Options:
Previous day, Next day
- # Session Start: Thu Aug 20 00:00:00 2015
- # Session Ident: #testing
- # [00:54] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [01:01] * Joins: jyasskin (~textual@public.cloak)
- # [01:01] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [02:18] * Joins: jyasskin (~textual@public.cloak)
- # [03:05] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [03:16] * terri is now known as terri_offline
- # [03:37] * Joins: gitbot (~gitbot@public.cloak)
- # [03:37] -gitbot:#testing- [web-platform-tests] hayatoito closed pull request #2097: Remove test-003.html (master...remove_event_test003) https://github.com/w3c/web-platform-tests/pull/2097
- # [03:37] * Parts: gitbot (~gitbot@public.cloak)
- # [04:09] * Joins: gitbot (~gitbot@public.cloak)
- # [04:09] -gitbot:#testing- [web-platform-tests] TakayoshiKochi opened pull request #2099: Fix event path trimming test case (master...event_dispatch_003) https://github.com/w3c/web-platform-tests/pull/2099
- # [04:09] * Parts: gitbot (~gitbot@public.cloak)
- # [04:57] * Joins: jyasskin (~textual@public.cloak)
- # [05:08] * Quits: jyasskin (~textual@public.cloak) (Client closed connection)
- # [05:09] * Joins: jyasskin (~textual@public.cloak)
- # [05:52] * Joins: MK (kawamori@public.cloak)
- # [06:25] * Joins: gamekathu (~gamekathu@public.cloak)
- # [06:42] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [06:42] * Quits: hgl (~hgl@public.cloak) (Ping timeout: 180 seconds)
- # [07:09] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [07:56] * Joins: jyasskin (~textual@public.cloak)
- # [08:26] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [09:32] * Joins: dom (dom@public.cloak)
- # [10:24] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [10:41] * Disconnected
- # [11:31] * Attempting to rejoin channel #testing
- # [11:31] * Rejoined channel #testing
- # [13:21] * Quits: gamekathu (~gamekathu@public.cloak) ("Page closed")
- # [13:33] <gsnedders> is all metadata now in the test files?
- # [13:36] <gsnedders> "Tests must be HTML, XHTML or SVG files." — can we not have arbitrary XML?
- # [13:38] <Ms2ger> In the test files or in the file name
- # [13:39] <gsnedders> in the file name? where is this documented?
- # [13:40] <Ms2ger> I have no idea
- # [13:40] <Ms2ger> But we have -manual and something with https
- # [13:44] <gsnedders> And what is the relation between the hg copy of the csswg-tests and the git copy?
- # [13:44] <gsnedders> Should I be ignoring one of them? Why do they both still exist?
- # [13:44] <MikeSmith> the git repo is just a mirror, right?
- # [13:45] <gsnedders> MikeSmith: but there are loads of issues and PRs open on it, so should I be submitting new tests on GitHub and pretended the non-mirror hg doesn't exist?
- # [13:45] <gsnedders> I am so confused.
- # [13:45] <gsnedders> And I bloody well worked on this stuff for years. >_>
- # [13:45] <MikeSmith> plinss can probably clear that up
- # [13:46] <MikeSmith> I doubt he's online yet though
- # [13:46] <MikeSmith> but anyway I think he's the one to talk to about all things csswg-tests
- # [13:46] <gsnedders> http://testthewebforward.org/docs/github-101.html tells you to use git, but then if you look at the github page it just says it's a mirror
- # [13:47] <gsnedders> "Please note that although we will grant write access directly to the Mercurial repo, it is strongly advised to use GitHub for test submissions to enable reviewers to use its built-in review tools. Direct submissions to Mercurial should be limited to administrative or housekeeping tasks, very minor changes that don't require a review, or from advanced users of the system."
- # [13:47] <gsnedders> If everything is done through Github nowadays, why on earth does the Mercurial repo still exist?
