Options:
- # Session Start: Tue Jan 27 00:00:00 2009
- # Session Ident: #html-wg
- # [00:13] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
- # [00:15] * Quits: MichaelC (Michael@128.30.52.30) (Quit: ChatZilla 0.9.84 [Firefox 3.0.5/2008120122])
- # [00:49] <Philip> MikeSmith: About putting tests on dev.w3.org: Is there some existing infrastructure that could be used for running cross-domain tests?
- # [00:50] * karl didn't know there was a new chair or maybe staff contact in the group - "With this, I am happy to publish this working draft as a WD." http://lists.w3.org/Archives/Public/public-html/2009Jan/0367.html
- # [00:50] <Hixie> ?
- # [00:51] <anne> karl, Sam asked for objections against publishing, Hixie gave one, MikeSmith addressed it, Hixie was ok with the change
- # [00:52] <MikeSmith> Philip: If you mean a test harness that supports such, I don't know. Maybe the CSS WG or mobile-Web test suites WG have something.
- # [00:52] <anne> CSS WG does not
- # [00:52] <karl> anne: that might be my loosy English grammar :)
- # [00:53] <anne> ah, I see what you mean now, I don't think Hixie intended to say that :)
- # [00:53] <Philip> MikeSmith: Not necessarily a harness, just a domain that is not dev.w3.org but can serve the same files
- # [00:53] <MikeSmith> karl: yeah, I think you could just read that as "to have this published as..." or whatever
- # [00:54] <karl> MikeSmith: this makes more sense indeed
- # [00:54] <MikeSmith> Philip: I see. I would think we could get something set up as needed
- # [00:54] <gavin> "working draft as a WD" is funny
- # [00:54] <anne> different ports and http vs https would be cool as well
- # [00:55] <Philip> MikeSmith: It's not something I'd care much about myself; just a thought I had when seeing Microsoft's tests using www.asp.net as well as samples.msdn.microsoft.com
- # [00:55] <MikeSmith> ah
- # [00:55] <Philip> anne: That doesn't sound like something that would make the W3C sysadmin people very happy :-)
- # [00:55] <Hixie> karl: see /msg
- # [00:56] <MikeSmith> Well, I would vote for starting to get the tests checked in and sorting out the problems as we need
- # [00:57] <MikeSmith> I'm recalling now that part of the reason we never made muhc progress on getting tests checked it is that we started doing a lot of hand-wringing about how we should structure the subdirs and what harness to use and even what version-control system
- # [00:57] <Hixie> imho the solution to that kind of problem is to assign a test "czar" and give them full authority and responsibility
- # [00:58] <Philip> I thought the reason we didn't make much progress was that nobody had tests that they wanted to bother checking in to the W3C system
- # [00:58] <Hixie> i've seen this kind of thing (specifically with test suites even) half a dozen times at least, every time there's a committee involved it devolves into a rat hole with no progress
- # [00:58] <anne> there was fuss over the things MikeSmith mentions as well
- # [00:59] <Philip> At least that's my reason - I have a load of canvas tests but I don't currently see any value in moving them off philip.html5.org, because anyone who wants to use them already knows where they are and nobody wants to review or modify them
- # [01:00] <Dashiva> And then you get hit by a bus one day
- # [01:00] <Philip> Then someone can download the source code that's also hosted on philip.html5.org and move it elsewhere :-)
- # [01:01] <Philip> Also, if I was hit by a bus then I wouldn't care about canvas tests
- # [01:01] <anne> i was hoping we could some day merge the various <canvas> test suites, but maybe that is not necessary
- # [01:01] <karl> Philip: what would give value to your tests?
- # [01:02] <karl> Would a simple mirroring be ok? with the license agreement which goes with it
- # [01:03] <Philip> anne: I've never worried much about what the other test suites are doing, since I've just attempted to comprehensively test everything I can think of and hopefully it covers everything the others are doing
- # [01:03] <Philip> though in practice that probably doesn't work perfectly
- # [01:04] * Hixie wonders why karl is ignoring him
- # [01:04] <anne> it might be too much effort to just find out
- # [01:04] <karl> anne: possibly
- # [01:04] <Philip> In particular, it might be nice to look at other browsers' test suites (i.e. WebKit's) since they're likely to have regression tests for real problems
- # [01:05] <Philip> karl: I'm not sure what would give value, and I don't see any particular benefits in mirroring on w3.org
- # [01:05] <Philip> I'm not at all opposed to doing things in the W3C space, I just don't care enough to do anything about it :-)
- # [01:06] <MikeSmith> Philip: at the time we set up the CVS test dir, Maciej said he had a number of tests he would really like to check in somewhere
- # [01:06] <MikeSmith> Hixie: I agree about having a test czar. Problem is getting someone to volunteer to do it.
- # [01:07] <Hixie> MikeSmith: yeah, not sure what to do about that. there is a dearth of volunteers.
- # [01:07] <MikeSmith> yep
- # [01:07] <Hixie> MikeSmith: though apparently not for writing the about: spec, which got about half a dozen volunteers in about an hour
- # [01:07] <Hixie> go figure
- # [01:07] <karl> Philip: what I can see but not on your side indeed, is: w3c uri policy, unique point of information for devs who might not know your site, and the hit by a bus factor (which I do not wish)
- # [01:07] <MikeSmith> heh
- # [01:07] <Hixie> (exaggerating a bit though not much!)
