00:37
<Hixie>
did shelley just formally object to the free market philosophy?
01:00
<gavin_>
where?
01:18
<Lachy>
Hixie, why are you surprised? Shelly objects to everything that doesn't result in getting what she wants
02:13
<Lachy>
http://twitter.com/minusfive/status/2524514401
02:13
<Lachy>
"@Lachy "you misunderstand. <font> is non-conforming, validators will give an error" - please share your sources? #html5"
02:19
<Hixie>
http://www.whatwg.org/specs/web-apps/current-work/#font
02:27
<Lachy>
thanks. I was looking for where the spec said that authors could only used the defined elements and attributes, but that list is good enough
02:32
<Lachy>
damn, I really can't keep up with the absurd volume of e-mail on public-html and whatwg. I had 0 mails left in either of those lists earlier today (after ignoring a whole bunch), and it's now up to 46 unread (combined) in just 12 hours
02:33
<Lachy>
and that's already after having read quite a few of them in the mean time
02:36
<Lachy>
I'll have to read them all in the morning. Hopefully there won't be another 50 by the time I wake up!
02:37
<tantek>
Lachy - email doesn't scale. Eventually you'll give up on efail.
02:37
<Lachy>
tantek, what's the alternative?
02:38
<Lachy>
the big problem is that the codec and summary threads are largely repetitive.
02:40
<tantek>
Lachy - the alternative is to put content on the wiki, and have email/IRC only for sharing updates/URLs
02:40
<tantek>
http://tr.im/wikibetter
02:41
<tantek>
more on EMAIL is EFAIL: http://tantek.com/log/2008/02.html#d19t2359
02:43
<Lachy>
tantek, I strongly disagree. Wikis are subject to edit wars and are not optimised for discussion purposes.
02:44
<Lachy>
they're even worse than forums for discussion purposes
02:45
<Lachy>
anyway, bed time. Good night.
03:08
<Hixie>
Lachy: try joining www-font
03:09
<Hixie>
50 a day so far this month
03:11
<Hixie>
http://blog.halindrome.com/2009/07/w3c-you-ignorant-slut.html?showComment=1247014132596#c5389136712076824446
03:15
<othermaciej>
oh jesus, www-font
03:23
Hixie
ponders the bibtex vocabulary
03:23
<othermaciej>
omg I got +1d by John Foliot
03:26
<Hixie>
the main use case for the bibtex stuff was drag-and-drop of content from one page into another (e.g. a wikipedia entry) with the destination having a script that automatically took the biblio data from the first page and added it to the second page's references
03:27
<othermaciej>
it seems to me that, on the one hand, that use case doesn't sound terribly common, and on the other hand, it could be served via script and agreed convention in absence of native support
03:27
<Hixie>
yeah
03:28
<Hixie>
ok i'm gonna strip the bibtex vocabulary out. it wasn't that popular in the first place.
03:30
<othermaciej>
I do think personal information and calendar events are of greater general interest
03:30
<Hixie>
yeah
03:57
<heycam>
re flexbox and css animation on http://wiki.whatwg.org/wiki/Companion_specifications, i think both have active editors
03:59
<dbaron>
does anybody know where the message was where the XHTML2 WG announced that XHTML2 would use the 1999 namespace?
04:00
<othermaciej>
I remember seeing it in the w3c archives but I don't remember where
04:01
<dbaron>
yeah, that's my current problem
04:02
<othermaciej>
I think it was posted by Shane McCarron
04:05
<dbaron>
well, I can cite http://lists.w3.org/Archives/Public/public-html/2007Jun/0313.html since I haven't found anything better
04:05
<othermaciej>
I think this might be the email I was thinking of: http://lists.w3.org/Archives/Public/public-xhtml2/2007Jun/0014.html
04:06
<othermaciej>
other than that, I've only seen discussion in minutes I think
04:06
<dbaron>
http://lists.w3.org/Archives/Public/www-html-editor/2006JulSep/0137 too
04:06
<othermaciej>
ah, that might have also been one I was thinking of
04:07
othermaciej
pats self on the back for remembering who posted it
04:07
<dbaron>
See, I remember one that Lachy wrote a good response to that I wanted to link to
04:07
<dbaron>
but I can't find it
04:07
<dbaron>
maybe it was on a member-confidential list
04:11
<othermaciej>
I see this from Lachlan: http://lists.w3.org/Archives/Public/www-html/2006Oct/0014.html
04:12
<dbaron>
no, I remembered something else more directly about the namespace
04:12
<dbaron>
but I have another link that makes the same point, although not quite as well, but well enough
04:13
<othermaciej>
xhtml2 is dead anyway, are you looking to desecrate the body?
