/irc-logs / w3c / #html-wg / 2007-11-21 / end

Options:

# Session Start: Wed Nov 21 00:00:00 2007
# Session Ident: #html-wg
# [00:12] * Joins: sbuluf (epcnwhg@200.49.132.113)
# [00:19] * Joins: Lachy (Lachlan@213.236.208.22)
# [00:22] * Quits: jgraham_ (james@81.86.210.2) (Quit: This computer has gone to sleep)
# [00:35] * Quits: Lachy (Lachlan@213.236.208.22) (Quit: This computer has gone to sleep)
# [00:43] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
# [00:43] * Joins: jgraham_ (james@81.86.210.2)
# [01:01] * shepazu_ is now known as shepazu
# [01:03] * Joins: Lachy (Lachlan@84.215.41.149)
# [01:13] * Quits: tH (Rob@87.102.22.166) (Ping timeout)
# [01:16] * Quits: Lachy (Lachlan@84.215.41.149) (Quit: This computer has gone to sleep)
# [01:50] * Quits: jgraham_ (james@81.86.210.2) (Quit: This computer has gone to sleep)
# [01:58] * Joins: olivier (ot@128.30.52.30)
# [02:01] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [02:04] * Joins: Lachy (Lachlan@84.215.41.149)
# [02:06] * Joins: gavin_ (gavin@99.227.30.12)
# [02:13] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
# [02:22] * Joins: mjs (mjs@64.81.48.145)
# [02:53] * Parts: shepazu (schepers@128.30.52.30)
# [03:28] * Joins: shepazu (schepers@128.30.52.30)
# [03:35] * Joins: Zeros (Zeros-Elip@69.140.40.140)
# [03:35] * Quits: Zeros (Zeros-Elip@69.140.40.140) (Quit: Leaving)
# [03:39] * Quits: Lachy (Lachlan@84.215.41.149) (Quit: Leaving)
# [03:39] * Joins: Lachy (Lachlan@84.215.41.149)
# [03:40] * Quits: Lachy (Lachlan@84.215.41.149) (Quit: This computer has gone to sleep)
# [03:46] * Quits: shepazu (schepers@128.30.52.30) (Ping timeout)
# [03:57] * Joins: shepazu (schepers@128.30.52.30)
# [04:08] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [04:13] * Joins: gavin_ (gavin@99.227.30.12)
# [04:48] * Quits: shepazu (schepers@128.30.52.30) (Quit: Trillian (http://www.ceruleanstudios.com)
# [04:51] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
# [05:09] * Joins: Zeros (Zeros-Elip@67.154.87.254)
# [05:56] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
# [06:16] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [06:21] * Joins: gavin_ (gavin@99.227.30.12)
# [06:21] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
# [06:27] * Joins: olivier (ot@128.30.52.30)
# [06:29] * Quits: trackbot-ng (trackbot-n@128.30.52.30) (Client exited)
# [06:34] * Joins: trackbot-ng (trackbot-n@128.30.52.30)
# [06:34] * trackbot-ng is loading HTML Issue Tracking data...
# [06:34] * trackbot-ng found 15 users
# [06:34] <trackbot-ng> Tracking ISSUEs and ACTIONs from http://www.w3.org/html/wg/tracker/
# [06:35] * Joins: mjs (mjs@64.81.48.145)
# [06:55] * Quits: heycam` (cam@130.194.72.84) (Quit: bye)
# [06:59] <hsivonen> if the HTTP WG is not adding new features and is not defining error handling, what are they defining?
# [07:00] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
# [07:04] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
# [07:13] <Hixie> hsivonen: they're clarifying errors or something
# [07:14] <hsivonen> hmm. clarifying errors without specifying what to do with them...
# [07:17] <Hixie> no i mean errors in the spec
# [07:18] <hsivonen> oh
# [07:26] * Joins: heycam` (cam@210.84.13.228)
# [07:27] * Quits: heycam (cam@210.84.13.228) (Quit: bye)
# [07:27] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
# [07:28] * Joins: mjs (mjs@64.81.48.145)
# [07:28] * heycam` is now known as heycam
# [08:23] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [08:28] * Joins: gavin_ (gavin@99.227.30.12)
# [08:30] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
# [08:44] * Joins: paullewis (paullewis@82.242.109.217)
# [08:44] * Quits: paullewis (paullewis@82.242.109.217) (Quit: paullewis)
# [08:45] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
# [09:30] * Joins: edas (edaspet@82.120.100.44)
# [09:36] * Joins: tH_ (Rob@87.102.22.71)
# [09:36] * tH_ is now known as tH
# [09:43] * Joins: Lachy (Lachlan@84.215.41.149)
# [09:44] * Joins: zcorpan (zcorpan@88.131.66.80)
# [09:53] * Joins: shepazu (schepers@128.30.52.30)
# [09:58] * Quits: Zeros (Zeros-Elip@67.154.87.254) (Ping timeout)
# [09:58] * Joins: aroben (aroben@67.160.250.192)
# [10:02] * Quits: Lachy (Lachlan@84.215.41.149) (Quit: This computer has gone to sleep)
# [10:12] * Quits: Philip (philip@80.177.163.133) (Ping timeout)
# [10:13] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
# [10:14] <MikeSmith> anne - are you there?
# [10:17] * Joins: ROBOd (robod@89.122.216.38)
# [10:18] * Joins: Lachy (Lachlan@213.236.208.22)
# [10:24] * Joins: Lachy_ (Lachy@213.236.208.22)
# [10:30] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [10:34] <anne> am now
# [10:35] * Joins: gavin_ (gavin@99.227.30.12)
# [10:36] * Quits: shepazu (schepers@128.30.52.30) (Quit: Trillian (http://www.ceruleanstudios.com)
# [10:36] <MikeSmith> anne - can you please change the publication date in design-principles doc and this-version URI to 23 November?
# [10:37] <MikeSmith> btw, what tool do you use for generating the doc?
# [10:37] <MikeSmith> I notice you have both Overview.src.html and Overview.html
# [10:39] <anne> I use the CSS Spec Generator
# [10:40] <MikeSmith> OK, yeah, I remember you saying that now
# [10:40] <anne> done
# [10:40] <MikeSmith> thanks
# [10:44] <anne> just have to add -F date=2007-11-23 to the curl command for generating the spec :)
# [10:46] <hsivonen> isn't that side effects on GET?
# [10:46] <hsivonen> or is it POST?
# [10:48] <anne> I think both work
# [10:50] <anne> but I'm sure why GET is a problem here btw
# [10:50] <hsivonen> so what happens if your browser has W3C HTTP Basic Auth credentials in store and you navigate to a page that has <img src=''> or ping='' pointing to the spec generator?
