Options:
Previous day, Next day
- # Session Start: Tue Jan 20 00:00:00 2015
- # Session Ident: #testing
- # [00:06] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
- # [00:39] * terri is now known as terri_offline
- # [01:37] * heycam is now known as heycam|away
- # [01:46] * Joins: terri (~terri@public.cloak)
- # [01:47] * Quits: terri (~terri@public.cloak) ("Page closed")
- # [01:56] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [02:01] * heycam|away is now known as heycam
- # [03:30] * Joins: tripu (antonio@public.cloak)
- # [03:34] * heycam is now known as heycam|away
- # [03:57] * heycam|away is now known as heycam
- # [04:33] * Joins: ptressel (~chatzilla@public.cloak)
- # [05:15] * Quits: artb (~ArtB@public.cloak) ("Leaving.")
- # [05:21] * Joins: artb (~ArtB@public.cloak)
- # [05:23] * Quits: tripu (antonio@public.cloak) (Ping timeout: 180 seconds)
- # [05:30] * Quits: artb (~ArtB@public.cloak) ("Leaving.")
- # [05:38] * Quits: hober (~ted@public.cloak) (Client closed connection)
- # [05:38] * Quits: sangwhan (~sid12645@public.cloak) (Client closed connection)
- # [05:38] * Joins: hober (~ted@public.cloak)
- # [05:38] * Joins: ptressel_ (~chatzilla@public.cloak)
- # [05:38] * Joins: sangwhan (~sid12645@public.cloak)
- # [05:38] * Joins: astearns_ (~sid15080@public.cloak)
- # [05:38] * Quits: lmclister______ (~sid13822@public.cloak) (Ping timeout: 180 seconds)
- # [05:38] * Quits: scheib (~sid4467@public.cloak) (Ping timeout: 180 seconds)
- # [05:38] * Joins: lmclister______ (~sid13822@public.cloak)
- # [05:39] * Joins: scheib (~sid4467@public.cloak)
- # [05:41] * Quits: ptressel (~chatzilla@public.cloak) (Ping timeout: 180 seconds)
- # [05:41] * ptressel_ is now known as ptressel
- # [05:41] * Quits: astearns (~sid15080@public.cloak) (Ping timeout: 180 seconds)
- # [05:41] * astearns_ is now known as astearns
- # [05:46] * heycam is now known as heycam|away
- # [05:48] * Joins: tripu (antonio@public.cloak)
- # [07:49] * Quits: tripu (antonio@public.cloak) ("Leaving")
- # [07:50] * Joins: tripu (antonio@public.cloak)
- # [08:49] * Joins: dom (dom@public.cloak)
- # [09:17] * Joins: Lachy (~Lachy@public.cloak)
- # [09:33] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [09:35] * Quits: Lachy (~Lachy@public.cloak) ("Textual IRC Client: www.textualapp.com")
- # [09:37] * Joins: zcorpan (~zcorpan@public.cloak)
- # [10:03] * Joins: darobin (rberjon@public.cloak)
- # [10:12] * Joins: zcorpan_ (~zcorpan@public.cloak)
- # [10:19] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
- # [10:47] * Joins: zcorpan (~zcorpan@public.cloak)
- # [10:47] * Quits: zcorpan_ (~zcorpan@public.cloak) (Client closed connection)
- # [11:25] <zcorpan> Ms2ger: should https://critic.hoppipolla.co.uk/r/3770 be reopened?
- # [11:27] <Ms2ger> Yep, thanks
- # [11:30] <zcorpan> no green button in https://github.com/w3c/web-platform-tests/pull/1548
- # [11:37] <jgraham> Needs a rebase
- # [11:37] <Ms2ger> Guess what I'm doing :)
- # [11:40] <jgraham> Talking on irc?
- # [11:40] <Ms2ger> Now, yes
- # [11:41] * Joins: gitbot (~gitbot@public.cloak)
- # [11:41] -gitbot:#testing- [web-platform-tests] Ms2ger closed pull request #1548: Rewrite EventListener.js. (master...EventListener) https://github.com/w3c/web-platform-tests/pull/1548
- # [11:41] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [11:41] <jgraham> What do I win?
