor: in practice we've gotten very little useful input (or indeed any input) from microsoft
12:12
<AryehGregor>
Yes, but they at least claim they'll implement HTML5 at some point, which is something.
12:13
<AryehGregor>
I really meant from an implementation perspective, not feedback.
12:13
<AryehGregor>
It's not much good if Microsoft doesn't implement it.
12:13
<Hixie>
i don't think the w3c affected that really
12:13
<AryehGregor>
Maybe not. Too late now, I guess.
12:14
<AryehGregor>
Hixie, please tell your superiors at Google that they need to kill Microsoft faster, thanks.
12:14
<jgraham>
Hixie: The exact weighing function isn't exactly part of the "method"
12:14
<Hixie>
google has no intention of killing microsoft, we actually wish microsoft would compete better to give us more of a run for our money
12:14
<Hixie>
but that's another story
12:15
<Hixie>
jgraham: fair enough
12:15
Philip`
imagines a Google monopoly would be about as bad as a Microsoft monopoly
12:15
<gsnedders>
But they "do no evil"! :P
12:15
<AryehGregor>
Philip`, you mean like all the terrible effects we've been seeing from their monopoly on search?
12:16
<jgraham>
Google monopoly? Do you go round a little board and buy up different websites and build adverts on them to collect money?
12:16
<AryehGregor>
Slightly lower market share than MS in the OS market, but still.
12:16
<Hixie>
AryehGregor: google has nowhere near a monopoly on search
12:17
<Philip`>
Google doesn't seem to have shown any reluctance to develop and deploy their own private web technologies that are largely focused on the needs of other parts of Google, without much input from other vendors
12:18
<AryehGregor>
Well, that's called "not being a bunch of hippie open-source freaks", not "being monopolistic". :)
12:19
<AryehGregor>
As long as they don't try to lock you in to anything.
12:19
<AryehGregor>
Which generally they don't.
12:19
<AryehGregor>
There's some tie-in between their services, but usually pretty weak.
12:19
<AryehGregor>
Okay, I was supposed to leave ten minutes ago, bye.
12:19
<Hixie>
later
12:40
<boblet>
interesting—HTML5-style charset declaration isn’t stronger than user-declared character encoding in FF3.5.5, but http-equiv style one is
12:41
<Hixie>
yeah okcool.de has a comment to that effect
12:42
<Hixie>
i haven't been able to get more info on it -- do you have test cases showing this?
12:43
<boblet>
Hixie: is that to me? I’ll up one…
12:43
<Hixie>
yes :-)
13:19
<boblet>
Hixie: crap—testing methodology fail. Sorry. Here’s the question that made me check it: http://doctype.com/doesnt-html-5-meta-charset-tag-work
13:19
<boblet>
Hixie: I couldn’t reproduce the error
13:29
<boblet>
Can <nav> be used for site search or pre/next page links? I think no, but heard yes for search…
13:31
<boblet>
(I’ve also seen “read more…” links marked as <nav> btw ;-)
13:32
<daedb>
I can see <nav> being used for prev/next page links, but I wouldn't use it for search or read more links...
13:53
<Philip`>
boblet: Seems odd
13:53
<Philip`>
Could be his HTML/text editor detects the meta http-equiv and saves as UTF-8 in that case, and ISO-8859-1 otherwise, or something
13:54
Philip`
doesn't know what else it would be, especially since he says he's setting the HTTP Content-Type charset too
13:56
<Philip`>
boblet: I hope <nav> is okay for next/previous, since the multipage HTML5 spec uses it for that
14:12
<boblet>
Philip`: yeah that Doctype.com q is peculiar. hopefully I’ll hear back
14:16
<boblet>
Philip`: Looking at the HTML5 multipage spec, next/prev page links are part of a larger nav block. Do you think it’d apply on eg the meter/noscript links on http://dev.w3.org/html5/markup/nav.html ?
14:17
<boblet>
Philip`: Mike’s used div.nav span.nav-prev and span.nav-next
14:19
<daedb>
boblet: I'd probably use <nav> for those links...