- # [01:08] <MikeSmith> I think at one time DanC was keen on getting work going on tests. I don't know if leading the test-suite work would fit in with his current responsibilities
- # [01:08] <karl> MikeSmith: hehe, DanC is a test gourmet
- # [01:08] <Hixie> he mentioned that he was interpreting his contract with adobe as license to write tests, iirc
- # [01:09] <Hixie> so he might be in a good position to do so
- # [01:09] * Parts: hyatt (hyatt@98.201.21.231)
- # [01:11] <takkaria> what would be being a test czar entail?
- # [01:11] <MikeSmith> takkaria: deciding what the structure for the tests directories is, deciding what test harness to use
- # [01:12] <Philip> Does the czar get to invade other Working Groups to expand his empire?
- # [01:13] <karl> Philip: only in bad movies and cheap novels
- # [01:13] <Philip> s/harness/harnesses/ hopefully
- # [01:14] <MikeSmith> Philip: no, just the normal lobbing of missiles at each other, with demands to the UN to issue formal statements that everybody then subsequently ignores
- # [01:14] <MikeSmith> our core problem is that we are mostly an all-volunteer army
- # [01:15] <MikeSmith> what we really need is a Web-technologies conscription/draft system
- # [01:15] <MikeSmith> draft smart Web developers into 6 months of public service as an editor
- # [01:16] <Philip> Most armies work alright without conscription because they pay their volunteers
- # [01:16] <Philip> Also they get cool uniforms, and can shoot people
- # [01:16] <takkaria> the test czar could get a tshirt saying as much
- # [01:16] <Philip> I think those are three pillars that the W3C should adopt
- # [01:17] <takkaria> and can shout at web browser vendors, which is almost like shooting them
- # [01:19] <Philip> takkaria: If you consider shouting at someone to be "almost like" shooting them, I would hate to imagine you in the crowd at a football match
- # [01:20] <Hixie> he'd fit right in in the UK, no?
- # [01:20] * takkaria grins
- # [01:21] <takkaria> well, I might be interested in doing testsuite management, but I haven't got much time for that in the next few weeks
- # [01:23] <MikeSmith> takkaria: how long is a few weeks? 3 weeks?
- # [01:24] <MikeSmith> W3C: Cash. Fashion. Guns.
- # [01:24] <takkaria> something like that. realistically, second half of february / beginning of march
- # [01:25] * karl has Morricone music in the head now. thanks MikeSmith
- # [01:27] <Philip> MikeSmith: Sounds like a good premise for the next in the GTA series
- # [01:27] <MikeSmith> takkaria: If you can commit to that, I will suggest to the chairs that they take you up on the offer.
- # [01:27] <takkaria> is the w3c exclusively-cvs? the thought of managing testsuite directory structures in cvs fills me with dread
- # [01:28] <MikeSmith> takkaria: yeah, W3C is currently CVS-only -- at least as far as what is supported by the systeam
- # [01:29] <MikeSmith> though I think DanC has a mercurial or git server running somewhere
- # [01:29] <takkaria> hmm
- # [01:29] <MikeSmith> I share your dread about working with CVS, but not much we can do about it in the short term
- # [01:30] <MikeSmith> we go to war with the version-control system that we have, not the version-control system that we'd like to have
- # [01:30] <takkaria> :)
- # [01:30] <MikeSmith> Philip: as far as uniforms, we already have those:
- # [01:30] <MikeSmith> http://www.flickr.com/photos/sideshowbarker/2067567428/
- # [01:30] <pimpbot> Title: SW3C1Ws on Flickr - Photo Sharing! (at www.flickr.com)
- # [01:30] <takkaria> well, I'd be happy to put my name down
- # [01:30] <MikeSmith> takkaria: cool
- # [01:30] <takkaria> though I'll have to talk to the chairs about what they would need me to do--I'm sure I can pick it up
- # [01:31] <MikeSmith> OK, I think we can take discussion about it to the list. But I will bring it up with Sam and Chris directly as well.
- # [01:32] * takkaria nods
- # [01:33] <Philip> The pain of organising complex hierarchies in CVS sounds like it could be greater than the pain of switching to a not-yet-supported VCS
- # [01:35] <Philip> (particularly since DVCSs seem to make it relatively non-painful to work without a centralised repository, so it'd be alright at first even if the W3C couldn't actually host the files itself)
- # [01:35] <takkaria> something like using git for a masster copy which committed to cvs whenever it was updated would probably make life easier
- # [01:36] * Joins: jun (fujisawa_j@114.48.147.199)
- # [01:36] * Parts: jun (fujisawa_j@114.48.147.199)
- # [01:36] <MikeSmith> fwiw, the team well understands that there are better version-control systems that CVS, and the plan is not to stay with the CVS forever
- # [01:36] <MikeSmith> in the mean time, we need to work within the limitations of what we have at hand
- # [01:37] <Philip> Do our limitations have to include relying on the W3C's facilities?