- # [11:41] <Ms2ger> More reviews
- # [11:43] <Ms2ger> jgraham, actually, you get to figure out http://build.servo.org/builders/linux2/builds/579/steps/shell_1/logs/stdio
- # [11:47] <jgraham> Yay!
- # [11:49] <jgraham> Which particular part of that? The failure or the fact that it's throwing errors all over the place?
- # [11:51] <Ms2ger> The throwing errors part
- # [11:51] * Joins: artb (~ArtB@public.cloak)
- # [12:08] <Ms2ger> jgraham, looks like it's the glutin change
- # [12:11] <jgraham> Yeah so I haven't looked in much detail, but it's the harness erroring out because reftest files aren't being written
- # [12:12] <jgraham> Which is a bug in the harness, but also presumably a bug in servo
- # [12:12] * Quits: tripu (antonio@public.cloak) (Ping timeout: 180 seconds)
- # [12:13] <Ms2ger> I can't even run Servo locally \o/
- # [12:14] <jgraham> You installed the dep?
- # [12:15] <Ms2ger> Yeah
- # [12:15] <Ms2ger> It didn't link before I did that
- # [12:15] <Ms2ger> Now it's X Error of failed request: BadRequest (invalid request code or no such operation)
- # [12:16] * Joins: zcorpan_ (~zcorpan@public.cloak)
- # [12:17] <jgraham> Time to put on your sheriff hat?
- # [12:17] <jgraham> ;)
- # [12:21] <jgraham> Anyone know if the reftest.list files in CSS are actually correct?
- # [12:21] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
- # [13:25] * Joins: zcorpan (~zcorpan@public.cloak)
- # [13:25] * Quits: zcorpan_ (~zcorpan@public.cloak) (Client closed connection)
- # [14:05] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [14:07] * Quits: terri_offline (~terri@public.cloak) (Ping timeout: 180 seconds)
- # [14:10] * Joins: plh (plehegar@public.cloak)
- # [14:10] * Quits: gsnedders (~gsnedders@public.cloak) (Ping timeout: 180 seconds)
- # [14:21] * Joins: tripu (antonio@public.cloak)
- # [14:24] * Joins: gsnedders (~gsnedders@public.cloak)
- # [14:40] * Joins: terri_offline (~terri@public.cloak)
- # [14:40] * terri_offline is now known as terri
- # [14:45] * Joins: zcorpan_ (~zcorpan@public.cloak)
- # [14:45] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
- # [14:55] * Quits: ptressel (~chatzilla@public.cloak) ("ZZZzzz...")
- # [15:07] * Joins: anssik (~uid10742@public.cloak)
- # [15:30] * Quits: zcorpan_ (~zcorpan@public.cloak) (Client closed connection)
- # [15:39] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [15:48] * Joins: zcorpan (~zcorpan@public.cloak)
- # [15:49] * Joins: zcorpan_ (~zcorpan@public.cloak)
- # [15:49] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
- # [15:55] * Quits: zcorpan_ (~zcorpan@public.cloak) (Client closed connection)
- # [16:15] * Joins: zcorpan (~zcorpan@public.cloak)
- # [16:17] <jgraham> "Yay!
- # [16:17] <jgraham> s/!/"/
- # [16:18] <Ms2ger> ?
- # [16:18] <jgraham> 'utf8' codec can't decode byte 0x96 in position 2644: invalid start byte
- # [16:18] <Ms2ger> \o/
- # [16:18] <jgraham> This is a "XHTML" CSS test with no other encoding information
- # [16:25] <gsnedders> that's really odd.
- # [16:25] <gsnedders> the toolchain should fail on it.