- # [13:47] * MikeSmith didn't write that part of the docs nor ever touch it so can't really say
- # [13:48] <MikeSmith> anyway, all good questions
- # [13:48] <MikeSmith> it's just that I don't think anybody who's normally around here can answer them authoratively
- # [13:48] <gsnedders> I left you guys to handle all of this when I vanished because I trusted you to keep things sane, pff! :P
- # [13:48] <MikeSmith> hahah
- # [13:48] <MikeSmith> see what happens!
- # [13:49] <MikeSmith> so the lesson for you is to never disappear again
- # [13:49] <MikeSmith> because clearly things just go all to hell while you're gone
- # [13:49] <gsnedders> is there any easy way to modify an HTML page programmatically *without* changing the serialisation of it all? hmmm
- # [13:50] <Ms2ger> gsnedders, hey, wpt is nice
- # [13:50] <Ms2ger> gsnedders, setting input.value?
- # [13:52] <gsnedders> input.value?
- # [13:52] <Ms2ger> gsnedders, and I fixed the description of csswg-test
- # [13:52] <gsnedders> Basically I want to add reftest link elements to hundreds of tests.
- # [13:52] <Ms2ger> Oh
- # [13:52] <Ms2ger> No
- # [13:52] <gsnedders> And I don't want to do that by hand.
- # [13:52] <Ms2ger> I mean, not if you use a parser
- # [13:53] <gsnedders> Yeah, exactly :\
- # [13:54] <gsnedders> this is part of the reason why I so much prefer external metadata
- # [13:58] <gsnedders> Ms2ger: that's not true, the new description, given the README says the hg copy is the canonical one
- # [13:58] <Ms2ger> For all intensive porpoises, it's true
- # [14:03] <gsnedders> tl;dr: I have all the Opera references, and the mapping from test to ref
- # [14:04] <gsnedders> I guess I'll add a moz-style manifest file to presto-testo as a first step
- # [14:04] <Ms2ger> Nice
- # [14:04] <Ms2ger> Do you have them by spec?
- # [14:08] <gsnedders> I'll just order the huge manifest file somehow, so we'll get a directory at a time
- # [14:11] <gsnedders> I'm wondering what to do with all the Opera XPath tests, because it seems like they should be in wpt
- # [14:11] <gsnedders> because whether we like it or not XPath is part of the platform
- # [14:12] <Ms2ger> That's document.evaluate()?
- # [14:12] <gsnedders> yeah
- # [14:13] <Ms2ger> We already have domxpath/ in wpt, feel free to add there
- # [14:14] <gsnedders> why did searching for "xpath" in the page not find that?
- # [14:14] <gsnedders> browsers!
- # [14:14] <gsnedders> and I don't know why we have two tests from Opera there rather than the thousand there are…
- # [14:15] <gsnedders> weird
- # [14:16] <Ms2ger> commit 3243c6ba43022564efe314c8f970062a3b5bb0b7
- # [14:16] <Ms2ger> Author: James Graham <jgraham@opera.com>
- # [14:16] <Ms2ger> Date: Mon Oct 15 10:34:08 2012 +0200
- # [14:16] <Ms2ger> Tests for interaction between XPath and the HTML parser
- # [14:16] <Ms2ger>
- # [14:16] <Ms2ger> --HG--
- # [14:16] <Ms2ger> extra : rebase_source : 06e4394fbd06e5b31a716d4080e8a3badc12e6ae
- # [14:17] <gsnedders> Doesn't really help know from where in Opera's test tree they're from, sadly
- # [14:17] <gsnedders> probably some bug directory that jgraham wrote as part of the HTML parser work
- # [14:17] <gsnedders> and the bug directories aren't public, because that'd be a nightmare
- # [14:17] <gsnedders> even though half the interesting stuff is there
- # [14:18] <Ms2ger> Yeah, sorting tests never seemed to be Opera's strong suit
- # [14:18] <gsnedders> they were sorted!
- # [14:18] <gsnedders> just not in a massively sane way
- # [14:18] <gsnedders> though most sorting schemes have some drawback or other
- # [14:20] <Ms2ger> How about jstest-futhark and jstest-core-2, which have mostly the same tests?