- # [01:37] <MikeSmith> Philip: no, not necessarily
- # [01:37] <MikeSmith> takkaria's suggestion about using git and pushing the commits to CVS would be fine by me
- # [01:38] * Philip remembers seeing lots of XHTML2 stuff on aptest.com, so it doesn't seem that uncommon to rely on external services
- # [01:38] <Philip> That sounds reasonable to me
- # [01:39] <Philip> though is it worth pushing the commits to CVS, rather than just uploading them as static files into a web server directory somewhere?
- # [01:40] <takkaria> that's a fair point
- # [01:41] <MikeSmith> I guess static copies would work fine, though I can imagine some people might be happy to have them available in CVS too
- # [01:41] <Philip> Maybe it would be worthwhile just so people who don't understand git could do "cvs up; edit some test file; cvs diff > patch-to-send-to-mailing-list"
- # [01:42] <MikeSmith> I will say that one advantage of just using CVS would be that it's a lower barrier of entry for potential contributors
- # [01:42] <MikeSmith> in that there are more people who easy access/setup for CVS client tools than for git
- # [01:43] <takkaria> hmm
- # [01:43] <takkaria> I'm not so sure that's quite so true these days
- # [01:43] <takkaria> though I guess in companies it might be
- # [01:43] <takkaria> e.g. Windows has git with a GUI these days
- # [01:43] <MikeSmith> I do have to say that I can see git or some other distributed VCS being much more valuable for projects where you need to do a lot of branching and merging, and need to do it quickly and easily
- # [01:44] <takkaria> and 'git clone branch; edit some test files; git diff > patch' is not more complex than cvs
- # [01:44] <MikeSmith> true
- # [01:45] <Philip> It's not more complex but it's different, and people already understand cvs and have it installed, so it's at least twice as much work if they have to have git too
- # [01:45] * takkaria nods
- # [01:45] <takkaria> well, I don't see there'd be any problem having a cvs mirror around
- # [01:46] <MikeSmith> takkaria: anyway, this is exactly why we need a test czar
- # [01:46] <MikeSmith> some people will object to whatever ends up getting decided
- # [01:47] <MikeSmith> and in the end, the choice of VCS and directory structure is not critical to the success of the work
- # [01:48] * Philip isn't sure that's necessarily true
- # [01:48] <Philip> The success of the work is dependent on volunteers contributing to the work, and that is strongly dependent on how easy it is to contribute
- # [01:48] <MikeSmith> so it's more important that whatever choice is made is what makes it easiest for the person(s) leading and actually doing the work to manage it in the way that works best for them
- # [01:49] <Philip> and issues like VCS can have a fairly significant effect on how easy it is to contribute, and how much time and effort you have to waste trying to get all the latest versions and create patches and review things without getting lost
- # [01:50] * Philip would agree that it makes sense for decisions to be made only by the people who will be directly affected by those decisions
- # [01:51] <MikeSmith> The problem is everybody has different opinions about what constitutes "easiest".
- # [01:52] <Philip> You can solve that problem by beating up everyone who is wrong
- # [01:52] <Lachy> that's only a problem if you consider everyone's opinion but your own to be equally valid
- # [01:52] <MikeSmith> Fortunately, we can migrate among VCS systems later as needed (if people directly involved in the work decide they want to)
- # [01:52] <takkaria> well, from my experience running an open-source game, I would say that the VCS choice is not all that big a deal these days
- # [01:53] <takkaria> people who want to contribute to things will go to reasonable effort to do so
- # [01:53] <MikeSmith> right
- # [01:54] <takkaria> and there is definitely at least one DVCS which is easily obtainable on all of {Windows, OS X, Unix}
- # [01:54] * Philip 's experience from a not-quite-open-source game is that it is a big deal, because the artists have learnt how to use TortoiseSVN and won't understand anything else :-)
- # [01:55] <takkaria> heh
- # [01:55] <takkaria> there's TortoiseDarcs and TortoiseGit and TortoiseTheMoonOnAStick these days
- # [01:56] <Philip> Are they anything like TortoiseCVS/TortoiseSVN, or are they just random GUIs with "Tortoise" stuck in front of their names?
- # [01:56] <takkaria> yeah, branched from the same code
- # [01:56] <takkaria> so essentially the same interface
- # [01:56] <Philip> Ah, right
- # [01:56] <Philip> Do they still have a nice simple right click -> "update", right click -> "commit" model?
- # [01:56] <takkaria> yeah
- # [01:57] <Philip> rather than the crazy checkout/pull/update/clone/branch/whatever model that the command-line tools tend to favour
- # [01:58] <takkaria> also, we're dealing with people who are competent enough to write browser tests
- # [01:59] <Philip> (I think DVCS wouldn't work on my game anyway because there's 5GB of stuff in the repository and nobody is going to want to have a clone of it all :-( )
- # [02:12] <MikeSmith> takkaria: you make a good point -- it's hard imagine that somebody competent enough to write a browser test that will pass review will not be competent enough able to install and learn whatever VCS client they need
- # [02:13] <MikeSmith> which reminds me: the other thing you'd need to decide on is how to handle review
- # [02:14] <MikeSmith> I think we had been discussing before that ideally nothing should get checked in unless it had been reviewed first
- # [02:14] <MikeSmith> http://dev.w3.org/html5/tests/readme.txt
- # [02:17] * Quits: dbaron (dbaron@63.245.220.241) (Quit: 8403864 bytes have been tenured, next gc will be global.)