# [10:51] <anne> not much
# [10:51] <anne> the spec generator takes a spec and some other variables as input and outputs a new spec
# [10:51] <anne> just like a search engine takes a key word as input and outputs a result page
# [10:51] <hsivonen> ah. I thought it wrote the result on w3.org
# [10:54] <anne> I do a cvs diff and cvs commit afterwards
# [10:56] <MikeSmith> anne - I sent to the transition request and just now sent the publication request and Cc'ed you and mjs
# [10:57] <anne> thx
# [10:57] <MikeSmith> I'll be traveling tomorrow, so please follow up with DanC and/or webreq to get it actually published
# [10:57] <MikeSmith> in case I'm away when the OK for the transition comes back
# [10:58] * Joins: marcos (root@124.171.166.8)
# [10:59] <anne> sure
# [11:02] * Quits: marcos (root@124.171.166.8) (Ping timeout)
# [11:02] <MikeSmith> anne - btw, great to finally have document() support in Opera
# [11:02] <MikeSmith> and somebody seems to have fixed a truckload of XSLT bugs since the last time I checked
# [11:02] <anne> jl
# [11:03] <MikeSmith> ah, yeah
# [11:04] * mjs is amazed at how many big web apps use XSLT for no apparent reason
# [11:05] <mjs> one of the reasons it's hard to keep web engine bloat under control is that it's hard to credibly rule out support for a feature - if any browser implements it and sites start using it, all engines eventually have to add it
# [11:05] <mjs> this tends towards perhaps more growth than is optimal
# [11:07] <anne> definitely more bugs :)
# [11:08] * anne wonders what karl is on about
# [11:09] * Joins: Julian (chatzilla@217.91.35.233)
# [11:09] <MikeSmith> web developers like have choices I guess. And different developers prefer different approaches, for whatever reasons. Some crazy ones actually seem to like XSLT
# [11:10] <MikeSmith> the fact that browsers now both support document() and node-set() means they can actually maybe do something useful with it
# [11:10] <MikeSmith> well, at least Webkit and Opera support node-set()
# [11:10] <MikeSmith> and MSIE
# [11:10] <MikeSmith> not sure about Mozilla
# [11:14] * Quits: aroben (aroben@67.160.250.192) (Ping timeout)
# [11:17] * Joins: Philip (philip@80.177.163.133)
# [11:17] <MikeSmith> hsivonen - has anybody made changes to the microsyntaxes page you set up?
# [11:17] <anne> nope
# [11:17] <anne> http://wiki.whatwg.org/wiki/Special:Recentchanges
# [11:19] * Joins: jgraham_ (james@81.86.210.2)
# [11:21] <hsivonen> MikeSmith: no. (I'm not surprised considering that it's basically me outsourcing my itch instead of letting people spontaneously scratch their own itches)
# [11:22] * Quits: jgraham_ (james@81.86.210.2) (Quit: This computer has gone to sleep)
# [11:23] <mjs> anne: me neither
# [11:23] <anne> it seems to me that the decision to do something is up to the whole WG too, not just authors
# [11:23] <mjs> I guess he thinks most web developers have a strong need to tell other web developers what to do
# [11:24] <anne> there is such a subset
# [11:25] <anne> i don't like how he implies that browser vendors are evil or something
# [11:25] <anne> i understand the need for easier non-normative text for authors, but that's tutorials
# [11:26] <MikeSmith> Am I mistaken or does the HTML5 spec does not impose any constraints on the syntax of the ID attribute value other than that it must not contain any space characters?
# [11:26] <anne> (and so far the only tutorials are created by Hixie and I...)
# [11:26] <anne> MikeSmith, you're correct
# [11:26] <mjs> yes
# [11:26] * mjs just wrote tests for that part of the spec recently
# [11:26] <mjs> by the way, do we have a way to submit test cases yet?
# [11:27] <MikeSmith> mjs - no, we do not
# [11:27] <mjs> I'd love to get the ones I wrote at the Tech Plenary off of my hard drive
# [11:27] <mjs> I was hoping something would get decided
# [11:27] <anne> e-mail them to www-archive for now?
# [11:27] <MikeSmith> e-mail them to www-archive at least
# [11:27] <MikeSmith> yeah, what anne said :)
# [11:28] <mjs> I guess that is more reliable than my hard drive, probably not quite as easy to find stuff
# [11:28] <hsivonen> mjs: more to the point, it seems to me that karl wants to let the subset of Web developers who have a strong need to tell other Web developers what to do tell it in the w3.org space
# [11:28] <anne> can we have some subversion/distributed repository on the W3C for HTML tests?
# [11:28] <MikeSmith> Or we could for now just decide on a subdirectory of html5 on dev.w3.org to put them in
# [11:28] <MikeSmith> anne - not a supported one
# [11:28] <mjs> I am wondering if we have a planned convention for how test cases should be named, also
# [11:28] <mjs> or if there is a proposed directory structure
# [11:28] <MikeSmith> I think DanC already has an unofficial mercurial repository set up
# [11:28] <mjs> should it be by numeric section of the spec?
# [11:29] <mjs> or plain english names?
# [11:29] <mjs> or what?
# [11:29] <anne> html/common-attributes/id/xxx.htm
# [11:29] <MikeSmith> mjs - we have had no discussion about that yet as far as I know
# [11:29] <anne> where xxx is a number
# [11:29] * mjs was kinda hoping we could use cvs like we do for everything else
# [11:29] <zcorpan> certainly not numeric section of the spec, since that changes all the time
# [11:29] <mjs> anne: could we maybe do xxx-some-reasonable-name.htm
# [11:29] <MikeSmith> I think Hixie at least thinks it's probably too early to start writing too many test cases
# [11:29] <anne> mjs, for people who can dream up reasonable names, sure :)
# [11:29] <mjs> anne: I hate number-only names, it is hard for me as an engine developer to tell what the test is testing when it fails
# [11:30] <anne> mjs, supposedly it has some information in <title> too
# [11:30] <mjs> MikeSmith: some parts of the spec are clearly pretty stable
# [11:30] <mjs> likely most things that overlap HTML4.01
# [11:30] <mjs> probably also <canvas>
# [11:31] <zcorpan> using the file name instead of (or as well as) <title> would work for me
# [11:31] <anne> actually, the HTML4 features had lots of interop issues
# [11:31] <mjs> anne: the regression test tool WebKit uses lists filenames that fail without human interaction, to see the title I have to click
# [11:31] <mjs> so, somewhat selfishly, I like for the tests to have semi-understandable names
# [11:31] <mjs> anne: well, this is why test cases need review
# [11:31] <MikeSmith> mjs - So, I guess if you and others already have test cases for those, we really should try to get something set up now.
# [11:31] <mjs> maybe ones that are less certain should go into an "unstable" dir
# [11:31] <mjs> I think we need some test suite editors to review incoming test cases
# [11:32] <mjs> MikeSmith: I made a pretty exhaustive set of tests for conformance requirements for the "id" attribute
# [11:32] <mjs> MikeSmith: I'd make more in my spare time if I had a place to put 'em
# [11:32] <mjs> maybe even donate some test cases from WebKit's regression tests if any are appropriate
# [11:32] <MikeSmith> OK, that's a very good motivation for me to get something set up then.
# [11:33] <mjs> it would make my life easier if it was cvs or svn or git
# [11:33] <mjs> so I don't have to learn yet another VCS
# [11:33] <mjs> personally I don't think distributed version control is very useful for a test suite
# [11:33] <mjs> because there's really no need for branches
# [11:34] <anne> now dev.w3.org has autmated checkout this should be possible
# [11:34] <MikeSmith> ACTION: Michael(tm) to work on getting a non-CVS VCS system set up for storing test cases, by 2007-11-29
# [11:34] * trackbot-ng noticed an ACTION. Trying to create it.
# [11:34] <trackbot-ng> Created ACTION-26 - Work on getting a non-CVS VCS system set up for storing test cases, by 2007-11-29 [on Michael(tm) Smith - due 2007-11-28].