- # [14:24] <gsnedders> legacy of multiple JS engines co-existing at once, IIRC
- # [14:25] <gsnedders> and trying to have testsuites that should entirely pass
- # [14:25] <gsnedders> (which I'm never convinced is a worthwhile goal)
- # [14:25] <gsnedders> (you just need to know which tests are expected to fail)
- # [14:26] <Ms2ger> Right
- # [14:27] <Ms2ger> Multiple engines shouldn't be an issue if they can have separate expectations
- # [14:29] <gsnedders> there was really just a bunch of policies, some good and some bad, which led to the clusterfuck
- # [14:29] <Ms2ger> Anyway, water under the bridge
- # [14:29] * Ms2ger wonders if it'd be hard to deduplicate those automatically
- # [14:30] <MikeSmith> hmm, wondering why basically all of the html5lib tests in the wpt repo fail in Chrome
- # [14:30] <MikeSmith> e.g., http://w3c-test.org/html/syntax/parsing/html5lib_tests19.html
- # [14:31] <MikeSmith> ah
- # [14:31] <MikeSmith> nm
- # [14:31] <Ms2ger> Failed to read the 'contentDocument' property from 'HTMLIFrameElement': Blocked a frame with origin "http://w3c-test.org" from accessing a cross-origin frame.
- # [14:31] <MikeSmith> "Failed to read the 'contentDocument' property from 'HTMLIFrameElement': Blocked a frame with origin "http://w3c-test.org" from accessing a cross-origin frame."
- # [14:31] <MikeSmith> yeah
- # [14:31] <Ms2ger> Data urls?
- # [14:31] <MikeSmith> must be
- # [14:31] <MikeSmith> gsnedders would know
- # [14:32] <MikeSmith> we had discussion about this recently somewhere
- # [14:32] <MikeSmith> wonder if we could switch to using srcdoc instead
- # [14:32] <MikeSmith> for those
- # [14:34] <Ms2ger> Servo doesn't have srcdoc :)
- # [14:34] <Ms2ger> Not sure we actually run those tests
- # [14:35] <MikeSmith> oh
- # [14:35] <gsnedders> MikeSmith: it's cross-origin data URIs
- # [14:35] <MikeSmith> yeah
- # [14:35] <gsnedders> MikeSmith: but I don't really know about html/syntax/parsing much, mostly I just maintain the upstream stuff :P
- # [14:35] <MikeSmith> hai
- # [14:42] <MikeSmith> gsnedders: I'm guessing https://gsnedders.html5.org/html5lib-tests/runner.html hasn't been updated in quite a while
- # [14:42] <MikeSmith> and I suppose there there's no other way to run them online
- # [14:43] <MikeSmith> so that I can give somebody else a link to them so they can run them themselves in whatever browser/version they want
- # [14:50] * Joins: Ms2ger` (~Ms2ger@public.cloak)
- # [14:50] * Quits: Ms2ger (~Ms2ger@public.cloak) ("Leaving")
- # [14:53] <gsnedders> MikeSmith: hah, no
- # [14:53] <gsnedders> MikeSmith: that's /anicent/
- # [14:55] <Ms2ger`> <script src="http://sloth.whyi.org/~whiskey/cgi-bin/delay.py?delay=600"></script>
- # [15:09] <gsnedders> hah
- # [15:21] <gsnedders> Ms2ger`: where is that? that shouldn't have made it out, really
- # [15:22] <Ms2ger`> $ git grep whiskey
- # [15:22] <Ms2ger`> jstest-core-2/demos/inspector/index.html: <a href="http://i100.ryd.student.liu.se/~whiskey/js/inspector/inspector.py">
- # [15:22] <Ms2ger`> jstest-core-2/scheduler-regression/interruptedInlineScript3.html: <script src="http://sloth.whyi.org/~whiskey/cgi-bin/delay.py?delay=600"></script>
- # [15:22] <Ms2ger`> jstest-core-2/scheduler-regression/interruptedInlineScript4.html: document.write ("<" + "script src=\"http://sloth.whyi.org/~whiskey/cgi-bin/delay.py?delay=600\"></" + "script>");
- # [15:22] <Ms2ger`> jstest-futhark/demos/inspector/index.html: <a href="http://i100.ryd.student.liu.se/~whiskey/js/inspector/inspector.py">
- # [15:22] <gsnedders> hahaha
- # [15:24] <gsnedders> but yeah, those testsuites were mostly abandoned long ago… slightly surprising that they didn't get hit by my work on removing external servers, but whatever
- # [16:09] <Ms2ger`> <script type="sun/javascript">
- # [16:33] * Joins: jyasskin (~textual@public.cloak)
- # [16:46] * Quits: dom (dom@public.cloak) ("")
- # [16:49] * Joins: dom (dom@public.cloak)
- # [16:53] * Joins: playjax (~chatzilla@public.cloak)
- # [16:58] <Ms2ger`> gsnedders, fyi: https://github.com/w3c/csswg-test/pull/820
- # [17:00] * terri_offline is now known as terri
- # [17:02] <gsnedders> Ms2ger`: taht the ancient email?