- # [02:18] * Quits: sryo (sryo@190.245.204.198) (Ping timeout)
- # [02:27] * Joins: dbaron (dbaron@63.245.220.241)
- # [02:32] * Quits: Lachy (Lachlan@85.196.122.246) (Quit: Leaving)
- # [02:32] * Joins: Lachy (Lachlan@85.196.122.246)
- # [02:34] <takkaria> yeah, wfm
- # [03:04] * Quits: adele (adele@17.244.18.87) (Quit: adele)
- # [03:32] * Quits: dbaron (dbaron@63.245.220.241) (Quit: 8403864 bytes have been tenured, next gc will be global.)
- # [04:23] * Joins: Zeros (Zeros-Elip@68.50.195.181)
- # [04:24] * Quits: Zeros (Zeros-Elip@68.50.195.181) (Quit: Leaving)
- # [04:58] * Joins: billyjackass (MikeSmith@mcclure.w3.org)
- # [05:04] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Ping timeout)
- # [05:08] * billyjackass is now known as MikeSmith
- # [05:18] * Joins: adele (adele@24.7.123.50)
- # [05:18] * Quits: adele (adele@24.7.123.50) (Client exited)
- # [05:18] * Joins: adele (adele@24.7.123.50)
- # [06:15] * Quits: heycam (cam@130.194.72.84) (Quit: bye)
- # [06:20] * Quits: hober (ted@206.212.254.2) (Client exited)
- # [06:20] * Joins: hober (ted@206.212.254.2)
- # [06:20] * Quits: gavin_ (gavin@99.226.207.11) (Ping timeout)
- # [06:25] * Joins: gavin_ (gavin@99.226.207.11)
- # [06:33] * Joins: heycam (cam@124.168.42.25)
- # [06:38] * Quits: takkaria (takkaria@208.75.87.166) (Quit: Reconnecting)
- # [06:38] * Joins: takkaria (takkaria@208.75.87.166)
- # [06:39] * Joins: Zeros (Zeros-Elip@68.50.195.181)
- # [06:44] * Quits: hober (ted@206.212.254.2) (Quit: ERC Version 5.3 (IRC client for Emacs))
- # [08:28] * Quits: Zeros (Zeros-Elip@68.50.195.181) (Quit: Leaving)
- # [08:45] * Joins: zcorpan (zcorpan@88.131.66.80)
- # [09:14] * Quits: marcos (marcos@87.196.212.70) (Quit: marcos)
- # [09:23] * Quits: adele (adele@24.7.123.50) (Quit: adele)
- # [09:41] * Joins: tlr (tlr@128.30.52.30)
- # [09:44] * Joins: darobin (robinb@81.66.15.230)
- # [10:21] * Joins: gsnedders_ (gsnedders@86.136.52.180)
- # [10:21] * Quits: gsnedders (gsnedders@86.136.52.180) (Connection reset by peer)
- # [10:41] * Quits: anne (annevk@77.163.243.203) (Client exited)
- # [10:42] * Joins: anne (annevk@77.163.243.203)
- # [10:49] * Quits: darobin (robinb@81.66.15.230) (Ping timeout)
- # [11:04] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Ping timeout)
- # [11:05] * Joins: ROBOd (robod@89.122.216.38)
- # [11:12] * Quits: anne (annevk@77.163.243.203) (Client exited)
- # [11:12] * Joins: anne (annevk@77.163.243.203)
- # [11:32] <pimpbot> bugmail: "[Bug 6474] New: Event handler DOM attributes on HTMLDocument" ( message in thread) <http://lists.w3.org/Archives/Public/public-html-bugzilla/2009Jan/0038.html>
- # [11:33] * Joins: darobin (robinb@82.233.247.234)
- # [11:42] * Quits: Lachy (Lachlan@85.196.122.246) (Quit: This computer has gone to sleep)
- # [11:47] <jgraham> FWIW one possible other way to handle review is to have some way of checking in unreviewed stuff and then some way of marking it reviewed. Could use some sort of extended metadata (I have no idea what is actually supported by the various DVCSs) or just parallel directory trees with reviewws/unreviewed
- # [11:50] <Philip> I would assume the DVCS way is to add tests into a branch, and then merge them across when they're reviewed
- # [11:51] <Philip> but I have no idea if that's sensible or correct or the only way to do it
- # [11:52] <jgraham> Yeah, I'm not sure what the best way to do it is either but the idea is to make contributing tests easy (there is no need to jump through hoops looking for review) and to make reviewing tests easy (there is a big pool of tests that need review)
- # [11:53] <jgraham> It will also highlight when we have the problem that there are lots of tests that have gone unreviewed
- # [11:53] * Joins: Lachy (Lachlan@213.236.208.22)
- # [11:55] <Philip> I suppose there's an issue of making it easy to run unreviewed tests, vs not making it easy but being prompt at reviewing tests so people don't want to run unreviewed ones
- # [11:57] * Quits: Lachy (Lachlan@213.236.208.22) (Quit: Leaving)
- # [11:57] * Joins: Lachy (Lachlan@213.236.208.22)
- # [11:57] <jgraham> I don't think people running unreviewed tests is a problem as long as they know what they are doing
- # [11:57] <Philip> I'd be happy as long as there's not a process like "send each test case individually to the mailing list for review" when I've got seven hundred canvas tests :-)
- # [11:58] <jgraham> In any case anything that has the effect of making the barrier to entry for tests smaller is good