04:15
<dbaron>
I was just writing a short post-mortem.
04:16
<othermaciej>
to me one of the weirdest events in the history of xhtml2 was Sam Ruby's "merger" proposal on his blog which seemed to capture no interest whatsoever in the HTML WG
04:16
<othermaciej>
even though I believe there was much intense private discussion preceding it
04:20
<sayrer__>
dbaron, sam ruby wrote on that
04:20
<dbaron>
sayrer__, on what?
04:23
<MikeSmith>
dbaron: I seem to remember looking for an archived message with such an announcement, and never being able to find one
04:23
<MikeSmith>
I think they may have never actually announced it
04:23
<dbaron>
MikeSmith, http://lists.w3.org/Archives/Public/www-html-editor/2006JulSep/0137 is one
04:23
<MikeSmith>
OK
04:24
<sayrer__>
dbaron, http://www.w3.org/html/wg/tracker/actions/105
04:25
<dbaron>
I'm looking for a particular message that was probably second half of 2006 or maybe early 2007
04:25
<dbaron>
or was, rather
04:28
<sayrer__>
http://lists.w3.org/Archives/Public/public-html/2007Oct/0385.html
04:28
<sayrer__>
getting warmer
04:28
<sayrer__>
but if you're not looking
04:28
<sayrer__>
nevermind then
04:52
<roc>
wow
04:52
<roc>
Tom Lord's latest to www-font really ups the ante
04:54
<roc>
http://lists.w3.org/Archives/Public/www-font/2009JulSep/0341.html for those following along at home
04:58
<dbaron>
I think http://lists.w3.org/Archives/Public/www-html-editor/2006OctDec/0014.html actually probably was the post I was thinking of
04:59
<dbaron>
and wasn't as clear as I remembered in explaining what "backwards compatible" meant
05:47
<othermaciej>
my car crash curiosity is aroused
05:49
<othermaciej>
roc: was he high when he posted that?
05:50
<roc>
I hope so
05:50
<othermaciej>
true, it would be much more disturbing if he wasn't
05:50
<othermaciej>
I wonder how he came to be so very interested in this issue
06:26
<shepazu>
Lord's comment is more or less what Chris Wilson argued to me and David Storey at Google I/O (minus the wise holy streetperson)
06:27
<shepazu>
no, correction... it was me and Arun, not David Storey
06:30
<sayrer__>
"Lord's comment is more or less what Chris Wilson argued to me "
06:30
<sayrer__>
lol
06:31
<sayrer__>
then I got to the end and the joke was over
06:31
<shepazu>
sayrer: I meant that in the sense of "Lord's Prayer"... I'm a lapsed Catholic, and I still have messianic visions of the lotus and the Beast
07:02
<othermaciej>
shepazu: I doubt that Chris Wilson argued the "also support straight TTF" part
07:17
<shepazu>
othermaciej: correct... just the bit that says "The EOT-lite part makes sense because, done correctly, it allows interop with already deployed versions of IE."
07:40
<hsivonen>
Hixie: I saw the discussion between you and maciej on IRC, but I'm unsure which email is expecting my reply. Could you help me with an URI, please?
07:41
<Hixie>
sure, hold on
07:42
<Hixie>
http://lists.w3.org/Archives/Public/public-html/2009Jul/0227.html
07:42
<Hixie>
but see also related thread messages
07:42
<hsivonen>
thanks
07:44
<othermaciej>
hsivonen: we were discussing whether non-fatal warnings might be a better kind of less-sever diagnostic than down-played errors as they currently stand, and also the possibility of applying that to summary=""
07:44
<othermaciej>
note: it might be legitimate to make lack of alt="" a nonfatal warning
07:44
<othermaciej>
since the spec says you should include alt except in exceptional circumstances
07:44
<hsivonen>
I don't like downplayed errors
07:45
<hsivonen>
I also don't like prescribed warnings
07:45
<hsivonen>
but I like downplayed errors less
08:08
<hsivonen>
my spirits sink reading sicking's description of what IE does with deferred doc.write
08:09
<hsivonen>
(the innerHTML part)
08:10
<Hixie>
when i specced defer="" i left a comment in the spec to the effect that i knew it didn't match IE but i was hoping no-one would notice :-/
08:11
<othermaciej>
I can't think of any possible thing to do with deferred document.write that would be sane
08:15
<Philip`>
Hixie: The mails to commit-watchers seem to indicate that changes to source are not getting propagated to index
08:16
<Hixie>
that is correct
08:16
<Hixie>
the changes in the recent checkins went to the IETF
08:17
<gsnedders>
Philip`: Pfff. Coward! Just read the spec and check what was updated.