- # [17:03] <Ms2ger`> Based on something I pushed to bitbucket in 2011
- # [17:03] <Ms2ger`> So "ancient" pretty much covers it
- # [17:03] <gsnedders> I think that's an email from me from years ago :)
- # [17:04] <Ms2ger`> Quite possible
- # [17:05] <Ms2ger`> Apparently I didn't bother mentioning where I got them :)
- # [17:10] <ato> Ms2ger`: I just
- # [17:10] <gsnedders> Ms2ger`: looks like it's those tests :)
- # [17:10] <ato> Ms2ger`: I just `git grep whiskey`’d my repo, but did not find any scotch )-:
- # [17:10] <gsnedders> Ms2ger`: see my comments
- # [17:10] <Ms2ger`> ato, in a Mozilla tree, look for bourbon instead :)
- # [17:11] <ato> Hah
- # [17:12] <AutomatedTester> yea... crazy north american's confused with the idea that Bourbon == Whiskey
- # [17:12] <AutomatedTester> but then again they call whiskey Scotch
- # [17:12] <AutomatedTester> ¯\_(ツ)_/¯
- # [17:23] * Joins: ddddd (~ddddd@public.cloak)
- # [17:23] * Quits: ddddd (~ddddd@public.cloak) ("Page closed")
- # [17:42] <Ms2ger`> I forgot how aggressively Microsoft fought the idea of having a testsuite for anything more than implementation reportds
- # [17:46] * Quits: dom (dom@public.cloak) ("")
- # [17:46] * Quits: playjax (~chatzilla@public.cloak) (Client closed connection)
- # [17:54] * Joins: bhill2 (~bhill2@public.cloak)
- # [17:56] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [17:58] <AutomatedTester> Ms2ger`: oh how they change?
- # [17:58] * Joins: bhill2 (~bhill2@public.cloak)
- # [18:01] <Ms2ger`> AutomatedTester, one hopes so
- # [18:51] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [19:07] <plinss> gsnedders: in the csswg-test repo you can have arbitrary XML, that’s fine
- # [19:07] <plinss> the hg repo is still considered our canonical repo as we have a lot of tooling built around it
- # [19:08] <plinss> the github mirror is a live bi-directional mirror, so submissions are welcome in either
- # [19:10] * Joins: bhill2 (~bhill2@public.cloak)
- # [19:23] <gsnedders> Ms2ger`: it was a weird time in the WG, the chairs pushing very very very hard to get CSS 2.1 to REC
- # [19:23] <gsnedders> Ms2ger`: most of us in the WG didn't care
- # [19:24] <gsnedders> Ms2ger`: most of us wanted the testsuite to be better even if it delayed REC
- # [19:24] <gsnedders> Ms2ger`: MS were some of the few who didn't, they wanted it to REC
- # [19:29] <gsnedders> plinss: wouldn't it kinda make sense to have hg as a read-only mirror, given the use of GitHub for review, keeping it just for the tooling?