14:20
<Hixie>
boblet: weird. there's a comment in the source of okcool.de that talks about this also, and i can't get anyone to explain to me why.
14:21
<Philip`>
boblet: I imagine he used that because the W3C pubrules don't like HTML5 markup
14:23
<Hixie>
who's doing webidl these days?
14:37
<Hixie>
indeed, is there a bug tracker for webidl?
14:37
<boblet>
Hixie: re: charset, maybe it’s just a problem somewhere else (server header, tools…)
14:37
<Hixie>
maybe...
14:38
<boblet>
Philip`: ok that makes sense
14:38
<Hixie>
hmm, webidl has a bug component in bugzilla, but not bugs
14:38
<boblet>
I’m surprised that <nav> applies to prev/next links, as they’re often just a single link (not a block), and there’s already rel="next/prev"
14:39
<Philip`>
Judging by the relevant code (around http://mxr.mozilla.org/mozilla1.9.1/source/intl/chardet/src/nsMetaCharsetObserver.cpp#319 (for Firefox 3.5)), it looks like http-equiv charset and meta charset are reported indistinguishably, so it shouldn't be possible for one to work and not the other
14:39
<boblet>
Hixie: would you use <nav> for site search form?
14:40
<Hixie>
would you have a "skip navigation" link for a site search form?
14:40
<Hixie>
Philip`: yeah, i never found any distinctions when doing the testing way back when
14:41
<Philip`>
(That does indicate <meta charset=utf8 foo=bar won't work, though)
14:41
<Philip`>
Uh, <meta charset=utf8 foo=bar>
14:45
<Hixie>
oh?
14:46
<Philip`>
Hmm, but that's not what happens in practice
14:47
<Philip`>
(The GetCharsetFromCompatibilityTag is only in the 'else' of a thing that checks for the number of attributes)
14:51
<Philip`>
Maybe that's not the code that's actually used
14:51
<boblet>
Hixie: I wouldn’t, but I can see some would. Yeah, that’s the nub
14:52
<Hixie>
boblet: i think it'd be fine to use <nav> for that, but also fine not to
14:53
<Hixie>
basically, don't use <nav> unless you think <section> would also be appropriate, with an <h1>Navigation</h1>.
14:53
<boblet>
Hixie: I really love that a lot of the structural element usage is almost up to authors. That’s what most people hate, of course, but still :)
14:53
<Hixie>
it'd be extremely difficult, and only mildly useful, to be more specific than we already are
14:53
<Philip`>
Whatever code Firefox 3.5 is using, it also accepts things like <meta notcharset="utf-8">
14:53
<Hixie>
and we're already waaaay more specific than html4 ever was
14:54
<Philip`>
so it can't just be the attribute-based extracter
14:54
<boblet>
makes it harder to write best practice info, but it’s much more interesting
14:55
<Hixie>
best practice is "don't use <div> and don't use class=''", basically
14:55
<Hixie>
but often you have to use one or the other
14:55
<Hixie>
since html isn't _that_ expressive, even with all the new stuff
14:56
<boblet>
Hixie: why not class? that seems to becoming a best practice with eg @stubbornella’s CSS performance info
14:58
<boblet>
I’ve actually started using her “h2 .h2 {}” style classes, then in html <section><h1 class="h2">
14:59
<Hixie>
o_O
14:59
<Hixie>
why would you do that
15:00
<boblet>
not as pure HTML but better than “section section h1, article section h1 {}”
15:00
<Hixie>
oh well, yes, we need to figure out a better css selector solution for nested section headers
15:00
<Philip`>
boblet: Why not use <h2>?
15:00
<Hixie>
i meant theoretical best practice in the future, not best practice now
15:01
<boblet>
also I think class="h2" has semantic meaning
15:01
Hixie
wonders when breakfast service is _supposed_ to start at his cafe
15:01
<boblet>
Philip`: HTML5-style headings—reset to <h1> with each section
15:01
<Hixie>
no, you can use whatever level you want
15:01
<Hixie>
<section><h2></h2></section> and <section><h1></h1></section> mean the same thing
15:02
<boblet>
well, I think it was maintain correct heading levels taking nesting into account, or reset to <h1> each time you start a new section, no?