08:17
<Hixie>
they were WebSocket changes
08:17
<Philip`>
Oh
08:17
<Philip`>
Whoops
08:57
hsivonen
wonders if zeldman's comments have different moderation behavior at differt times of day
08:57
<hsivonen>
that is, if you go to queue when it's night on Zeldman's time zone and straight to public when it's day
08:59
<hsivonen>
well, maybe my comment appears some time
09:07
hsivonen
wonders if Google is really going to keep Dalvik apps around or if Android is going Palm Pre now that they are doing the Chrome OS thing
09:07
<Hixie>
ok
09:08
<Hixie>
opinions on the new obsolete section? http://www.whatwg.org/specs/web-apps/current-work/#obsolete-features
09:08
<othermaciej>
hsivonen: Google just recently announced a native C++ SDK for Android I think
09:08
<Hixie>
see also changes under DOCTYPEs here: http://www.whatwg.org/specs/web-apps/current-work/#the-initial-insertion-mode
09:08
<othermaciej>
hsivonen: so it sounds more like they are going the other way with it
09:09
<gsnedders>
othermaciej: So how will they cope with non-ARM Android devices?
09:09
<othermaciej>
gsnedders: no idea! fat binaries?
09:09
<hsivonen>
looks like Web apps are the only thing that works across all these devices...
09:09
<gsnedders>
othermaciej: How will they cope with arch_that_doesn't_exist_yet?
09:10
<othermaciej>
gsnedders: you should probably ask someone who knows and understands the strategy
09:10
<hsivonen>
gsnedders: the question is even more interesting for Google Native Client if it catches on
09:10
<othermaciej>
iPhone's strategy is to lock down the hardware platform and severely limit number of available models so you know what your native apps are targeting
09:10
<gsnedders>
othermaciej: But I don't know anyone who does, so asking in IRC is the best thing I can do :)
09:10
<T-->
Hixie, just out of curiosity, what's the rationale of choosing iframes over framesets?
09:11
<hsivonen>
Hixie: "na name"
09:12
<Hixie>
T--: they're more flexible, and we can't really get rid of them (they're used everywhere)
09:12
<Hixie>
hsivonen: fixed
09:12
<othermaciej>
I would personally lean towards profile="" in the obsolete but conforming bucket, but say HTML processors MUST NOT vary their processing based on it
09:13
<othermaciej>
wording looks good to me though
09:13
<othermaciej>
I think this is an improvement over "downplayed errors"
09:14
T--
doesn't understand the ?flexible?.
09:14
<hsivonen>
Hixie: where's the normative statement that validators must warn about those?
09:14
<Hixie>
hsivonen: last subsection of that h2-level section
09:14
<Hixie>
i can move it up if you like
09:14
<hsivonen>
Hixie: works for me as is
09:15
<Hixie>
i've moved it up
09:15
<hsivonen>
ok
09:28
<jgraham>
So "obsolete" is the new "deprecated"
09:28
Hixie
growls at bug 7089
09:28
<Hixie>
jgraham: it always has been, hasn't it?
09:29
<jgraham>
Well I mean "obsolete"+warning
09:29
<Hixie>
the big difference between Transitional and the warnings is that you don't ever see "Your document is conforming!" if you have warnings, you see "You document is conforming, but with warnings"
09:30
<Hixie>
with transitional, once you had the transitional doctype, you never again saw anything clearly telling you that you weren't strict
09:30
<Hixie>
at least, i reckon that's a difference
09:30
<jgraham>
Given how popular compiler warnings are I can't imagine it will make a big difference
09:30
<Hixie>
true
09:30
<Hixie>
oh well, we'll see
09:30
<Hixie>
there's not much on the list
09:31
<othermaciej>
jgraham: are you saying compiler warnings are popular, or unpopular?