- # [19:29] <plinss> gsnedders: we still have a number of active contributors that really prefer hg
- # [19:30] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [19:30] <plinss> the bi-directional sync seems to be working well, no real reason to remove it
- # [19:30] <gsnedders> plinss: but that takes them out of the normal review process
- # [19:31] <plinss> those pushing to hg tend to be the reviewers
- # [19:31] * Joins: bhill2 (~bhill2@public.cloak)
- # [19:32] <gsnedders> so, what, reviewer tests don't get reviewed?
- # [19:33] <plinss> no, all tests get reviewed at some point, they just don’t all have to be reviewed on github
- # [19:33] <plinss> we also have shepherd which predated github
- # [19:34] <gsnedders> bah, it seems silly to have multiple parallel processes.
- # [19:34] <gsnedders> but maybe that's just me :)
- # [19:34] <plinss> agreeed, but that’s the world we live in
- # [19:35] <gsnedders> where is most of the discussion about the CSS testsuite nowadays? #css-test seems as dead as it almost always has been, so I presume it' sjust mailing list as always?
- # [19:35] <plinss> yes
- # [19:36] <gsnedders> I'm probably going to do the rest of the massive dump of refs for tests that Opera had
- # [19:37] <gsnedders> and hope they actually get into the testsuite this time
- # [19:37] <plinss> that would be awesome
- # [19:57] * Quits: bhill2 (~bhill2@public.cloak) (Client closed connection)
- # [20:01] * Joins: bhill2 (~bhill2@public.cloak)
- # [20:12] <Ms2ger`> plinss, r? https://github.com/w3c/csswg-test/pull/820
- # [20:12] <Ms2ger`> gsnedders, fwiw, Servo has the automated tests from the css21 suite in CI
- # [20:23] * Quits: Ms2ger` (~Ms2ger@public.cloak) ("nn")
- # [20:28] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [21:00] <gsnedders> I' starting to fear we don't have the data any more. Argh! Oh well, we can create it all again…
- # [21:01] <gsnedders> tl;dr: the metadata was stored in two places, one of which no longer exists. they may not have been entirely in sync. and the Opera summer party (and 20th anniversery party) is today so I don't have anyone with access to prod.
- # [21:06] <gsnedders> presto-testo/css/selectors/reftests contains crazy tests
- # [21:06] <gsnedders> I want to know who did that!
- # [21:07] <gsnedders> I presume they';re programmatically generated, bu t still
- # [21:11] <gsnedders> https://github.com/operasoftware/presto-testo/pull/49
- # [21:12] <gsnedders> I really want an easy way to add link elements to HTML files, because I feel like I'm gonna be doing that a lot
- # [21:13] <gsnedders> 7655 tests to add refs to
- # [21:13] * Joins: jyasskin (~textual@public.cloak)
- # [21:13] <gsnedders> suggestions anyone?
- # [21:57] <jgraham> gsnedders: Use a keyboard?
- # [21:57] <jgraham> Also, whyi!
- # [22:14] * Quits: jyasskin (~textual@public.cloak) ("My computer has gone to sleep. ZZZzzz…")
- # [22:45] * Quits: bhill2 (~bhill2@public.cloak) ("Leaving...")
- # [22:56] * Joins: bhill2 (~bhill2@public.cloak)
- # [22:56] * Joins: bhill2_ (~bhill2@public.cloak)
- # [23:03] * Quits: bhill2 (~bhill2@public.cloak) (Ping timeout: 180 seconds)
- # [23:27] <gsnedders> jgraham: that will take a long time!
- # [23:28] * Quits: bhill2_ (~bhill2@public.cloak) (Client closed connection)
- # [23:36] * terri is now known as terri_offline
- # Session Close: Fri Aug 21 00:00:00 2015
Previous day, Next day
Think these logs are useful? Then please donate to show your gratitude (and keep them up, of course). Thanks! — Krijn