15:02
<Hixie>
aha, 8am
15:02
<Hixie>
another hour
15:02
<Hixie>
no, it resets inside each <section>
15:03
<Hixie>
the highest <hx> within each section (er, lowest, i guess, closest to <h1>) is the heading
15:03
<boblet>
but not to <h1>? interesting
15:03
<Hixie>
well the best practice is you should have only one <hx> per <Section>
15:03
<Hixie>
so it doesn't make any difference
15:03
<boblet>
so you could skip levels, eg <h1></h1> <section><h3></h3>…
15:04
<Hixie>
you can use <body><h6></h6><section><h3></h3><nav><h1></h1></nav></section></body> means the same as <body><h2></h2><section><h4></h4><nav><h6></h6></nav></section></body>
15:04
<Hixie>
...which means the same as <body><h1></h1><section><h1></h1><nav><h1></h1></nav></section></body>
15:04
<Hixie>
each section does its own heading hierarchy
15:05
<boblet>
but wouldn’t that mean that it’s easier to just start at <h1> for each new section?
15:05
<boblet>
assuming you’re doing crazy stuff like applying styles via classes? :)
15:05
<Hixie>
apparently not, since you use classes :-)
15:05
<boblet>
haha
15:05
<Hixie>
<h1 class="h2"> is not as easy as <h2> :-)
15:06
<boblet>
hrm… I must ponder this more :) at this rate I’ll never release this site
15:07
<workmad3>
and even though <body><h1></h1><section><h2></h2></section></body> means the same thing no matter what level of heading you use to a machine, there are still ways of doing it that are clearer to people :)
15:07
<Hixie>
indeed
15:08
<workmad3>
and clearest to a human is probably either <h1> all the time or <hx> appropriate to the sectioning level...
15:08
<Hixie>
agreed
15:08
<workmad3>
(which I believe are the two examples given in the spec :) )
15:09
<boblet>
You might want to check out http://j.mp/6FMCDi — @stubbornella’s OOCSS presentation
15:10
<Hixie>
will do
15:10
<boblet>
it’s quite a different way to approach things. It’s definitely changed some of my perceptions
15:11
<boblet>
workmad3: I guess <h1 class="h2"> is just the first one with a cop-out for easier CSS selector writing
15:11
<boblet>
(at least, that’s what I thought)
15:11
<workmad3>
boblet: I personally see that as an abomination :)
15:12
<boblet>
hahaha
15:12
<workmad3>
and besides, why use a <h1> there... it's *much* simpler to use a <div>... so then you have <div class="h1">, <div class="em">, <div class="font">...
15:13
<boblet>
I can definitely see how on a large popular site writing long CSS selector chains would gradually kill you tho
15:13
<Hixie>
woah, that suggestion on slide 68 (the <h1 class="h2"> thing) is crazy
15:13
<Hixie>
if you're finding that you're applying h2 styles to an h3, then your semantics are wrong
15:13
<boblet>
Hixie: you’re killing me here
15:13
<boblet>
:D
15:13
<Hixie>
the rest of it so far is good
15:14
<Hixie>
i also disagree with slide 92
15:14
<boblet>
I think it‘s to address the ballooning of selectors problem; like specifying each case of clearfix rather than applying a .clearfix or .group style
15:15
<Hixie>
i think styling elements is the way to go, and styling classes should be the exception
15:15
<Hixie>
but that might just be that i think you should define defaults first, which seems to be the same, but said differently
15:15
<boblet>
well, she does say “(unless defining defaults)”
15:17
<Hixie>
yeah
15:17
<Hixie>
i strongly agree with what she says at the end
15:17
<Hixie>
i don't understand why we don't have better tools
15:18
<boblet>
@stubbornella addresses that too http://j.mp/7i7Dfe — CSS Wish List
15:19
<boblet>
adding programatic concepts like mixins to CSS
15:19
<workmad3>
boblet: check out sass :)
15:20
<boblet>
haha—planning to do so for this site (have it open in a tab somewhere here)
15:20
<workmad3>
sass does have mixins... and the ability to do maths, etc. in css files
15:21
<workmad3>
and it gets compiled down to normal CSS and can even be minimised, etc. :)
15:22
<boblet>
workmad3: yeah I’m looking forward to checking out stylesheet size differences (why I didn’t start with it)
15:22
<boblet>
ok thanks for the food for thought all. Bed calls
15:22
<workmad3>
I need chocolate and coffee myself
15:22
<workmad3>
but then it's only 3:30 p.m. for me :)
15:23
<boblet>
Will hopefully have an HTML5 site with the HTML5 articles I’ve written so far on it tomorrow to announce
15:23
<boblet>
later
15:26
<ray>
h1 class="h2"?