09:31
<jgraham>
othermaciej: I assume they are popular because compiling software tends to give me so many of them ;)
09:31
<othermaciej>
heh
09:32
<othermaciej>
WebKit builds with -Werror (or equivalent) and with many optional warnings on
09:32
<othermaciej>
(although in MSVC we have to turn some off, because they are stupid)
09:32
<jgraham>
(In seriousness, I know that many projects have a zero-warnings policy but it hardly seems to be normal and in web-developert terms probably translates onto the people who would care about the strict/transitional distinction)
09:33
<othermaciej>
for example, if you use a high enough warning level and don't silence this one specifically, MSVC will complain if you write if (intVariable) { something(); }
09:33
<othermaciej>
it wants you to say if (!!intVariable) { something(); }
09:34
<jgraham>
what's the difference?
09:34
<othermaciej>
in generated code and actual effect? nothing
09:34
<othermaciej>
but for the former, it complains about an implicit conversion to bool
09:35
<Hixie>
and !!a isn't an implicit conversion?!
09:35
<Philip`>
(if (intVariable != 0) seems nicer to read)
09:35
<othermaciej>
and warns you that it may be slow, or something
09:35
<othermaciej>
I'm not giving this as an example of a good warning!
09:36
<Philip`>
The Intel C++ Compiler has "info" (I think) messages, as well as errors and warnings, which say things like "order of argument evaluation is undefined" that aren't serious enough to be warnings
09:38
<Hixie>
ok bed time
09:38
<Hixie>
nn
09:39
<Philip`>
I always get tempted to waste time avoiding any messages the compiler might output, even if they're really pointless and I shouldn't care
09:39
<Philip`>
since I'm just incapable of ignoring them
09:39
<Philip`>
and it takes a lot of effort to convince myself to lower the verbosity setting
09:39
<othermaciej>
for WebKit we basically choose to make compiler or analysis tool messages either fatal or silenced
09:42
<hsivonen>
othermaciej: that flavor of compiler warning would have helped me avoid one classic if (foo = bar) bug in the HTML5 parser
09:42
<othermaciej>
hsivonen: gcc warns you if you write if (foo = bar)
09:42
<othermaciej>
hsivonen: it tells you that if you really meant to do that, you should put parentheses around it
09:42
<Lachy>
Hixie, in #conforming-but-obsolete-features section, you basically give the same list twice, but written in slightly different ways.
09:42
<othermaciej>
that's much better than the GCC warning
09:43
<Lachy>
oh, I see. The first is for authors. The latter is the UA conformance requirements
09:43
<Lachy>
Didn't make sense before I turned on the highlighting stylesheet
09:45
<hsivonen>
othermaciej: what really happened here is that I made the change quickly late in a review cycle for landing, so I only checked it compiled--I didn't take a good look at warning/assert smoke
09:45
<hsivonen>
so I guess no warning would have saved me
09:45
<othermaciej>
-Werror would have saved you
09:46
<othermaciej>
because then it wouldn't have compiled
09:47
<Lachy>
Hixie, with the "Hide UA text" stylesheet enabled, the link to "obsolete permitted DOCTYPEs" from the "Conforming but obsolete features" section no longer works because that's written in a .impl section
10:03
<Lachy>
re that discussion on www-font, it would suck to even allow EOT-lite, cause then we'll get stuck with DRM-encumbered fonts from major font vendors, even though the DRM is entirely ineffective and only serves as a hinderence to authors
10:03
<Lachy>
s/allow/require/
10:05
<Lachy>
just make ttf/otf mandatory. Authors can still provide EOT versions as a fallback font that will work in older versions of IE, at least until support for ttf/otf is widely deployed in more up to date versions of IE
10:06
<Lachy>
Are Microsoft still objecting to supporting TTF/OTF?
10:06
<othermaciej>
Web fonts, the other format war
10:07
<othermaciej>
Chris Wilson gave a really clear flat refusal
10:08
<Lachy>
It really makes me wonder where Microsoft's interests lie: In providing the best development platform for authors, or for giving in and promoting the absurd needs of the font-foundries?!
10:10
<Lachy>
I guess it's the same underlying philosiphy that Microsoft uses to push various forms of DRM on other types of media
10:11
<Philip`>
What is the argument against permitting fonts in a format that can't easily be reused outside the web? (separate from the arguments against not supporting TTF/OTF)
10:12
Philip`
is interested since his font subsetting tools makes fonts that can't easily be reused outside the web (or outside the context where they were originally used), and wonders if that's considered a bad thing
10:13
jgraham
hasn't been following the discussion but thought the issue was more to do with tying fonts to specific domains making testing a pain
10:13
<jgraham>
s/testing/deployment/
10:13
<Lachy>
EOT places restrictions on what domains a font can be used on and expects browsers to enforce those restrictions.