- # [11:58] * jgraham doesn't want to review 700 Canvas tests
- # [12:00] <Philip> It's not that bad, they're mostly only a few lines each :-)
- # [12:00] <jgraham> Still a few * 700 = many
- # [12:02] <Philip> The source file for all the tests is only, uh, ten thousand lines
- # [12:03] <Philip> Okay, maybe it's quite a bit
- # [12:05] <Philip> But I don't care much about review because I'm happy to wait until someone (e.g. me or an implementer) notices unexpected failures and investigates the relevant tests; most of the rest already work fine so there's not much value in checking them all again
- # [12:07] <jgraham> Philip: Hence the value in running unreviewed tests :)
- # [12:08] <Philip> I see your point :-)
- # [12:11] <Lachy> Philip, yeah, that method works as long as the both the test and the implementation are wrong for the same reason, in which case you get false positives
- # [12:11] <Lachy> *aren't wrong
- # [12:12] <Lachy> although I guess the test would get caught when a non-buggy implementation runs the test and fails
- # [12:14] <jgraham> Lachy: It is unreasonable to expect tests, even reviewed ones, to be bug free, so there will always be bugs caught be implementations anyway
- # [12:14] <Lachy> that's true.
- # [12:15] <Lachy> the process should just allow people who are experienced at writing good test cases to submit without review, whereas those from people less experienced should need review until they've shown that they do a good job
- # [12:17] <Lachy> that way the bulk of the tests will get checked in without too much bureaucracy and we still end up with reasonable quality TCs
- # [12:17] <Lachy> however, there's still the problem created by the spec changing after a TC was written. It often takes a while to notice the out of date tests.
- # [12:19] <jgraham> If you could rely on the ids in the spec being stable TCs could have the ID in them somewhere and get marked unreviewed again when that section of the spec changed. It would suck for editorial changes though
- # [12:19] <hsivonen> it would be nice if TCs had stable URIs
- # [12:20] <hsivonen> in order to use that also for validator stress testing
- # [12:22] * jgraham onders if someone should start making a list of requirements
- # [12:22] <jgraham> *wonders
- # [12:23] * Quits: Julian (chatzilla@217.91.35.233) (Connection reset by peer)
- # [12:30] * Joins: myakura (myakura@221.191.119.20)
- # [12:32] * Joins: ed_ (ed@88.131.66.80)
- # [12:33] <pimpbot> bugmail: "[Bug 6475] New: Missing event handler attributes" (1 message in thread) <http://lists.w3.org/Archives/Public/public-html-bugzilla/2009Jan/0039.html>
- # [12:33] * Quits: ed_work (ed@88.131.66.80) (Ping timeout)
- # [12:56] * Joins: maddiin (mc@87.185.243.173)
- # [13:03] <Philip> Lachy: It works as long as the test and *all* implementations are wrong for the same reason; and in that case, we can just define that they're right and the spec is wrong
- # [13:04] <Philip> For stability, I have a file containing extracted sentences from the spec and giving an ID to each, and then each test case lists the IDs of the sentences that are relevant to that test
- # [13:05] <Philip> so if the spec changes, the extracted sentences will no longer match and the conversion tool will complain, and I can find everything that refers to it and update them
- # [13:05] <Philip> and it at least works in fairly straightforward cases (though major section-wide changes need more manual work)
- # [13:06] <Philip> (but there aren't that many test cases for each section, so it's not too much of an issue)
- # [13:21] * Quits: jwatt (roslea@83.87.4.17) (Ping timeout)
- # [13:34] * Quits: karl (karlcow@70.81.32.202) (Quit: O public road, I say back I am not afraid to leave you, yet I love you, you express me better than I can express myself.)
- # [14:00] * Joins: Julian (chatzilla@217.91.35.233)
- # [14:03] * Joins: MichaelC (Michael@128.30.52.30)
- # [14:32] * tlr is now known as tlr-bbl
- # [14:33] * Quits: Julian (chatzilla@217.91.35.233) (Client exited)
- # [14:34] * Joins: Julian (chatzilla@217.91.35.233)
- # [15:02] * Joins: karl (karlcow@70.81.32.202)
- # [15:09] <anne> karl, heh, that's true, no more studies to class values now, they ruined that game :)
- # [15:09] <anne> though for HTML6 it might work again I suppose
- # [15:12] * Quits: Julian (chatzilla@217.91.35.233) (Connection reset by peer)
- # [15:20] * Joins: Julian (chatzilla@217.91.35.233)
- # [15:21] * tlr-bbl is now known as tlr
- # [15:28] <karl> anne: ;) community effect.