# [11:36] * Joins: edaspet (edaspet@82.120.100.44)
# [11:37] <mjs> MikeSmith: does it really have to be non-CVS?
# [11:37] <mjs> I mean, I have no great love of CVS
# [11:37] <mjs> but it seems adequate for this task and has the huge advantage of already being set up
# [11:37] <MikeSmith> mjs - the one big reason I would really prefer not to use CVS is because of lack of support for atomic changesets
# [11:37] <mjs> and familiar to many likely contributors
# [11:37] <MikeSmith> aye
# [11:37] <mjs> MikeSmith: does that matter for a test suite?
# [11:38] <mjs> who cares if 5 test cases are an atomic commit or not?
# [11:38] * Quits: edas (edaspet@82.120.100.44) (Ping timeout)
# [11:38] <mjs> keep in mind that a test suite is a pretty specialized kind of software project
# [11:38] <MikeSmith> I don't know. You guys tell me. You've written a lot more test suites than I have
# [11:38] <anne> we eventually moved our internal testsuite from CVS to SVN
# [11:38] <mjs> my opinion is no, but anne has probably written more test cases than I have
# [11:38] <mjs> WebKit's test suite is in WebKit's SVN
# [11:39] <mjs> we do make use of atomic commits, but mainly to make test commits atomic with the code change they test
# [11:39] <mjs> for regression tests
# [11:39] <MikeSmith> If nobody else feels strongly that we need support for atomic changesets, I am happy to agree to setting it up in CVS
# [11:39] <anne> personally I have no preference
# [11:39] <mjs> I have no preference either
# [11:40] * Joins: edas (edaspet@82.120.100.44)
# [11:40] <MikeSmith> then, let's set up on dev.w3.org today
# [11:40] * MikeSmith wonders who else might be contributing significant numbers of test cases
# [11:40] <MikeSmith> jgraham, zcorpan, hsivonen ?
# [11:41] <mjs> as far as VCS's go, SVN is probably my currently most familiar, but I care far less about that than having something ready to go ASAP
# [11:41] <MikeSmith> Lachy ?
# [11:41] <mjs> we can always migrate later
# [11:41] <anne> dev.w3.org/html-tests/
# [11:41] <Lachy> yes
# [11:41] <MikeSmith> mjs - true
# [11:41] <mjs> Phillip "canvex" Taylor
# [11:41] <anne> or dev.w3.org/html5/tests/
# [11:42] <MikeSmith> anne - yeah, that would be better, IMHO
# [11:42] * Quits: edaspet (edaspet@82.120.100.44) (Ping timeout)
# [11:42] <Lachy> MikeSmith, did you want me for something?
# [11:42] <anne> the dir needs to have something about the MIT license
# [11:42] <MikeSmith> Lachy - was asking if you might be contributing test cases for HTML5
# [11:43] <anne> so that when you commit within that dir you automatically license it under that license
# [11:43] <Lachy> oh right, just reading the IRC logs now
# [11:43] <mjs> Lachy: MikeSmith was trying to figure out what people might contribute a lot to the test suite
# [11:43] <Lachy> I was also thinking about how I could contribute to the HTML5 for authors document
# [11:44] * Lachy wants to be an editor
# [11:45] <MikeSmith> Lachy - Web API group needs editors :)
# [11:45] <hsivonen> Lachy: do you represent the right community or do you represent implementors? ;-)
# [11:45] <Lachy> I'm already an editor there
# [11:45] <anne> I would suggest focusing on something else as well
# [11:45] <MikeSmith> you can be both an editor and a test-case producer
# [11:46] <Lachy> hsivonen, I was a professional web developer for 4 years before joining Opera. I think I'm qualified
# [11:46] <MikeSmith> anne - do you have perms to create new dirs on dev.w3.org ?
# [11:46] <anne> yeah
# [11:46] <hsivonen> Lachy: cool
# [11:47] <Lachy> MikeSmith, which specs in webapi need an editor?
# [11:47] <MikeSmith> so do we have enough time right now to have a discussion here about subdirectory structure?
# [11:48] <mjs> what should we use as the process for revewing test suites?
# [11:48] <mjs> er
# [11:48] <mjs> test cases
# [11:48] <mjs> I'd like it if we could designate some people to be test case code reviewers
# [11:48] * Quits: sbuluf (epcnwhg@200.49.132.113) (Quit: sbuluf)
# [11:49] <mjs> for pre-commit review
# [11:49] <anne> http://dev.w3.org/html5/tests/
# [11:49] <mjs> perhaps we should let Hixie recommend a set of test suite reviewers (I hope his set would include more than just him or we'll be pretty badly bottlenecked)
# [11:50] <MikeSmith> ideally would be great to have Hixie review everything
# [11:50] <MikeSmith> but I suppose that may not be practical
# [11:50] <anne> Simon Pieters, Philip Taylor , Lachlan Hunt, Henri Sivonen can all review
# [11:50] <anne> and me
# [11:50] <mjs> I think a lot of people could do very adequate test case review
# [11:50] <mjs> such as the set anne listed
# [11:50] <anne> I would expect people to delegate tests to other people if they are not sure
# [11:50] <MikeSmith> sounds fine to me
# [11:50] <mjs> (plus obviously Hixie)
# [11:51] <MikeSmith> so we will need a readme saying as much, I guess
# [11:51] <mjs> can we set up a separate mailing list for test case review maybe?
# [11:51] * anne adds a readme
# [11:51] <MikeSmith> along with an announcement to the appropriate list
# [11:51] <MikeSmith> once we get the subdirectory structure decided
# [11:51] <mjs> to avoid spamming public-html
# [11:51] <Lachy> public-html-testsuite would be good
# [11:51] <mjs> I like subdirectory structure by short name of section/subsection
# [11:52] <Lachy> (like public-css-testsuite for the css wg)
# [11:52] <mjs> then for individual test cases nnn-text-name.htm
# [11:52] <MikeSmith> can we use the fragment identifiers as short names?
# [11:52] <anne> they're not stable at all
# [11:52] <MikeSmith> damn
# [11:52] <MikeSmith> can we maybe suggest that they be made stable?
# [11:53] <anne> maybe
# [11:53] <anne> I think we should do it a bit ad-hoc
# [11:53] <mjs> I say we let the test suite reviewers reorganize it as necessary
# [11:53] <mjs> and also nominate additional test suite reviewers as they see fit
# [11:54] <anne> /tests/video/ /tests/audio/ tests/global-attributes/id/ /tests/parsing/ /tests/canvas/2d/
# [11:54] <anne> /tests/input/checkbox/
# [11:54] <anne> or maybe /tests/forms/input/checkbox/
# [11:55] <mjs> tests/elements/embedding/video
# [11:55] * anne points at http://tc.labs.opera.com/html/
# [11:55] <anne> what would you put in embedding though?
# [11:55] <mjs> (not sure if that's worth it)
# [11:55] <anne> well anyway, something like that
# [11:56] <mjs> in embedding itself? I dunno if there are tests that are generic to categories of elements
# [11:56] <anne> on a first come first serve basis unless it's really dreadful in which case someone moves a bunch of tests
# [11:56] <MikeSmith> everybody OK with public-html-testsuite as the mailing-list name?
# [11:56] <mjs> yes
# [11:56] <anne> maybe public-html-tests
# [11:56] <anne> nm, i'm ok
# [11:56] <MikeSmith> anne - see Lachy comment about public-css-testsuite
# [11:56] <mjs> I defer to whatever the test suite reviewers decide as far as organization goes
# [11:56] <anne> no need to discuss this
# [11:56] <mjs> putting elements at top level might make sense
# [11:57] <mjs> I wonder if there'd be any common tests for audio/video of the HTMLMediaElement inteface
# [11:57] <mjs> maybe makes more sense to test it twice
# [11:58] <Lachy> mjs, sure, the test scripts for video and audio can probably be shared between them
# [11:59] <Lachy> so the scripts only need to be written once, and then generate equivalent tests for audio and video
# [12:00] <mjs> I like organizing by spec section because then it is more clear which sections are entirely missing tests
# [12:01] <anne> i don't trust Hixie in keeping the sections stable
# [12:01] <anne> but i'm willing to do it that way
# [12:01] <anne> we can see how it goes
# [12:02] <mjs> it would mean potential work to reorganize the test suite when things change
# [12:02] <anne> and breaking URIs
# [12:02] <anne> but maybe that's ok
# [12:05] <anne> i'll add to the readme that we might break URIs during reorganizing
# [12:20] <MikeSmith> I'm submitting the request to create the testsuites mailing list; for initial subscribers, I have Hixie, anne, zcorpan, Philipp "canvex" Taylor, Lachy, hsivonen