10:14
<Lachy>
(I could be remembering that wrongly, but I know the font foundries have been pushing for a feature like that in one form or another)
10:15
<Philip`>
If you could e.g. tweak a TTF/OTF so that it works fine in web browsers but is kind of broken if used in many other applications, e.g. by setting the embedding bits or setting the font name to "", would that be considered harmful to the world?
10:25
<Lachy>
yes, because it imposes what could be considered to be a Technological Protection Measure on the font, and could be subject to anti-circumvention clauses of laws like the DMCA.
10:25
<Lachy>
It would be the font equivalent of the failed broadcast flag proposal
10:27
<Lachy>
also, that's one of the proposals that was discussed on www-style a long time ago
10:28
<Lachy>
and I recall someone from the Linux community saying that support for the flag would have to be integrated into the systems font handling, and thus wouldn't protect against other uses on the system anyway
13:12
<takkaria>
I don't think the new warning model will be like Transitional, on reflection, because you can't turn the warnings off by putting a toggle in your source file
13:13
<gsnedders>
takkaria: You feeling better?
13:13
jgraham
was about to ask that
13:13
<takkaria>
well, I'm standing up and lighgt doesn't hurt now. :)
13:13
gsnedders
sticks tongue out at jgraham
13:14
<takkaria>
still a bit of a headace but that's bearable
13:17
<jgraham>
Nice weather we have here
13:19
<gsnedders>
Yeah, absolutely lovely.
13:42
<hsivonen>
Lachy: the WHATWG WordPress isn't up-to-date
13:43
<hsivonen>
Lachy: Akismet is out of date, too
13:46
<Lachy>
hsivonen, yeah, I noticed that yesterday
13:46
<Lachy>
I suppose I could update it now
13:48
<hsivonen>
hmm. did I just blog with a grammar error in the title?
13:48
<Lachy>
upgarded now
13:48
<hsivonen>
should it be "Help test" instead?
13:48
<Lachy>
yes
13:49
<Lachy>
or "Help With Testing..."
13:49
<Lachy>
but Help Test seems better
13:49
<hsivonen>
Lachy: thanks
13:50
<hsivonen>
fixed.
13:51
<Lachy>
plugins upgraded now too
13:51
<hsivonen>
thanks
13:51
<Lachy>
I should also check that our wiki installation is up to date as well later
13:52
<hsivonen>
it's been quite a while since the last "this week"
13:52
<Lachy>
but that takes a little more effort, since I don't have a script for it
14:04
<hsivonen>
how should this quote be read? http://twitter.com/markbirbeck/status/2388701600
14:10
<Lachy>
I have no idea what it means for something to "be highest impact"
14:11
<karlcow>
trying to look for more background in http://twitter.com/#search?q=%23swdag2009
14:11
<takkaria>
RDFa is falling as part of a meteor shower of general web technology, and out of all of the meteorites that will hit the earth, RDFa will be highest impact
14:12
<Dashiva>
Meaning it will hit a mountain?
14:12
<Lachy>
hsivonen, I'm curious what the practical difference will be as far as implementing downplayed errors vs. the new conforming but obsolete warnings.
14:13
<Dashiva>
I could read it as "be [the one with the] highest [degree of] impact"
14:13
<Lachy>
I thought downplayed errors meant that warnings should be issues, rather than full errors
14:13
<Lachy>
so it's not clear to me that these spec changes are anything more than superficial
14:15
<karlcow>
the ajax search call in twitter is killing my firefox
14:19
<hsivonen>
Lachy: warnings don't need new UI code, so they are easier
14:20
<Lachy>
hsivonen, sure, but how is issuing warnigns for the current spec different from issuing warnings for downplayed errors? Or would something else have been done for that?
14:21
<hsivonen>
Lachy: downplayed errors would have counted towards making the result turn into invalid
14:21
<hsivonen>
Lachy: and the vision for downplayed errors was that they'd sort and collapse differently
14:38
<Lachy>
hsivonen, that sounds like little more than UI design decisions that aren't really affected by what the spec actually said.