- # [15:33] <pimpbot> bugmail: "[Bug 6476] New: cross-origin media element loads and progress events" (1 message in thread) <http://lists.w3.org/Archives/Public/public-html-bugzilla/2009Jan/0040.html>
- # [16:04] <pimpbot> bugmail: "[Bug 6477] New: HTMLInputElement.width and height" (1 message in thread) <http://lists.w3.org/Archives/Public/public-html-bugzilla/2009Jan/0041.html>
- # [16:08] * Joins: rubys (rubys@75.194.128.17)
- # [16:08] * Joins: aroben (aroben@71.58.73.153)
- # [16:18] * Quits: rubys (rubys@75.194.128.17) (Ping timeout)
- # [16:19] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [16:24] <jgraham> Hmm. If the tracker is now characterised as the way to provide feedback on proposals I think something has changed again
- # [16:29] <DanC> certainly not "the" way; jgraham , what are you referring to?
- # [16:32] <jgraham> rubys's latest email
- # [16:32] <jgraham> ">and how feedback to such a proposal is collected.
- # [16:32] <jgraham> http://www.w3.org/html/wg/tracker/"
- # [16:42] <takkaria> hm, I guess a requirements doc would be useful for testsuite stuff
- # [16:49] <DanC> my attention is divided, jgraham ; thanks for the excerpt; I hope to look at it more soonish...
- # [16:51] * Quits: gsnedders_ (gsnedders@86.136.52.180) (Quit: gsnedders_)
- # [16:58] <takkaria> what does the W3C regarding licence on testsuites?
- # [16:58] * Joins: marcos (marcos@87.196.212.70)
- # [16:59] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Client exited)
- # [16:59] <anne> we could use MIT for HTML WG tests for now per rigo per last Boston TP meeting
- # [17:00] <takkaria> ah, that's good
- # [17:00] <takkaria> PS: http://wiki.whatwg.org/wiki/Testsuite. if you have any more requirements, please edit there or talk here
- # [17:00] <pimpbot> Title: Testsuite. - WHATWG Wiki (at wiki.whatwg.org)
- # [17:02] <Philip> Should we publish test results too?
- # [17:02] <Philip> (Not necessarily running the tests - just collecting and presenting the results after other people have done all the tedious work)
- # [17:03] <takkaria> I guess part of that is making the tests easy to run so that data can be collected either
- # [17:03] <takkaria> *easily
- # [17:04] * Philip likes having pretty colours like in http://philip.html5.org/tests/canvas/suite/tests/results.html
- # [17:05] <anne> takkaria, I just added something for that
- # [17:05] <anne> takkaria, when possible tests should just use scripting
- # [17:05] <Philip> (and gathering the results involves opening one or four pages, waiting for the browser to have a bit of a think, and then clicking a button to upload to a CGI script, which is easy enough for me to bother collecting results)
- # [17:06] * Joins: scotfl (scotfl@70.64.14.62)
- # [17:06] * Quits: scotfl (scotfl@70.64.14.62) (Quit: scotfl)
- # [17:06] <takkaria> anne: thanks
- # [17:07] <jgraham> takkaria: The main requirement is how to refer to you. Test Czar is somewhat passe so I suggest "Supreme Emporer of All Testing, May All His PASS Shine Green For a Thouand Years"
- # [17:07] <Philip> Could that page have non-requirements too, e.g. it's not a requirement to have a consistent test harness for everything? (because it doesn't make sense to test e.g. <canvas> and parsing in the same system)
- # [17:07] <anne> Darth Test
- # [17:07] <takkaria> jgraham: then I can have the emperor's new clothes, mr last week will like that
- # [17:08] <anne> after all, good tests are usually evil
- # [17:08] <takkaria> Philip: sure, go add it :)
- # [17:08] <Philip> (and nobody is going to just 'run the HTML5 test suite', because it'll (optimistically) be too huge for that)
- # [17:08] <Philip> takkaria: Hmph, now I've got to remember my wiki login details :-(
- # [17:09] <Philip> Oh, how handy, my browser remembered it for me
- # [17:09] <Philip> although it did give me a choice between logging in as either "Philip Taylor" or as "Philip Taylor", and it's hard to make an informed choice under those circumstances
- # [17:11] * Quits: darobin (robinb@82.233.247.234) (Client exited)
- # [17:16] * Joins: rubys (rubys@12.130.116.93)
- # [17:19] * Joins: gsnedders (gsnedders@86.136.52.180)
- # [17:20] * Joins: darobin (robinb@82.233.247.234)
- # [17:20] * takkaria adds some more test URIs to the fold
- # [17:25] <Philip> Mozilla has tests spread over lots of directories, in places like http://mxr.mozilla.org/mozilla-central/source/content/html/content/test/
- # [17:25] <pimpbot> Title: mozilla-central mozilla/content/html/content/test/ (at mxr.mozilla.org)
- # [17:26] <Philip> (generally named "test" or "reftests" or "crashtests" in a near ancestor directory of the code they're testing, it seems)
- # [17:28] <Philip> http://www.whatwg.org/specs/web-apps/current-work/status.cgi?action=get-all-annotations has links to more tests
- # [17:31] <gsnedders> Not all up-to-date though
- # [17:36] <takkaria> I'm more interested in looking at them for research purposes
- # [17:59] * Quits: myakura (myakura@221.191.119.20) (Quit: Leaving...)