- # [16:25] <gsnedders> hence they shouldn't have made it into the testsuite
- # [16:28] <jgraham> That is what I would have thought
- # [16:35] <jgraham> XMLSyntaxError: redefinition of the xmlns prefix is forbidden, line 14, column 23
- # [16:46] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
- # [17:14] <jgraham> So, comparing the list of tests you get from the CSSWG reftest.list files to those from the rules we discussed yesterday:
- # [17:14] <jgraham> http://pastebin.mozilla.org/8261095
- # [17:16] <jgraham> So there are actually only two files where we fail to extract a test that ought to exist given those rules not due to a cycle or a bug in the ts
- # [17:16] <jgraham> (the cases labelled "6 have inbound links")
- # [17:17] <jgraham> They seem to also have a rule that a file with -ref at the end of the name cannot be a test, but that suggests those are dead code (I should verify that)
- # [17:22] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
- # [17:30] * Quits: dom (dom@public.cloak) (Client closed connection)
- # [18:10] * Joins: bhill2 (~bhill2@public.cloak)
- # [18:11] * Joins: hjkh (~hjkh@public.cloak)
- # [18:13] * hjkh slaps hjkh around a bit with a large fishbot
- # [18:13] * Parts: hjkh (~hjkh@public.cloak)
- # [18:22] * Joins: zcorpan (~zcorpan@public.cloak)
- # [19:00] * Quits: tripu (antonio@public.cloak) (Ping timeout: 180 seconds)
- # [19:17] <jgraham> So, how do people actually feel about all this CSS stuff where files with -ref or _ref at the end of the name, or in a references/ directory, are always considered to be reference files (and never tests)
- # [19:17] <Ms2ger> Well
- # [19:17] <Ms2ger> Would be nice if we linted for files like that that aren't otherwise classified as refs
- # [19:19] <jgraham> I actually found, and started reading, the code that processes the CSS tests
- # [19:19] <jgraham> The approach there seems to be "all files with rel=[mis]match are tests, unless they have one of the identifying characteristics of a support file"
- # [19:20] <jgraham> So I expect A == B == A generates two tests
- # [19:20] <Ms2ger> With exactly the same result
- # [19:23] <jgraham> Yeah
- # [19:24] <jgraham> I mean, I might be wrong, I didn't read everything yet
- # [19:24] <Ms2ger> So I guess that's kinda neat, but also silly
- # [19:29] <jgraham> Well based on the findings from yesterday, it seems like not enforcing the rule that there must not be unlinked-to cycles just leads to bugs
- # [19:30] <jgraham> But I don't know if we should make it a requirement that refs follow one of these naming conventions
- # [19:30] <Ms2ger> I don't think we should
- # [19:30] <Ms2ger> But I don't think strongly :)
- # [19:31] <jgraham> Well the advantage is that we follow the CSS rules almost exactly, so there is less to argue about
- # [19:31] <gsnedders> I think the -ref/_ref or references/* is good, IMO
- # [19:31] <jgraham> The disadvantage is that I think it's silly.
- # [19:35] <Ms2ger> I particularly dislike references/
- # [19:37] <jgraham> Seems to be rather popular in CSS
- # [19:38] <Ms2ger> That's not exactly a ringing endorsement
- # [19:38] <jgraham> No, but it does mean that there will be tears if we try to change it
- # [19:39] <Ms2ger> Well, there won't be tears if we don't require it, maybe? :)
- # [19:39] <jgraham> It isn't required
- # [19:39] <jgraham> It's an option
- # [19:39] <jgraham> I particularly dislike that there are three ways to do it
- # [19:40] <jgraham> For example of the 19551 files matching */reference/* 12972 match */reference/*ref.*
- # [19:41] <Ms2ger> Ask plinss?
- # [19:44] * terri is now known as terri_offline
- # [19:48] <jgraham> Looking at it some of the 255 files (excluding generated files and support files) that would need to be fixed (and for which all the referrers would need to be fixed) are in css21
- # [19:49] <jgraham> So I guess I will be told "no because Process"
- # [19:50] <Ms2ger> I don't think so, actually
- # [19:51] <jgraham> Really?