14:40
<hsivonen>
Lachy: the non-UI effect is that the presence of one or more downplayed errors made the document invalid but the presence of normative warnings doesn't
14:49
<Lachy>
so it's a very fine line between "non-conforming, but ignorable errors" and "conforming, but with warnings"
14:52
<Philip`>
"non-conforming, but ignorable errors" wouldn't give you a badge
14:52
<Philip`>
(Maybe the other won't either, which makes them harder to distinguish)
15:01
<sayrer__>
Lachy: sometimes validator authors are nervous about stepping outside of what the spec tells
15:01
<sayrer__>
tells them is "invalid"
15:24
<Lachy>
it seems really strange that with so many browsers over the years, the idea of creating a browser OS keeps coming up. It happened with Netscape. It happened with Firefox, and now it's happening with Chrome. I suspect the only reason it didn't really happen with IE is that it was already so closely linked to Windows anyway, it wouldn't make much difference
15:24
<Lachy>
-- http://arstechnica.com/web/news/2009/07/google-chrome-os-lives-and-is-coming-to-a-netbook-near-you.ars
15:28
<karlcow>
[10:04] <sayrer__> Lachy: sometimes validator authors are nervous about stepping outside of what the spec tells
15:28
<karlcow>
indeed. And what ever option you are choosing you get shot for your choices.
15:29
<karlcow>
been there, seen that (aka w3c markup validator communitie*S*)
15:29
<sayrer__>
I think the author conformance requirements would be better in a document titled "How to write high quality HTML"
15:30
<sayrer__>
then the validator could say "Don't you want to write high quality HTML?"
15:31
<Dashiva>
"Why do you hate freedom?"
15:33
<Philip`>
"The Android browser has very little in common with Chrome on the desktop (for example, the Android browser uses Apple's SquirrelFish JavaScript engine instead of Chrome's V8)." - apart from having, like, all of WebKit in common?
15:33
<Lachy>
I guess WebKit is just a very minor and insificant part of the browser, compared with the JavaScript engine
15:34
<Lachy>
</sarcasm>
15:34
<MikeSmith>
Philip`: where do that quote come from?
15:34
<Lachy>
sayrer__, how to write high quality HTML should be explained in HTML5 guides and references. Normative conformance criteria belong in the spec
15:35
<Philip`>
MikeSmith: The Ars article to which Lachy linked
15:35
<MikeSmith>
ok
15:35
<sayrer__>
Lachy: sure
15:36
<MikeSmith>
Lachy: I suspect sayrer__'s point is that they shouldn't be normative conformance criteria
15:36
<Lachy>
but then that wouldn't be very useful
15:37
<sayrer__>
useful for producing interoperable markup?
15:38
<Lachy>
sayrer__, while we could very easily get interoperability between implementations without any authoring conformance criteria, normative conformance criteria are essential to ensure conformance checkers are interoperable and to help authors write markup properly
15:39
<sayrer__>
you mean lint tools must produce the same result for all input?
15:39
<karlcow>
interesting that the dicussion around canonical html is coming back :)
15:39
<sayrer__>
"properly" is not an objective quality
15:40
<sayrer__>
and I'll note that conformance checker interoperability is not something the WG is chartered to take on
15:40
<sayrer__>
but it seems tangential to me, since conformance checkers are supposed to be the means, not the end
15:40
<Lachy>
no, I mean conformance checkers and validators. Lints offen flag stuff that isn't necessarily non-conforming, but which you might like to consider fixing anyway
15:41
<Lachy>
s/offen/often/
15:41
<sayrer__>
why do you need checker/validator rather than a lint tool?
15:42
<Lachy>
it depends on your individual needs and the capabilities of the lint tool in use
15:43
<Lachy>
e.g. you might have a lint which checks that an HTML document fits the criteria necessary for being a syntactically correct polyglot document, but which doesn't check any other conformance criteria.
15:43
<sayrer__>
I guess I don't understand whats motivating your position. do you feel validators force people to write good markup in some way?
15:45
<Lachy>
no, validators are very useful QA tools to help ensure the quality of the markup and catch mistakes. Conformance criteria help to ensure that different people are using a comment set of criteria, whcih makes it easier for developers to work together
15:45
<sayrer__>
common set of criteria?
15:46
<sayrer__>
lint tools are "are very useful QA tools to help ensure the quality of the markup and catch mistakes."