15:30
<miketaylr>
<div class="span">
15:30
<ray>
html5 has a way to set the charset besides meta http-equiv?
15:31
<Hixie>
you can just say <meta charset="...">
15:31
<Hixie>
(or use HTTP headers)
15:33
<ray>
that's nice
15:33
<Hixie>
it even works in existing browsers
15:33
<ray>
meta http-equiv=blahblahblah was the only thing i couldn't type from memory
15:34
<Hixie>
you could type the DOCTYPE from memory? :-)
15:34
<ray>
:)
15:34
<Hixie>
i'm impressed :-)
15:37
<Hixie>
on an unrelated note, i'm amazed that my last e-mail to shelley actually convinced her
15:38
<Hixie>
i guess it pays to be thorough with each e-mail and to show how one is contradicting oneself rather than just replying to the latest comment each time, ignoring earlier ones
15:43
jgraham
would like to dispell the notion that only people who don't think the use cases should be addressed support microdata
15:43
Philip`
wonders where Shelley said she was convinced
15:44
<Hixie>
Philip`: she didn't, but she didn't reply to my e-mail, which is the only sign i've ever seen of her admitting she was wrong on anything
15:44
<Philip`>
It seems an equally likely hypothesis is that she had already made all the points she wanted to make and was fed up with going in circles and therefore stopped
15:46
<Hixie>
that'd be a first, if so
15:46
<Philip`>
jgraham: If you do that, you could also dispel the notion that the use cases are synonymous with the Semantic Web community
15:47
<Hixie>
anyone got a suggestion for a good e-mail that summarises the storage mutex issues?
15:48
Philip`
doesn't care about Semantic Webs or Linked Data or RDF or anything, but still thinks it's probably useful to have a way to easily encode structured data in a web page so screen-scrapers are easier to write
15:49
<Hixie>
that's basically google's position too
15:50
<jgraham>
Philip`: I habe a draft of an email that says more or less that that (I wrote it before complaining on IRC even) but I am never convinced that actually sending email is a good idea
15:50
<jgraham>
*have
15:53
gsnedders
grumbles about not having a clearly defined difference between es-discuss (where discussion is most about ES5) and es5-discuss (where discussion is mostly about ES5)
15:53
<Philip`>
gsnedders: Kind of like public-html and whatwg?
15:53
<Philip`>
(on good days)
15:53
<gsnedders>
But both mailing lists are on the same server.
15:53
<Philip`>
(where public-html discussion is mostly about HTML5 and not about HTML WG processes)
15:54
<Philip`>
s/good/rare/ perhaps
15:55
<jgraham>
gsnedders: Well es-discuss is mostly about es.next these days
15:55
<jgraham>
But I think there are too many mailing lists
15:55
<Hixie>
woo, breakfast time
15:55
<Hixie>
bbl
16:59
Philip`
thinks the HTML WG should have a poll to determine what poll options to provide in a second poll
16:59
<TabAtkins>
Philip`: I'm not sure that's a good idea. Can we have a poll about it?
17:00
<gsnedders>
TabAtkins: I don't think a poll for that is worthwhile. Poll?
17:02
<TabAtkins>
Poll? Poll poll poll-poll.