# [12:20] <MikeSmith> anybody else to add for now?
# [12:20] <hsivonen> MikeSmith: mjs?
# [12:20] <MikeSmith> yeah, I got mjs too
# [12:20] <mjs> I'd like to subscribe, yeah
# [12:24] <MikeSmith> OK, I've submitted the request for the list
# [12:25] <Lachy> awesome
# [12:25] * Lachy sets up another filter to manage my ever-growing list of mailing list subscriptions
# [12:25] <MikeSmith> seems like before announcing it (after it's actually set up), might be a good idea to have a subdirectory structure and instructions for submitting test cases
# [12:25] <MikeSmith> Lachy - heh
# [12:26] <Lachy> MikeSmith, did you go with pubilc-html-testsuite?
# [12:26] <MikeSmith> yep
# [12:28] <MikeSmith> because along with public-css-testsuite there are already public-xml-testsuite and public-xml-schema-testsuite
# [12:28] <MikeSmith> so seems like a reasonable convention to follow
# [12:29] <MikeSmith> hsivonen - so just to be clear, the microsyntaxes page should define the microsyntaxes based on what HTML5 says, not on any other references, right?
# [12:30] <anne> yes
# [12:32] <hsivonen> MikeSmith: based on HTML5 (with WF2 considered part of HTML5)
# [12:33] <anne> http://dev.w3.org/html5/tests/structure.txt
# [12:33] <Lachy> I wonder what happened to my subscription to public-html-comments. I definately subscribed when it was opened, but seemed to have been unsubscribed since.
# [12:33] <hsivonen> MikeSmith: when a syntax is imported by reference, it should describe the imported syntax (like media queries) but not as a copy-paste if the external spec is copyright-wise not copy-pasteable to a MIT-licensed doc
# [12:33] <MikeSmith> hsivonen - so I just added http://wiki.whatwg.org/wiki/MicrosyntaxDescriptions#ID
# [12:33] <Lachy> Oh well, re-subscribed
# [12:34] <hsivonen> MikeSmith: cool. thanks
# [12:34] <MikeSmith> Lachy - hmm. I guess everybody should check to make sure they are subscribed to public-html-comments
# [12:35] <MikeSmith> because that is where public comments will go on the HTML Design Principles doc
# [12:35] <anne> You can verify here: http://cgi.w3.org/member-bin/MailingListQuery.pl?queryList=public-html-comments
# [12:35] <Lachy> ah, I see. It looks like I never actually responded to the confirmation message.
# [12:35] <MikeSmith> I mean those who care to see the comments
# [12:35] <anne> so far only 15 subscribers
# [12:36] <mjs> I need to subscribe to that
# [12:37] <Lachy> glad I didn't miss much discussion there. Only a 7 messages posted to it in 4 months
# [12:37] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [12:41] <MikeSmith> hsivonen - does the microsyntax page need to define what a space is?
# [12:42] <hsivonen> MikeSmith: I suppose it needs to.
# [12:43] <anne> I think you should replace "space characters" with "whitespace" and leave it
# [12:43] <anne> this is for authors, right?
# [12:43] * Joins: gavin_ (gavin@99.227.30.12)
# [12:43] <hsivonen> anne: for authors, yes
# [12:44] <hsivonen> if whitespace is enough for authors to fix their stuff, then that's enough
# [12:44] <anne> whitespace is more clear than "space characters"
# [12:44] <MikeSmith> aye
# [12:44] <hsivonen> ok
# [12:44] <anne> if it's not enough i'll guess you get complaints
# [12:44] <hsivonen> yeah
# [12:45] <hsivonen> spending too much time with the spec makes me err on the side of unuseful accuracy when deciding between simple and accurate
# [12:50] <hsivonen> I think that I should carve out all of the XSD regexp stuff and use custom datatypes that the UI can show elaborations for
# [12:53] <MikeSmith> Lachy - about Web API specs: Window Object, File Upload, and Clipboard Operations specs are all in need of editors
# [12:54] * Quits: edas (edaspet@82.120.100.44) (Ping timeout)
# [12:56] <anne> i would expect file upload to be part of web forms 2
# [12:57] * Joins: edas (edaspet@82.120.100.44)
# [12:57] * anne thinks there's more need for Web DOM Core 5 and Web DOM Events 5
# [12:58] <mjs> I would love to see DOM Core and DOM Events updates
# [12:58] <mjs> I wish I had free time to help with those
# [12:58] <mjs> or Window Object for that matter
# [12:59] <mjs> I'm not sure File Upload or Clipboard Operations are needed
# [12:59] <mjs> unless they are attempts to factor the relevant parts of HTML5
# [13:00] <anne> file upload was a separate attempt, that people thought would be better integrated in <input type=file>
# [13:01] <anne> clipboard was based on HTML5
# [13:02] <mjs> the fact that ability to upload files in HTML is tied to a specific control is somewhat helpful from a security POV
# [13:02] <anne> that was the reason to move it back
# [13:03] <anne> so there would be only one true way to upload files...
# [13:17] * Joins: myakura (myakura@122.17.145.60)
# [13:21] * Quits: Dashiva (noone@84.48.60.15) (Ping timeout)
# [13:21] * Joins: Dashiva (noone@84.48.60.15)
# [13:31] * Quits: Dashiva (noone@84.48.60.15) (Ping timeout)
# [13:32] * Joins: Dashiva (noone@84.48.60.15)
# [13:35] <MikeSmith> so here's a guy who I'm sure pocketed some decent cash for writing a really simple article on HTML5