- # [18:27] * Joins: hyatt (hyatt@98.201.21.231)
- # [18:28] * Quits: hyatt (hyatt@98.201.21.231) (Client exited)
- # [18:28] * Joins: hyatt (hyatt@98.201.21.231)
- # [18:38] * Quits: zcorpan (zcorpan@88.131.66.80) (Quit: zcorpan)
- # [18:41] * Joins: ChrisWilson (cwilso@131.107.0.73)
- # [18:45] <jgraham> rubys: When you point to the tracker as a means of "collecting" feedback what exact process do you have in mind? (Historically there has been some confusion regarding the tracker)
- # [18:47] <rubys> if there are issues needed to be worked, they can be tracked there. I agree that tracking is something that this group wasn't particular good at previously, but I plan to address that.
- # [18:48] * Joins: hober (ted@206.212.254.2)
- # [18:50] <DanC> my picture is that the maling list is the primary forum, and tracker is a helpful index on the side...
- # [18:51] <DanC> sometimes talking by phone is a good supplement, and tracker's agenda page is where we keep notes on which things to talk about soonish
- # [18:51] <jgraham> rubys: So, more specifically, when someone provides feedback on a spec what process do you see that feedback going through? What is the distinction between issues that end up on the tracker and those that do not?
- # [18:52] <jgraham> e.g. there as feedback that MikeSmith should add an audience section to his draft. Should that have ended up in the tracker?
- # [18:52] <jgraham> s/as/was/
- # [18:52] <rubys> editors respond to feedback on their own schedule (which is how Ian has been operating). Issues resolved can be closed. Issues that are unresolved will ultimately block progression to W3C CR status
- # [18:54] <jgraham> rubys: I still don't understand which issues should end up in the tracker in the first place
- # [18:55] <rubys> Issues that would block progress to W3C CR status. Issues such as the ones that Ian brought up on Mike's document.
- # [18:55] <rubys> And there are a number of such issues already on Ian's document.
- # [18:55] <DanC> so far, I'm content with ISSUE-59 (normative-language-reference) for all the stuff about Mike's document; I can imagine splitting other issues out
- # [18:57] <jgraham> rubys: So the job of those with tracker write access is to assess which issues would block CR and file them appropriately? Or this task will be for some subset of those people (the chairs?) or some superset e.g. anyone who thinks their issue is important enough?
- # [18:57] <DanC> the tracker roughly represents issues that people have escalated to the chairs, or that the chairs have chosen to track for whatever reason
- # [18:57] <rubys> my thinking was that publishing mike's document closed issue 59, but I can see an alternative approach of morphing that issue into a bucket for al other issues though that might make tracking progress on multiple issues more difficult
- # [18:58] <DanC> (closed? past tense?)
- # [18:59] <jgraham> DanC: Under your model of how things work (whih I think is quite close to my model) it seems like it is inappropriate to point to the tracker as the place where feedback is collected. So I am trying to undertsnad if rubys's model is different
- # [19:00] * Quits: Lachy (Lachlan@213.236.208.22) (Quit: This computer has gone to sleep)
- # [19:00] <jgraham> Since in your/my model feedback is collected on the mailing list and then some issues are specially anointed for some reason to be in the tracker
- # [19:00] <DanC> if we read "collected" as "indexed", perhaps sam's model isn't different...
- # [19:01] * DanC is learning too...
- # [19:01] <rubys> I'm quite ok with "indexed"
- # [19:01] <jgraham> DanC: That doesn't seem quite right because I assume that issues like errors in algorithms will not end up in the tracker
- # [19:02] <jgraham> (yet they would be in the index of all issues that need to be addressed)
- # [19:04] <rubys> if the person who notes the issue and the editor don't feel that the error needs to be tracked at the workgroup level, I'm quite OK with that.
- # [19:05] <rubys> Ian makes changes all the time based on feedback that he gets either publicly or privately. No need for an issue to track every change, even if those changes are important (as I presume they all are).
- # [19:05] <jgraham> But any individual can request their issue be put in the tracker and the request will be granted?
- # [19:06] <jgraham> (assuming some minimum level of merit e.g. the request is not satirical)
- # [19:08] * Quits: maddiin (mc@87.185.243.173) (Quit: maddiin)
- # [19:09] <DanC> yes; the issue tracking task force is trusted to judge "not satirical" and such
- # [19:10] <DanC> there's some discretion/judgement that the tracking is worthwhile
- # [19:10] * rubys agrees with DanC
- # [19:12] <takkaria> btw, in case you missed the discussion before, I started putting together a page on the requirements for the testsuite at http://wiki.whatwg.org/wiki/Testsuite
- # [19:12] <pimpbot> Title: Testsuite - WHATWG Wiki (at wiki.whatwg.org)
- # [19:12] * Quits: darobin (robinb@82.233.247.234) (Ping timeout)
- # [19:12] <DanC> takkaria, yeah, I'm glad to see that moving; I hope to catch up soonish
- # [19:13] <DanC> are you interested to talk about it on a teleconference, takkaria ?