17:50
<jgraham>
othermaciej: It would be nice if the issues list had the name or shortname of each issue somewhere
17:50
<othermaciej>
jgraham: good point
17:51
<othermaciej>
jgraham: I'll look into that when I get into the office
17:51
<jgraham>
othermaciej: Thanks
17:52
<gsnedders>
othermaciej: For the poll on microdata, is it one vote per W3C Member or one vote per WG representive
17:53
<jgraham>
gsnedders: It is not a poll
17:53
<othermaciej>
gsnedders: I have a three-part answer for that
17:53
<othermaciej>
1) it's not a vote
17:54
<othermaciej>
2) multiple representatives of an organization are free to each state their response
17:54
<othermaciej>
3) if we *do* somehow count (which we don't intend to), we will weight all responses from a single organization as one vote
17:54
gsnedders
had the understanding of the W3C process that it didn't matter how the outcome of the poll was judged, just whether it was a formal WG vote or not, jgraham
17:55
<othermaciej>
it is not a formal Vote
17:55
jgraham
wonders how that works if different respondents from the same organisation say different things
17:55
<othermaciej>
jgraham: might end up counting as 0.5 in favor of each option - but it likely won't come up, since we don't plan to count
17:56
<jgraham>
othermaciej: Sure :)
17:56
<othermaciej>
we just don't want people to complain or be suspicious if there are, say, 7 Apple responses or 10 Opera responses
18:11
<Philip`>
Ooh, impressive, 3 out of 3 of the times listed in the new telecon reminder email are correct and consistent
18:11
<Philip`>
(Too bad the date in the subject is wrong)
20:49
<MattCampbell>
Which html5lib tree implementation is best for round-tripping HTML? That is, I want to parse some HTML, possibly do some manipulations, and then serialize the tree back to HTML.
20:50
<MattCampbell>
I'm talking about the Python implementation.
20:55
<gauthierm>
if an autoplaying video is removed from the document using removeChild(), should it keep playing?
21:07
<kinetik>
gauthierm: yes
21:08
<gauthierm>
kinetik: thanks. Bug in FF 3.5 then.
21:08
<kinetik>
gauthierm: "Media elements must not stop playing just because all references to them have been removed; only once a media element to which no references exist has reached a point where no further audio remains to be played for that element (e.g. because the element is paused, or because the end of the clip has been reached, or because its playbackRate is 0.0) may the element be garbage collected."
21:08
<gauthierm>
ah sorry, I meant NOT a bug in FF 3.5
21:09
<kinetik>
i think there actually is a bug where the media is stopped early in some cases
21:09
<kinetik>
which should be fixed in 3.6
21:09
<gauthierm>
it is a bit weird that audio can keep playing with no way for the user to stop it
21:12
<Philip`>
I guess it helps if you want to write a page that has sound effects, since you don't need to manually keep all the audio elements in the page and then clean them up once they've finished playing
21:13
<gauthierm>
kinetik: is the bug you're referring to triggered when removing a media element that is currently playing?
21:14
<gauthierm>
FF seems to only continue playing if the media is set to autoplay and the element is removed before it starts playing.
21:14
<gauthierm>
WebKit stops when the element is removed in both cases.
21:22
<jgraham>
MattCampbell: It shouldn't matter in theory
21:22
<kinetik>
gauthierm: there was no self-reference to the media when playing, so it'd die at some random point after it was removed from the document and no longer referenced
21:22
<kinetik>
gauthierm: https://bugzilla.mozilla.org/show_bug.cgi?id=518659#c6
21:23
<jgraham>
I always use lxml although there are some bugs with that in corner cases (some of the XMLness checking code is known-wrong, it can't represent all doctypes, etc.)
21:24
<jgraham>
but it is the best tested
21:24
<MattCampbell>
If lxml is the best tested, then I'll use that.
21:24
<jgraham>
beautifulsoup should be avoided
21:28
<gauthierm>
kinetik: I'm a bit confused. Here is my test case: http://labs.silverorange.com/files/mozilla/video-testcase/test-case.html When the video starts playing and you click remove, audio stops instantly. If you click remove before the video loads (still a black box) it continues buffering and then starts playing in the background.
21:29
<MattCampbell>
How tolerant is html5lib's parser with regard to malformed markup? Can it handle any markup that a current browser can?