# [13:35] <MikeSmith> http://www.ibm.com/developerworks/xml/library/x-html5xhtml2.html
# [13:35] <MikeSmith> IBM developerworks pays good money for articles
# [13:36] <MikeSmith> I am certain they would be happy to publish something about html5lib
# [13:36] <MikeSmith> or the validator.nu
# [13:36] <MikeSmith> or video or canvas or offline-applications/client-side-db-storage
# [13:37] <MikeSmith> and many other parts of HTML5
# [13:37] <MikeSmith> if we cared to take the time to submit some article proposals to them
# [13:39] <MikeSmith> that article is only about 2000 words
# [13:39] <MikeSmith> and I would guess that they paid something like 200 or 300 USD for it
# [13:41] <anne> "XHTML V5"
# [13:41] <MikeSmith> heh, yeah
# [13:41] <MikeSmith> but honestly that might not be the fault of the author
# [13:42] <MikeSmith> that could be some "refinement" made by one of their copy editors
# [13:43] <anne> well, he uses it as: "XHTML V5 (HTML V5 rewritten as an XML dialect)"
# [13:43] <MikeSmith> yeah, reading that in context kind of reveals some muddled thinking
# [13:43] <anne> i guess the only weird thing is the "VX" part but that's consistent throughout
# [13:44] <zcorpan> perhaps put a comment in the "how can we improve this article" box?
# [13:44] <MikeSmith> the first rule of copyediting is to be consisent
# [13:44] <MikeSmith> even if you are consistently wrong
# [13:45] <anne> there's also the http://www.rootly.com/topics/technology/linux_unix/Is_XHTML_V2_a_valid_successor_to_HTML5/ confusion
# [13:46] <anne> zcorpan, done
# [13:53] <MikeSmith> zcorpan - lemme know if you have other XSLT stuff I can help with. I need to put to good use all the money the US government spent to send me to XSLT school (back when I was a dual document-transformation specialist/assassin in the CIA)
# [13:57] <MikeSmith> anne, mjs - the tarot cards are telling me that the design principles doc may even end up getting published today or tomorrow
# [13:57] <mjs> cool
# [13:57] <mjs> publishing is good
# [13:57] <MikeSmith> so it will soon just be a matter of waiting for all the love letters to come in from fans around the world
# [13:58] <MikeSmith> as the good feelings spread from public-html to public-html-comments
# [13:58] <anne> if I need to change the date again let me know
# [13:59] <MikeSmith> hmm
# [13:59] <MikeSmith> seems we might actually
# [13:59] <MikeSmith> I spoke perhaps a bit too soon
# [13:59] <MikeSmith> may need to wait until monday to publish
# [13:59] <MikeSmith> it is generally better to publish at the beginning of the week
# [14:00] <MikeSmith> as far as getting attention
# [14:00] <MikeSmith> getting people to actually notice the announcement, I mean
# [14:02] <DanC> MikeSmith, my hg stuff is http://homer.w3.org/~connolly/projects/
# [14:02] <MikeSmith> DanC - got it, thanks
# [14:02] <zcorpan> MikeSmith: ok. don't need anything atm, but i'll keep it in mind :)
# [14:03] <MikeSmith> DanC - for the value of getting a Design Principles announced well, seems better to wait to publish until Monday
# [14:03] <MikeSmith> or not?
# [14:04] <DanC> hmm... I think I'd rather have it done
# [14:05] * anne agrees with DanC
# [14:06] <DanC> MikeSmith, we could ask anne and other bloggers to wait until monday to blog so that the news doesn't get lost in the thanksgiving holiday
# [14:06] <DanC> but I don't even feel strongly about that
# [14:07] <DanC> I don't need this WD to be a huge news item. I'm happy for it to be mostly a beta-test of the WG publication process
# [14:08] * DanC isn't really supposed to be working now; is having breakfast with some friends in 23 minutes...
# [14:08] <hsivonen> if there's now organizational OK for publishing it, I think it would be good to get it over with even if Thanksgiving is coming
# [14:09] * DanC hasn't seen the transition request
# [14:09] <MikeSmith> DanC - sent to chairs@
# [14:09] * DanC sees it now
# [14:09] <DanC> 27 nov? hmm.
# [14:09] <mjs> hey DanC
# [14:10] <mjs> MikeSmith: so is the new test cases list set up? I guess maybe it should be sort of announced before I start submitting stuff
# [14:10] <DanC> MikeSmith, I don't see the status info we discussed in http://www.w3.org/html/wg/html5/principles/#sotd
# [14:10] <Philip> About tests: Would it be desirable to eventually put my canvas tests in that w3.org thing? They use a kind of unique system for generating the actual test-case HTML files
# [14:11] <anne> at least a copy of the actual tests, yes
# [14:11] <Philip> (and for naming the cases, etc)
# [14:11] <mjs> Philip: if they are valid conformance tests, then yes - we now have a cvs directory for tests and a mailing list for review
# [14:11] <mjs> if some test cases are script-generated we may eventually want systems for that or something
# [14:11] <DanC> oh ... "mini black out"; webmaster is on holiday
# [14:13] <Philip> I've currently got three HTML files for every test case - one with full details so you can see what it's trying to test, one with fewer details so you can see an overview of dozens of test case outputs at once, and one with no details so you can count passes/fails as easily as possible
# [14:13] <hsivonen> DanC: speaking of thanksgiving: last week in the telecon, I said the survey could be answered in a week, but I forgot about thanksgiving
# [14:14] <DanC> yeah, tactical error by the chair
# [14:14] <DanC> should I extend the survey?
# [14:14] <DanC> our next telecon isn't 'till later
# [14:14] <hsivonen> DanC: extending would help, yes
# [14:14] <Philip> (http://philip.html5.org/tests/canvas/suite/tests/2d.gradient.interpolate.multiple.html vs http://philip.html5.org/tests/canvas/suite/tests/framed.2d.gradient.interpolate.multiple.html vs http://philip.html5.org/tests/canvas/suite/tests/minimal.2d.gradient.interpolate.multiple.html )
# [14:14] <Philip> (though the last of those won't work nicely in WebKit since it can't detect pass/fail)
# [14:16] <anne> WebKit should just implement getImageData() ... :)
# [14:16] <Philip> mjs: All of them (~550) are script-generated, from two input files
# [14:16] <zcorpan> Philip: i'm very impressed by your testing framework (with spec annotation and all)
# [14:17] <Philip> zcorpan: It's kind of crazy but I think it works alright for me so far :-)
# [14:19] <DanC> ok, help me remember later today to extend http://www.w3.org/2002/09/wbs/40318/req-gapi-canvas/ and the other one (and to announce the minutes from last telcon, including the date/time of next telcon)
# [14:19] <MikeSmith> DanC - calendar doesn't say webmaster is on vacation
# [14:20] <DanC> I got that from IJ's msg of 21 Nov 2007 12:57:41 +0000
# [14:29] <MikeSmith> ah
# [14:33] <hsivonen> MikeSmith: Validator.nu now pulls microsyntax UI text from the wiki: http://html5.validator.nu/?doc=http%3A%2F%2Fhsivonen.iki.fi%2Ftest%2Fmoz%2Felaboration-demo.xhtml (cached in RAM when the process starts; not updated like)
# [14:33] <hsivonen> s/like/live/
# [14:38] <hsivonen> Hixie: see above. This suggest that scraping rel registies or something like that from a wiki would be doable
# [14:38] <hsivonen> Hixie: although, an append-only Web service would still be more robust
# [14:40] <anne> i don't think it should be append-only
# [14:40] <MikeSmith> hsivonen - interesting. but I guess doable doesn't necessarily mean it's really a good way
# [14:42] <hsivonen> Hixie: is it intentional that the empty string is a valid browsing context name?
# [14:42] <zcorpan> hsivonen: nice!
# [14:43] * zcorpan might fill in the wiki page later
# [14:45] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [14:51] * Joins: gavin_ (gavin@99.227.30.12)
# [14:52] * Joins: aaronlev (chatzilla@66.30.196.151)
# [14:58] <mjs> hsivonen: I hope this isn't overly stating the obvious, but Error 1 should say which required attributes are missing
# [14:59] <mjs> listing all attributes isn't very helpful
# [14:59] * Joins: edaspet (edaspet@83.199.120.104)
# [15:00] * Joins: ganfset (edaspet@83.199.120.104)
# [15:00] <hsivonen> mjs: I'm hoping to get a fix for that from upstream
# [15:01] * Quits: edas (edaspet@82.120.100.44) (Ping timeout)
# [15:01] <hsivonen> mjs: also, the message used to be more helpful before Hixie removed the "(required)" annotation from the spec