15:46
<sayrer__>
agree or disagree?
15:46
<sayrer__>
my feeling is that if you want validators to interoperate, you would make rules for validators
15:46
<Lachy>
partially agree. I've had experience with some lint tools that report useless information that I didn't need. But, they can be depending on your needs
15:47
<Lachy>
and also experience with lints that fail to report info that I do need
15:48
<Lachy>
sayrer__, the authoring conformance criteria in the spec do apply to validators. It says that explicitly in the conformance section
15:49
<sayrer__>
yes, I don't think that makes much sense.
15:49
<sayrer__>
validators aren't authors
15:49
<Lachy>
validators check markup on behalf of authors
15:50
<Lachy>
and if there weren't a common set of conformance criteria defined, then validators would vary significantly in what they report, and they wouldn't be particularly useful
15:51
<Lachy>
with a common set of criteria, developers can, in theory, use their favourite validation tools (perhaps based on the UI they like most) and be reasonably sure that they'll get equivalent results.
15:52
<sayrer__>
I am not sure how that became a goal, and it still seems tangential, since conforming HTML5 UAs will produce equivalent results no matter what
15:52
<Lachy>
I honestly do not understand your POV.
15:52
<sayrer__>
well, let's say one validator flags <font> and one doesn't
15:53
<sayrer__>
and the author happens to be using the validator that doesn't
15:53
<sayrer__>
so a <font> element has slipped through undetected
15:53
<sayrer__>
what is the interoperability consequence?
15:54
<Lachy>
that seems like a problem that only applies to your authoring-conformance-free spec
15:54
<sayrer__>
well, you are saying validators won't be useful without the requirements we have now
15:54
<sayrer__>
in a world without them, I expect some problems will occur if you are right
15:55
<sayrer__>
what is the problem that has occured here?
15:56
<Lachy>
from a browser POV, there isn't an interoperability issue. The issue is that different tools give authors different results. This reduces their ability to rely on tools to give useful results and could cause different people to argue about what's right and what's wrong, based on nothing more than the implementation decision of their favourite QA tools
15:59
<sayrer__>
so all validators must produce identical results?
15:59
<sayrer__>
and is it so bad if people argue about what's right and wrong?
16:00
<Lachy>
ideally, yes, as far as issuing errors is concerned. They can issue additional useful warnings based on the user's need
16:00
<Lachy>
or, actually, the user's desires
16:00
<sayrer__>
is that requirement in the spec?
16:01
<Lachy>
there's nothing in the spec to say that they can't issue additional warnings
16:01
<sayrer__>
is there a requirement that they must not issue additional errors?
16:01
<sayrer__>
otherwise, different tools could give different results
16:01
<Lachy>
no, tools are only allowed to issue errors defined in the spec
16:02
<sayrer__>
oh, I missed that. where is it?
16:04
<Lachy>
if it issues an error, or even treates something as a fatal error, where the spec doesn't say it can, then it has violated the actual criteria specified which doesn't say to issue an error
16:07
<sayrer__>
so then it couldn't issue an error for elements developed later
16:07
<sayrer__>
or elements in other namespaces
16:07
<sayrer__>
afaik, there are no fatal errors in HTML5
16:08
<sayrer__>
so far, I don't think there's anything in the spec that will enforce the level of interoperability you claim
16:08
<Lachy>
a spec that defines new elements would also update the relevant conformance criteria
16:08
<Lachy>
elements in other namespaces have their own criteria applying to them defined in their respective specs
16:11
<Lachy>
the parsing requirements allow implementations to treat errors as fatal instead of applying the corrective steps specified.
16:12
<Lachy>
"The error handling for parse errors is well-defined: user agents must either act as described below when encountering such problems, or must abort processing at the first error that they encounter for which they do not wish to apply the rules described below."