21:31
<Philip`>
MattCampbell: It will parse any sequence of bytes and never stop with a fatal error (excepting some of the bugs with lxml etc)
21:32
<Philip`>
and will recover from errors in a way that should be sufficiently compatible with current browsers that future browsers will be happy to adopt the same parsing algorithm
21:32
<kinetik>
gauthierm: that's because removing an active media element causes it to pause
21:35
<kinetik>
er, so, my original answer was incorrect, sorry
21:35
<MattCampbell>
jgraham: Mainly out of curiosity, why should BeautifulSoup be avoided?
21:36
<Philip`>
MattCampbell: It lacks features (like namespace support) and has quite a few bugs that we don't how to work around, if I remember correctly
21:38
<MattCampbell>
So it seems that html5lib + lxml.etree might be suitable for screen-scraping apps, which is what BeautifulSoup was intended for.
21:41
<Philip`>
It ought to work for that
21:41
<Philip`>
If not, please file bugs :-)
21:42
<Philip`>
(A while ago the BeautifulSoup author sounded interested in replacing its parser with html5lib, which would be good if you prefer that API, but I don't know if anything's happened with that)
21:43
<MattCampbell>
I have no particular API preference; since lxml seems to be the most tested, I'll use that.
21:46
<Philip`>
Some of the bugs with lxml may cause fatal errors if you try parsing lots of random real pages, but the bugs ought to be easy to fix and then it'll be robust
21:47
<gauthierm>
kinetik: ah, that makes more sense :) There is a bug in FF then because the networkState is NETWORK_LOADING when the element is removed but FF is NOT acting as if the pause() method was called. I will file on bugzilla.
21:47
<kinetik>
gauthierm: great, thank you
22:20
<Hixie>
TabAtkins: man, i hope you're able to keep track of all the change proposal changes in doing your updates
22:20
<Hixie>
i can barely work out what the proposal is anymore
22:21
<TabAtkins>
It's, well… It's a good thing I procrastinate.
22:22
<Hixie>
heh
22:22
<Hixie>
you have just a few more hours, i guess :-)
22:28
<othermaciej>
you have until 9 AM Pacific Time tomorrow, basically (i.e. the time of the telecon)
22:29
<TabAtkins>
The hope is that from the time I start to the time I turn it in, no more change proposals.
22:31
<hober>
TabAtkins: some feedback on your counter proposal
22:31
<hober>
TabAtkins: for what it's worth, here's what I think is one of the more compelling reasons for having features in HTML like <time>, the microdata attributes, etc.
22:32
<hober>
TabAtkins: HTML's existing extensibility mechanisms (class="" etc.) are pretty good, but could use some enhancing to better support groups like the Microformats community (who build on top of HTML.)
22:32
<hober>
TabAtkins: <time>, microdata, etc. simply are that enhancing of HTML's extenisbility mechanisms over what is in HTML4.
22:32
hober
didn't even mention RDFa
22:32
<hober>
:)
22:33
<TabAtkins>
Heh, but it's sort of dishonest then. RDFa could also be used to do that. Microdata just does it better and more easily.
22:33
<hober>
... which is why we should incorporate Microdata into HTML5 and not RDFa, no? :)
22:34
<TabAtkins>
It's why we should *support* Microdata. It doesn't, by itself, mean we should keep it in HTML5.
22:34
<TabAtkins>
But it is a good support for the actual reasons.
22:35
<hober>
Hmm. No, I think the point I'm trying to make *does* support "keeping microdata in HTML5", in much the same way that my argument supports "keeping class='' in HTML5"
22:36
<TabAtkins>
Not if there was compelling evidence that Microdata could serve the same noble purpose in its own separate spec.
22:37
<hober>
I have an insufficiently powerful imagination, I guess: I can't see the difference between Microdata and the other extensibility points of HTML. No one actually wants class='' to be in a different spec, right?
22:38
hober
studiously ignores the XHTML Role Attribute Module (because what is role='' if not "class='' 2: class harder"?)
22:40
<TabAtkins>
Oh, don't worry. I don't see a difference either. I'm just saying.
23:15
<Hixie>
i really don't understand why we would keep class="" in html5 if we didn't keep microdata, personally