# [15:01] <mjs> it's certainly improving
# [15:01] <mjs> making nice error messages is hard
# [15:02] <hsivonen> upstream being oNVDL as developed by SyncRO
# [15:02] <mjs> I should probably file my old complaints about <figure><legend> and <details><legend> as tracker ISSUEs
# [15:02] * Quits: edaspet (edaspet@83.199.120.104) (Ping timeout)
# [15:03] <MikeSmith> hsivonen - which really goes back to messages from jing, right?
# [15:03] <hsivonen> MikeSmith: yeah
# [15:04] * Quits: Lachy_ (Lachy@213.236.208.22) (Quit: Leaving)
# [15:08] <mjs> oh hey I missed http://www.w3.org/html/wg/tracker/issues/12
# [15:09] <mjs> I guess I should put in my 2c at some point
# [15:09] * Philip wonders if anyone mentioned http://canvex.lazyilluminati.com/misc/ref/ref.html in the recent discussion about HTML5 for authors
# [15:09] <mjs> though I don't think that should be an ISSUE at all, at least not on the HTML5 spec product
# [15:16] <anne> I think introducing some of the new elements simply requires changes in browsers
# [15:16] <anne> you will not be able to style them in all old UAs, but that shouldn't be much of an issue
# [15:16] <mjs> it violates the Degrade Gracefully principle
# [15:17] <mjs> especially given that using a new tag name instead of <legend> would work better
# [15:17] * Quits: ganfset (edaspet@83.199.120.104) (Ping timeout)
# [15:17] <mjs> especially since "legend" isn't even a very good match semantically
# [15:17] <mjs> and if used for <figure> because <caption> causes compatibility issues
# [15:18] <mjs> s/if/is/
# [15:18] <zcorpan> <fcaption>
# [15:18] <mjs> and for <details> it is just silly; if such a control has anything, it is a label or heading, certainly not a legend
# [15:19] <anne> it degrades gracefully
# [15:20] <anne> you get the content and all
# [15:20] <mjs> the author doesn't even have the possibility of styling the content to look decent
# [15:20] <mjs> in which case why use <figure> at all
# [15:21] <mjs> better to use <div class="figure"><span class="caption">
# [15:21] <anne> that was the case for <table> back then as well
# [15:21] <zcorpan> mjs: yeah, though currently that's better anyway because of ie
# [15:21] <mjs> <legend> has basically no advantage over using new elements
# [15:21] <mjs> see my thread on this
# [15:21] * zcorpan doesn't disagree btw
# [15:22] <anne> we could use <label>
# [15:22] <mjs> I think even <label> would be better than <legend> though it may interfere with existing stylesheets
# [15:22] <mjs> (which was Lachy's objection to it)
# [15:22] <zcorpan> might confuse some ATs too
# [15:22] <anne> * { } might interfere with anything we introduce
# [15:23] <anne> ATs don't ignore <label> when it's not associated with anything?
# [15:23] <zcorpan> haven't tested
# [15:23] <mjs> sure, but popular sites do style label { } without further qualification currently
# [15:23] <mjs> at least in my brief survey
# [15:23] <anne> yeah, that seems fine though
# [15:24] <mjs> though other sites use "form label" or ".someFormClass label"
# [15:24] <Philip> http://lists.w3.org/Archives/Public/public-html/2007Sep/0389.html
# [15:24] <anne> you add <details><label> and "details label" in your style sheet
# [15:24] <anne> or use the same style if you want that
# [15:24] <zcorpan> it looks weird though when you use <details> in forms for further <input>
# [15:25] <zcorpan> <details><label>More</label> <label>Foo: <input></label></details>
# [15:25] <anne> <div class=something> will always be more compatible than <something>, but I don't think that should be the only design input
# [15:25] <mjs> whole thread here: http://lists.w3.org/Archives/Public/public-html/2007Sep/0375.html
# [15:26] <mjs> we can give authors something that is more compatible in practice than <legend>
# [15:26] <zcorpan> <dcaption> and <fcaption> for <details> and <figure> respectively would work for me and make selectors less verbose in style sheets
# [15:26] <zcorpan> compare with <tbody>
# [15:27] <zcorpan> (it's not <tcaption> for tables though)
# [15:27] <mjs> The options I liked best from the original discussion were <figcaption> and <detailslabel>
# [15:27] <mjs> but f and d might be ok
# [15:28] <mjs> I think just <label> might be ok too but it will require authors to edit stylesheets to use it
# [15:28] <mjs> which doesn't seem like a huge deal to me but Lachy thought it was important
# [15:28] <mjs> anyway
# [15:28] <anne> i don't see the point
# [15:29] <anne> they have to edit the style sheet anyway
# [15:29] <anne> confusion with other <label> elements inside <details> seems a more legit point
# [15:29] * mjs is also sad that <https://bugzilla.mozilla.org/show_bug.cgi?id=311366> is not fixed
# [15:29] <mjs> I'd guess it probably won't make Firefox 3 at this point
# [15:29] <zcorpan> i don't like using <label> because the element we're after should have an explicit "contexts in which this element is allowed" (just like <caption> for tables isn't just a <p>)
# [15:30] <zcorpan> authors think "<label>, that's inline and is for forms"
# [15:30] <zcorpan> thus <details><label>...</label> <p>... *looks* wrong
# [15:31] <mjs> well, details is sort of a control
# [15:31] <Philip> HTML4 lets <map> be used with about the same semantics as HTML5's <nav>, but <nav> was added because overloading elements was considered undesirable
# [15:31] <mjs> it actually makes sense for its label to be a <label>
# [15:31] <zcorpan> Philip: indeed
# [15:32] <mjs> anyway I think new elements are fine too
# [15:32] <mjs> certainly better than <legend>
# [15:56] <MikeSmith> anne - Peter Karlsson not at Opera any longer?
# [15:57] <anne> correct
# [15:57] <anne> although I'm not sure about the exact date of leaving us
# [16:08] <hsivonen> are there hashed ID references pointing to anything but <map>s?
# [16:09] <anne> no
# [16:10] <hsivonen> the spec text could be simplified, then
# [16:12] <anne> it's interesting that "#foo bar" is a valid hashed ID reference
# [16:12] <anne> though the element it points to is in error
# [16:12] <hsivonen> yeah
# [16:13] * zcorpan updated http://simon.html5.org/specs/xml-stylesheet5
# [16:13] <hsivonen> also, I'm not sure what the validator should say about "#"
# [16:13] <anne> I wonder if Hixie tested references like "#foo%20bar"
# [16:13] <anne> hsivonen, compat talisman
# [16:14] <hsivonen> anne: compat talisman?
# [16:14] * anne was kidding
# [16:14] <hsivonen> anne: I mean usemap="#"
# [16:14] <hsivonen> what's the error? considering that id="" does not put an entry in the map
# [16:16] <anne> it doesn't point to an element
# [16:16] <anne> so that's the error
# [16:16] <anne> but that's per the impl algo
# [16:16] <hsivonen> ok
# [16:17] <anne> per the production algo there is some room for ambiguity I suppose
# [16:19] <hsivonen> I took the easy way out with dates and times (bad error messages)
# [16:19] <hsivonen> with integers and float, I try to have the patience to write proper error messages
# [16:21] <anne> integer: "One or more digits without a period. Optionally preceded by a minus."
# [16:22] <anne> maybe "One or more digits (0-9) without a period (.). Optionally preceded by a minus (-)."
# [16:22] <anne> and "For instance: 42"
# [16:22] * anne thinks there should be small examples too
# [16:23] <hsivonen> anne: yeah, not using Unicode names is probably better
# [16:23] * Joins: edas (edaspet@82.120.235.139)
# [16:23] <hsivonen> anne: it's a wiki :-)
# [16:23] <anne> yes, I'm going to edit now :)
# [16:33] * Joins: billmason (billmason@69.30.57.156)
# [16:34] <zcorpan> i'd like some opinions about xml-stylesheet. should i remove arbitrary restrictions in parsing? e.g. should the following PI be ignored? <?xml-stylesheet 1="2" href="test.css"?> (...because the first pseudo-attribute doesn't match the XML Attr production)
# [16:34] <zcorpan> that seems to be the case in ie, safari and opera, but firefox doesn't seem to mind
# [16:35] <zcorpan> i'm tempted to remove the arbitrary restrictions; it makes the spec simpler, for one :)
# [16:38] <anne> yeah, just say that specifications defining pseudo-attributes must make names that fall within a simple ASCII grammar