- # [19:14] <takkaria> I guess, but it'll be a few weeks before I can really do anything useful
- # [19:14] <DanC> ok
- # [19:15] <takkaria> I'll seek feedback on-list re: testsuite soon, anyway
- # [19:15] * takkaria hasn't got time to follow the list again until february
- # [19:22] * Joins: Lachy (Lachlan@85.196.122.246)
- # [19:39] * jgraham tries summarising his understanding of how tracker is to be used on the mailing list
- # [19:39] <jgraham> But I have probably got some of it rong, so please correct me
- # [19:39] <jgraham> *wrong
- # [19:39] * jgraham decides to go home
- # [19:48] * Joins: Sander (svl@86.87.68.167)
- # [20:04] <DanC> ooh... test contributions. (I guess Adrian Bateman's msg is old news to the rest of you)
- # [20:09] * Joins: adele (adele@17.244.18.87)
- # [20:12] * Joins: dbaron (dbaron@63.245.220.225)
- # [20:13] <Lachy> DanC, what test contributions are you talking about?
- # [20:13] <takkaria> the IE team's
- # [20:14] <Lachy> link?
- # [20:14] <DanC> http://lists.w3.org/Archives/Public/public-html/2009Jan/0359.html
- # [20:14] <pimpbot> Title: Microsoft's cross-document messaging and DOM Storage tests published from Adrian Bateman on 2009-01-26 (public-html@w3.org from January 2009) (at lists.w3.org)
- # [20:15] <DanC> issue-12: Mike notes CVS repository in response to Bateman's offer of test cases http://lists.w3.org/Archives/Public/public-html/2009Jan/0361.html
- # [20:15] * trackbot attempting to add a note to ISSUE-12.
- # [20:15] <trackbot> ISSUE-12 File naming format for test cases notes added
- # [20:15] <pimpbot> Title: Re: Microsoft's cross-document messaging and DOM Storage tests published from Michael(tm) Smith on 2009-01-26 (public-html@w3.org from January 2009) (at lists.w3.org)
- # [20:18] * DanC wanders off for a little food before next telcon...
- # [20:32] * Parts: rubys (rubys@12.130.116.93)
- # [20:34] * Joins: rubys (rubys@12.130.116.93)
- # [20:35] * Parts: rubys (rubys@12.130.116.93)
- # [20:44] * Quits: dbaron (dbaron@63.245.220.225) (Quit: 8403864 bytes have been tenured, next gc will be global.)
- # [20:44] * Joins: dbaron (dbaron@63.245.220.241)
- # [20:47] <pimpbot> planet: HTML Anchors with 'name' or 'id'? <http://stackoverflow.com/questions/484719/html-anchors-with-name-or-id>
- # [21:06] * Quits: karl (karlcow@70.81.32.202) (Quit: This computer has gone to sleep)
- # [21:19] * Joins: Zeros (Zeros-Elip@68.50.195.181)
- # [21:32] * Quits: dbaron (dbaron@63.245.220.241) (Ping timeout)
- # [21:51] * Quits: adele (adele@17.244.18.87) (Quit: adele)
- # [22:03] * Joins: adele (adele@17.255.28.183)
- # [22:03] * Quits: adele (adele@17.255.28.183) (Quit: adele)
- # [22:03] * Joins: dbaron (dbaron@63.245.220.241)
- # [22:06] * Joins: maddiin (mc@87.185.247.5)
- # [22:06] * Quits: ROBOd (robod@89.122.216.38) (Quit: http://www.robodesign.ro )
- # [22:08] * Joins: adele (adele@17.255.28.183)
- # [22:08] * Quits: adele (adele@17.255.28.183) (Client exited)
- # [22:08] * Joins: adele (adele@17.255.28.183)
- # [22:10] * Quits: adele (adele@17.255.28.183) (Quit: adele)
- # [22:16] * Joins: alyosha (anime4chri@74.93.182.234)
- # [22:17] <alyosha> What is the <figure><legend> supposed to look like by default?
- # [22:18] <alyosha> does it look like the <fieldset><legend> or does it have the caption(legend) at the bottom like the classic photography style?
- # [22:23] <alyosha> nvm, I'm asking on the other channel
- # [22:28] * Parts: alyosha (anime4chri@74.93.182.234)
- # [22:32] * Joins: Julian_ (chatzilla@217.91.35.233)
- # [22:42] * Quits: heycam (cam@124.168.42.25) (Quit: bye)
- # [22:52] * Joins: adele (adele@17.244.18.87)
- # [22:54] * Quits: adele (adele@17.244.18.87) (Client exited)
- # [22:54] * Joins: adele (adele@17.244.18.87)
- # [22:54] * Quits: Zeros (Zeros-Elip@68.50.195.181) (Quit: This computer has gone to sleep)
- # [23:10] * Quits: Julian_ (chatzilla@217.91.35.233) (Ping timeout)
- # [23:19] * Joins: heycam (cam@130.194.72.84)
- # [23:26] * Quits: adele (adele@17.244.18.87) (Quit: adele)
- # [23:33] * Joins: adele (adele@17.244.18.87)
- # [23:33] * Quits: adele (adele@17.244.18.87) (Client exited)
- # [23:33] * Joins: adele (adele@17.244.18.87)
- # [23:38] * Joins: sryo (sryo@190.245.204.198)
- # [23:41] <Hixie> ChrisWilson: ping
- # Session Close: Wed Jan 28 00:00:00 2009
The end :)