16:13
<sayrer__>
oh, that's very damaging to interoperability
16:13
<sayrer__>
but I thought that hedge existed for validators
16:13
<sayrer__>
it seems to have changed
16:14
<Lachy>
it depends on the needs of the specific implementation. Browsers, for example, couldn't really get away with aborting on errors in practice. But there are lots of other tools for which graceful error recovery is not essential
16:15
<Lachy>
A validator is one such use case for that
16:15
<Philip`>
Some (non-validator) tools really want streamability, and aborting on certain parse errors lets them do that
16:15
<sayrer__>
Lachy: not sure it is useful to claim they are conformant to the same spec
16:15
<sayrer__>
Philip`: yeah, a spider for instance
16:15
<Philip`>
(and they can streamably handle all conforming documents)
16:16
<sayrer__>
that last part is not true, since they can streamably all non-scripted conforming documents
16:16
<sayrer__>
streamably handle
16:16
<Lachy>
from memory, the adoption agency algorithm is one that affects streamability
16:17
<Philip`>
sayrer__: I'd expect most spiders would want to accept arbitrary input, and so they'd have to non-fatally handle invalid documents
16:17
<Lachy>
(hsivonen knows more about that, though)
16:17
<sayrer__>
Philip`: that's good point
16:18
<sayrer__>
I'll make a note about this
16:18
<Philip`>
sayrer__: I'm thinking of e.g. a tool that applies some kind of transformation to an HTML document, and doesn't want to buffer the whole document in memory
16:18
<Lachy>
sayrer__, tools that wish to incorporate an HTML5 parser into existing XML tool chains, and which require streamability, are likely to opt to abort on some errors
16:19
<Lachy>
e.g. a CMS that allows the author to write HTML, but which uses XML for everything else on the back end
16:19
<sayrer__>
they will probably opt out of more than just that
16:21
Philip`
wrote a web service that converts HTML to XHTML streamingly, but (if he remembers correctly) it also aborted on cannot-serialise-to-XML errors
17:58
<Philip`>
http://hacks.mozilla.org/2009/07/video-more-than-just-a-tag/ - "<video id="myVideo" src="myFile.ogv"/>" - isn't that syntax going to break badly?
18:11
<Lachy>
we really should do more to promote better quality fallback, beyond the useless "You're browser doesn't support this"
18:12
<Lachy>
even a simple <a href="video.ogg">Download and watch</a> would be better
18:23
<Dashiva>
I could imagine a script that generates fallback for video elements based on <source> and @src
18:51
<hsivonen>
Philip`: is your code for generating inputs that visit various tokenizer states publicly available?
19:05
<Philip`>
hsivonen: I think the latest version of the code is http://canvex.lazyilluminati.com/svn/tokeniser/ (particularly http://canvex.lazyilluminati.com/svn/tokeniser/test_gen.ml)
19:05
<Philip`>
though it's very out-of-date, and also very ugly
19:05
<Philip`>
and quite possibly quite buggy
20:15
<Hixie>
Lachy: there are lots of broken links from the author-only section
20:25
<Lachy>
ok
20:26
<Lachy>
that really makes it useless in many cases
20:32
<Hixie>
well like all the DOM stuff's <dfn>s are in the impl section
20:32
<Hixie>
so all the domintro green boxes point into the impl section
20:32
<Hixie>
not sure what to do about it
20:33
<jgraham>
Hixie: The sumamry attribute description looks OK to me but I suggest not referncing <caption> explicitly as a replacement, just the whole techniques section
20:33
<jgraham>
I think <a @name> could have a similar description
20:35
<jgraham>
like "This attribute was used in older versions of HTML to allow fragment identifiers o be associated with particular places in the document"
20:37
<Hixie>
how about saying it was similar to what ID does now
20:38
<Hixie>
checked in
20:40
<jgraham>
Hixie: perfect, thanks
20:46
<Hixie>
jgraham: btw, i think there's a difference between a whole spec, and a requirement in a spec
20:46
<Hixie>
jgraham: and in particular on the theora issue, the requirement in question is just importing another spec entirely, which i think is a particularly special case.
20:50
<jgraham>
Hixie: I agree that the theora thing is a somewhat different case
20:50
<jgraham>
But I think e.g. svg-in-HTML is quite a similar case
20:52
<jgraham>
in the sense that it is more useful to have a spec with support from 3/4 major vendors if one announces that they will not support it
20:52
<jgraham>
than to not have any spec at all
21:29
<Hixie>
jgraham: if they're going to implement regardless, then yes
22:12
<jgraham>
othermaciej: I think your proposed @summary text may be unnecessarily long and so unlikely to be read (the current text on alt has this problem)
22:12
<jgraham>
However I'm not sure
22:15
<othermaciej>
jgraham: it would be at the end of the already long section on how to provide table explanations
22:15
<othermaciej>
I think it is not overly long compared to the rest of that section
22:15
<othermaciej>
but
22:15
<othermaciej>
I am not wedded to my exact wording, it was just an example of how to provide the right information