# [16:38] <anne> you could actually just allow [a-z-] only and parse based on that
# [16:39] <zcorpan> not doing so would be pretty bad language design, but is there a need for this spec to require it?
# [16:39] <anne> this would limit it more than XML, but that's reasonable
# [16:39] <zcorpan> what if a spec wants to use hyphens or 1-9? or some other character?
# [16:40] <anne> fail
# [16:40] <hsivonen> anne: thanks; <code> added
# [16:41] <zcorpan> anne: i'm not sure i like arbitrary restrictions; also, implementations ignore unknown pseudo-attributes but ignore the whole PI if there's a syntax error
# [16:43] <anne> it's not that arbitrary, it's in line with HTML, SVG, CSS, XBL, and MathML
# [16:43] <anne> either way works though
# [16:43] <anne> just define it :)
# [16:45] <hsivonen> Hixie: "non-zero" is redundant in "positive non-zero integers"
# [16:46] <hsivonen> Hixie: why aren't percentages conforming in attributes where they need to be parsed anyway?
# [16:47] <hsivonen> whoa. lots of new email
# [16:47] <MikeSmith> smedero++
# [16:50] <anne> what's wiki syntax for disabling autolinking?
# [16:50] * anne created text for IRIs
# [16:53] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [16:58] * Joins: gavin_ (gavin@99.227.30.12)
# [17:03] * Joins: timbl_ (timbl@209.6.134.246)
# [17:07] <gavin_> mjs: it wouldn't hurt to voice that opinion in the bug
# [17:07] <anne> didn't jonas already do that?
# [17:08] <gavin_> a little while ago, yeah
# [17:08] <gavin_> wouldn't hurt to poke him again :)
# [17:08] <gavin_> unfortunately Blake is rather busy with school at the moment, and has other bugs on his list that are more important
# [17:08] * anne would feel bad for adding "please fix this" to a bug
# [17:09] <gavin_> you could also email blake directly
# [17:12] * Quits: myakura (myakura@122.17.145.60) (Quit: Leaving...)
# [17:12] * Quits: aaronlev (chatzilla@66.30.196.151) (Quit: ChatZilla 0.9.79 [Firefox 3.0b2pre/2007111805])
# [17:13] <anne> ok, done
# [17:26] * Joins: edaspet (edaspet@82.120.235.139)
# [17:27] * Joins: ganfset (edaspet@82.120.235.139)
# [17:28] * Quits: edas (edaspet@82.120.235.139) (Ping timeout)
# [17:29] * Quits: edaspet (edaspet@82.120.235.139) (Ping timeout)
# [17:31] * Quits: ganfset (edaspet@82.120.235.139) (Ping timeout)
# [17:45] * Quits: Lachy (Lachlan@213.236.208.22) (Ping timeout)
# [17:52] * Joins: aaronlev (chatzilla@66.30.196.151)
# [17:52] * Joins: Sander (svl@86.87.68.167)
# [18:19] * Joins: Lachy (Lachlan@84.215.41.149)
# [18:20] * Quits: Lachy (Lachlan@84.215.41.149) (Quit: Leaving)
# [18:20] * Joins: Lachy (Lachlan@84.215.41.149)
# [18:41] * Quits: billmason (billmason@69.30.57.156) (Connection reset by peer)
# [18:43] <anne> "I object to having <canvas> in HTML 5. There is no vehicle to apply accessibility semantics to the markup - unlike SVG."
# [18:44] <MikeSmith> anne - yeah, was just reading that too
# [18:44] <anne> -- Richard Schwerdtfeger, IBM
# [18:44] * Joins: billmason (billmason@69.30.57.156)
# [18:44] <anne> I guess he hasn't read the draft
# [18:44] * Quits: Julian (chatzilla@217.91.35.233) (Ping timeout)
# [18:45] <MikeSmith> regardless, I wonder what those suggesting it be omitted expect will happen -- that if we omit it from the spec, the browsers that already support it will quit supporting it?
# [18:46] <anne> i dunno
# [18:47] <anne> this non-discussion feels like overhead to me and it's slowing us down
# [18:47] <anne> (even more!)
# [18:50] <zcorpan> anne: you need to work on your /ignore flag
# [18:53] <Philip> MikeSmith: If it's omitted, that leaves room for the HTML WG to invent a new better mechanism for the relevant use cases
# [18:53] <Philip> and then the browser developers will see the new mechanism is sufficiently better that they'll implement it, and everyone will start using it, and canvas will be unloved and eventually fade away when nobody important relies on it any more
# [18:54] <anne> the practice for not creating a link automatically is to wrap the http:// part as in <nowiki>http://</nowiki>
# [18:54] <zcorpan> the html wg can invent new better mechanisms anyway
# [18:55] * MikeSmith sheds a tear for the sad and lonely fate of canvas outlined in Philip's tale
# [18:55] * anne notes that the tale assumes there's success
# [18:55] <MikeSmith> a tragedy worthy of Shakespeared
# [18:56] <MikeSmith> what this working group needs is a Bard
# [18:56] <Philip> zcorpan: The HTML WG can't publish something with two incompatible mechanisms for immediate-mode graphics, because that'd be silly
# [18:56] <anne> MikeSmith: DanC? :)
# [18:56] <MikeSmith> anne - yeah, true. DanC got the guitar
# [18:57] <Philip> so one of them would have to be omitted before a new one can be brought in to replace it
# [18:58] <Philip> Alternatively, the HTML WG could omit canvas and then everyone would refer to the WHATWG superset spec instead
# [18:59] <DanC> at your service
# [18:59] <anne> and the HTML WG would need a new editor...
# [18:59] * DanC missed something... scrolls back more...
# [19:00] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [19:00] <DanC> MikeSmith, re canvas and accessibility, note that your input to the survey represents the whole W3C team, so you should talk with Michael Cooper and Al G. and such about accessibility
# [19:01] <MikeSmith> DanC - I've not voted on the survey
# [19:01] <DanC> but you will, right?
# [19:01] * Philip still wants 3D canvas, but doesn't know quite enough to form a coherent idea of how it should work :-(
# [19:01] <DanC> I think having declarative formats is good, but pretending javascript doesn't exists is silly
# [19:02] <MikeSmith> true that
# [19:02] <DanC> see also TAG finding on rule of least power
# [19:02] <MikeSmith> I don't intend to vote on the survey
# [19:02] <DanC> you don't intend to fulfil your obligation as team contact? that's a problem.
# [19:03] <DanC> if the position of the team is divided, you should report the divided positions and abstain. but to not answer is to neglect duties.