- # [19:51] <Ms2ger> https://lists.w3.org/Archives/Public/public-css-testsuite/2015Jan/0023.html makes me think so
- # [19:51] <Ms2ger> But you never know, of course
- # [19:55] <jgraham> So a possible way forward could be:
- # [19:56] * jgraham stops to think
- # [19:57] <jgraham> Build a graph from all rel links
- # [19:57] <jgraham> References are things that have -ref in the name
- # [19:57] <jgraham> Tests are things that are in the graph and aren't references
- # [19:58] <jgraham> Unlinked cycles are an error
- # [19:59] <jgraham> References without a test linking to them are a lint error
- # [20:00] <jgraham> Not sure if I had something else
- # [20:00] <jgraham> But CSS could be fixed to follow these rules
- # [20:01] * jgraham decides the way home is the best place to consider this
- # [20:02] <Ms2ger> I guess email first before implementing
- # [20:03] <plinss> jgraham: the relevant code to identify tests vs references is at: http://hg.csswg.org/dev/w3ctestlib/file/0b71cc5e3fb6/Sources.py#l23
- # [20:03] <plinss> the logic is: if it’s in a ‘support/‘ directory, it’s a support file
- # [20:04] <plinss> if it’s in a ‘reference/‘ directory or has ‘ref-‘, ‘notref-‘, ‘-ref-‘, ‘-notref-‘, or ‘-ref’, ‘-notref’ in the file name it’s a reference
- # [20:04] <plinss> els it’s a test
- # [20:04] <plinss> tests do not have to have rel=‘[mis]match’, but if it does, it’s a reftest
- # [20:05] <plinss> tests _do_ have to have rel=‘help’ links
- # [20:05] <plinss> (or be otherwise associated with a spec)
- # [20:05] <plinss> reference relationships do not create or imply tests
- # [20:05] <plinss> tests may reference other tests or reference files
- # [20:07] * terri_offline is now known as terri
- # [20:07] <plinss> personally I don’t care if you want to change the naming convention, but then all the tests that have reference need to be updated to match
- # [20:09] <plinss> also, having a rel=‘[mis]match’ in a file does nothing to tell you if it’s a test or a reference
- # [20:09] <Ms2ger> How strongly do you feel about these rules?
- # [20:10] <plinss> it’s not about how I _feel_ about the rules, it’s about having a system that works for our needs, and the work involved to change things
- # [20:11] <Ms2ger> Right
- # [20:11] <Ms2ger> Remind me, are your needs written down somewhere?
- # [20:12] <plinss> what we have is documented on our wiki, I don’t know if it’s all written down or not
- # [20:13] <plinss> I don’t care how references are identifed or linked to tests, but we need files that are references (and not tests), and we need multiple references per test for some (as well as mismatch references)
- # [20:13] <plinss> we also need optional references, ie: the test passes if it matches one or the other
- # [20:14] <Ms2ger> By "multiple references", you mean "matches one of tho...
- # [20:14] <Ms2ger> Apparently not
- # [20:14] <plinss> I mean both
- # [20:14] <plinss> some tests need to match more thatn one reference (which match each other)
- # [20:14] <plinss> other tests need to match one of a set
- # [20:14] <Ms2ger> Got it
- # [20:15] <plinss> (fwiw: the needs weren’t all written down based on an arbitrary list someone came up with, but developed over time as we had to figure out how to test features)
- # [20:15] <Ms2ger> I didn't mean to imply anything else :)
- # [20:37] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [20:48] * Quits: anssik (~uid10742@public.cloak) ("Connection closed for inactivity")
- # [20:57] * Joins: darobin (rberjon@public.cloak)
- # [21:18] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
- # [21:18] * Joins: zcorpan (~zcorpan@public.cloak)
- # [21:55] * heycam|away is now known as heycam
- # [22:11] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
- # [22:38] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
- # [23:00] * Joins: gitbot (~gitbot@public.cloak)
- # [23:00] -gitbot:#testing- [testharness.js] Ms2ger opened pull request #103: Support callback interfaces. (master...callback-interfaces) https://github.com/w3c/testharness.js/pull/103
- # [23:00] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [23:12] * Joins: zcorpan (~zcorpan@public.cloak)
- # [23:45] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
- # [23:46] <jgraham> plinss: In wpt rel=[mis]match identifies a reftest and <script src="testharness.js"> identifies a testharness test
- # [23:46] <jgraham> manual tests require special naming
- # [23:47] <jgraham> Adding support for the multiple references is what I'm doing
- # [23:48] <jgraham> (so if you have a rel=match but [some other feature] you won't be considered a test but a ref instead)
- # [23:48] <jgraham> I want this to be basically compatible with CSS but with fewer options
- # Session Close: Wed Jan 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