# [19:03] <MikeSmith> It's not my understanding that it's my role as team contact to attempt to represent that sentiments of the entire team in voting on specific surveys within the group
# [19:04] <DanC> well, then we should talk about your understanding of the team contact role.
# [19:04] <MikeSmith> I think it is fairly safe to say that position of the team is divided on this
# [19:04] <MikeSmith> If it's necessary for me to actually vote on the survey in order to indicate that, I'm happy to do it
# [19:04] <DanC> if that's safe to say, then say it.
# [19:05] <DanC> yes, it's necessary (or at least: customary and expected.)
# [19:05] <anne> Philip, what exactly is needed beyond checking if the input to ECMAScript calls won't crash OpenGL?
# [19:05] <DanC> in particular, MikeSmith , it's your role to represent TimBL in this WG. and the interaction domain.
# [19:06] <Philip> anne: Minimum hardware requirements are one issue
# [19:06] <DanC> so if you don't know how TimBL would answer, then you should talk to him
# [19:06] * Joins: gavin_ (gavin@99.227.30.12)
# [19:06] <DanC> perhaps by sending mail to www-tag
# [19:06] <Philip> e.g. programmable shaders are nice, and make the API much simpler, but some oldish desktop hardware and some current mobileish devices don't support that
# [19:07] <Philip> so the API design (even if it's just choosing between OpenGL ES 1.x and 2.x) will exclude some people from 3D content
# [19:08] <anne> that seems natural
# [19:09] <Philip> Also it'd be kind of nice if it didn't rely on implementors using OpenGL, since DirectX drivers often work much better on Windows
# [19:10] <Philip> (but then programmable shaders use an OpenGL-specific language, and it's kind of unpleasant to automatically translate them into the DirectX-specific language)
# [19:12] * Quits: heycam (cam@210.84.13.228) (Ping timeout)
# [19:16] <zcorpan> Philip: we currently have two incompatible mechanisms for something else... data templates and repetition model
# [19:16] * Quits: Lachy (Lachlan@84.215.41.149) (Quit: Leaving)
# [19:17] * Quits: timbl_ (timbl@209.6.134.246) (Connection reset by peer)
# [19:17] * Joins: timbl_ (timbl@209.6.134.246)
# [19:17] * Joins: Lachy (Lachlan@84.215.41.149)
# [19:19] <Philip> (Also OpenGL 3 is significantly redesigning its API, so it'd be kind of silly to adopt the OpenGL 1.x API just as it's being abandoned by the rest of the world)
# [19:20] <anne> must be nice if you can just break everything each time around
# [19:21] <Philip> The desktop driver developers will have to support the old OpenGL API forever, so it's not perfectly nice
# [19:23] <Philip> and OpenGL 2.1 still supports all the stuff from OpenGL 1.0
# [19:24] <Philip> I don't know if GL3 will still provide the old API functions for compatibility or not
# [19:25] <Philip> but DirectX seems to have been reasonably successful in just providing versioned interfaces, so they can remove stuff with each new version
# [19:28] * Joins: heycam (cam@203.217.79.225)
# [19:45] * Joins: Julian (chatzilla@80.143.134.49)
# [20:05] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
# [20:24] * Quits: zcorpan (zcorpan@88.131.66.80) (Ping timeout)
# [21:00] * Quits: aaronlev (chatzilla@66.30.196.151) (Ping timeout)
# [21:08] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
# [21:08] * Quits: gavin_ (gavin@99.227.30.12) (Ping timeout)
# [21:13] * Joins: gavin_ (gavin@99.227.30.12)
# [21:14] * Joins: DanC_lap (connolly@128.30.52.30)
# [21:20] * Joins: jgraham_ (james@81.86.210.2)
# [21:27] * Joins: aaronlev (chatzilla@146.115.126.114)
# [21:34] * Quits: DanC_lap (connolly@128.30.52.30) (Ping timeout)
# [21:39] * Quits: heycam (cam@203.217.79.225) (Ping timeout)
# [22:12] <anne> DanC, yt?
# [22:12] <anne> hsivonen, too much e-mail to public-html? :)
# [22:13] <hsivonen> anne: yeah, and I don't feel like debating the status of WF2
# [22:20] * Quits: ROBOd (robod@89.122.216.38) (Quit: http://www.robodesign.ro )
# [22:27] * Joins: Zeros (Zeros-Elip@67.154.87.254)
# [22:33] * Joins: heycam (cam@130.194.72.84)
# [22:34] * Quits: Julian (chatzilla@80.143.134.49) (Ping timeout)
# [22:43] * Quits: aaronlev (chatzilla@146.115.126.114) (Ping timeout)
# [23:16] <mjs> gavin_: I would not want to nag, especially since Boris on Jonas already did (also am surprised that Boris and/or Jonas could not just do it themselves, they seem like heavy hitters)
# [23:16] <mjs> gavin_: or I could even code it myself but I am not inclined to fix Gecko bugs
# [23:17] <gavin> I'm sure Jonas or Boris could do it themselves, if they had no other work to do :)
# [23:18] <gavin> I really don't know how much work it involves, but if it were easy I'd imagine it would have been done by now
# [23:27] <anne> hsivonen, re: #whatwg, mailto: allows everything, such as 'mailto:"Anne van Kesteren" <annevk@opera.com>'
# [23:27] <anne> iirc
# [23:27] <Hixie> does that work anywhere?
# [23:27] <Hixie> and how does that work with the "?body=..." stuff?
# [23:27] * jgraham_ looked at the gecko bug but didn't manage to work out how residual style handling was supposed to work
# [23:28] <hsivonen> anne: ok. I wonder if a mailto5 is needed
# [23:28] <anne> Hixie, data:text/html,<a href='mailto:"Anne van Kesteren <annevk@opera.com>"'>test</a>
# [23:28] <anne> wfm in Opera at least
# [23:28] <Hixie> oh i meant what does the spec require
# [23:28] <Hixie> but good to know that it works in opera
# [23:28] <Hixie> doesn't work at all in firefox
# [23:29] <Hixie> it started an e-mail to |"Anne van Kesteren <annevk@opera.com>"@ariel.corp.google.com|
# [23:29] <anne> interesting
# [23:29] <Hixie> though i guess that could also be just a bug in pine
# [23:29] <Hixie> dunno who is supposed to do the parsing
# [23:29] <Hixie> i guess pine
# [23:30] <anne> yeah, iirc everything after mailto: is handed to the e-mail app
# [23:30] <anne> except maybe for percent encoding oddities
# [23:30] <anne> someone did research into that and filed a few bugs on browsers and even worked on a RFC
# [23:31] <mjs> gavin_: making a similar change in WebKit would be a matter of a day or two's work at most, but perhaps Gecko's parser is harier
# [23:31] <gavin> I believe that to be the case ;)
# [23:31] * Quits: Thezilch (fuz007@68.54.228.249) (Ping timeout)
# [23:32] <mjs> I imagine there is some rule to make a block-level element close pending inlines, unknown elements would just have to be changed to be treated like <div> instead of like <i> or <span>
# [23:32] * Hixie wonders what mjs and gavin are talking about
# [23:32] <mjs> Hixie: https://bugzilla.mozilla.org/show_bug.cgi?id=311366
# [23:32] <mjs> and likely challenge level thereof
# [23:33] <Hixie> aah
# [23:35] <hsivonen> anne: someone being Michael Puls II?
# [23:36] <anne> yeah
# [23:37] <mjs> (did anyone but me notice that Dean Edridge's trivial sample double-markup document is non-conforming?)
# [23:37] <hsivonen> anne: do you think integers need to have "without a period"?
# [23:37] <mjs> (even trivial cases are hard to get right)
# [23:38] <Hixie> mjs: you mean the xml:lang thing, or something else?
# [23:38] <hsivonen> anne: that is, wouldn't it be better to be silent about what isn't supposed to be there?
# [23:38] <hsivonen> mjs: I did
# [23:39] <mjs> both xml:lang and lang
# [23:40] * Joins: marcos (root@124.171.166.8)
# [23:40] <anne> hsivonen, yeah, it can be removed
# [23:40] <hsivonen> anne: ok
# [23:42] <hsivonen> summarizing an RFC in one paragraph is a fun writing task
# [23:43] <anne> just give examples and a pointer
# [23:51] * Quits: marcos (root@124.171.166.8) (Ping timeout)
# Session Close: Thu Nov 22 00:00:00 2007

The end :)