Options:
- # Session Start: Fri Jun 27 00:00:00 2014
- # Session Ident: #testing
- # [00:09] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [00:19] * Joins: MK (kawamori@public.cloak)
- # [00:37] * Joins: scott_gonzalez (~scott_gonzalez@public.cloak)
- # [01:11] * heycam|away is now known as heycam
- # [01:36] * Joins: Lachy (~Lachy@public.cloak)
- # [01:41] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [01:43] * Quits: lmclister (~lmclister@public.cloak) ("")
- # [02:06] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
- # [02:08] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [04:45] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [04:45] * Joins: glenn (~gadams@public.cloak)
- # [04:52] * Quits: glenn (~gadams@public.cloak) (Ping timeout: 180 seconds)
- # [05:18] * heycam is now known as heycam|away
- # [06:05] * heycam|away is now known as heycam
- # [06:43] * Joins: glenn (~gadams@public.cloak)
- # [09:00] * Joins: dom (dom@public.cloak)
- # [09:03] * Joins: gitbot (~gitbot@public.cloak)
- # [09:03] -gitbot:#testing- [web-platform-tests] dontcallmedom pushed 5 new commits to master: https://github.com/w3c/web-platform-tests/compare/7e9339dd8eaf...95a037c31763
- # [09:03] -gitbot:#testing- web-platform-tests/master 5bb0464 Zhiqiang Zhang: Import Geolocation API Test Suite...
- # [09:03] -gitbot:#testing- web-platform-tests/master 1c923b7 Zhiqiang Zhang: Remove tests covered by geo IDL tests in #348
- # [09:03] -gitbot:#testing- web-platform-tests/master 51f0807 Zhiqiang Zhang: Rewrite other geo tests using testharness.js as...
- # [09:03] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [09:27] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [09:44] * Quits: ptressel (~chatzilla@public.cloak) ("zzz")
- # [09:48] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [10:02] * Joins: zcorpan (~zcorpan@public.cloak)
- # [10:04] * Quits: zcorpan (~zcorpan@public.cloak) ("Leaving...")
- # [10:29] * Joins: Lachy (~Lachy@public.cloak)
- # [10:39] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [11:20] <jgraham> I have a bikeshed. Anyone got a paintbrush?
- # [11:20] <wilhelm> Is lead paint okay?
- # [11:21] <jgraham> Depends if you want to kill the idea :p
- # [11:22] <jgraham> So forCSP testing, it seems like there is a desire for substitutions in .headers files
- # [11:22] <jgraham> So far so fine
- # [11:22] <jgraham> But there is a requirement to do something like put the *same* random data in multiple places
- # [11:23] <jgraham> So random data is easy to add. Something like {{uuid()}}
- # [11:24] <jgraham> But something is needed for variable binding. So far I have {{$foo:uuid()}} and {{$foo}}, where the $ is required
- # [11:24] <jgraham> Those will both be replaced with the same uuid
- # [11:25] <jgraham> But I am open to other suggestions for syntax
- # [11:25] <jgraham> (the above is implemented and working, but it's not hard to change)
- # [11:25] * Joins: gitbot (~gitbot@public.cloak)
- # [11:25] -gitbot:#testing- [web-platform-tests] sizuhiko opened pull request #1079: remove createSR wrapper of createShadowRoot (master...submission/sizuhiko/remove_wrapper_of_create_shadow_root_separated) https://github.com/w3c/web-platform-tests/pull/1079
- # [11:25] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [11:26] <Ms2ger> Seems somewhat ugly, but I guess it's fine
- # [11:28] <Ms2ger> Btw, I've been working on cleaning up Philip`'s tests: https://github.com/Ms2ger/web-platform-tests/compare/philip
- # [11:28] <Ms2ger> Will probably PR tonight
- # [11:31] <jgraham> Yeah, I wasn't sure about requiring the $, but it at least means it's clear that {{host:uuid()}} {{host}} doesn't work
- # [11:33] <jgraham> (I guess the next bit of syntax will be {{$foo:uuid()|js}} which will js escape rather than html escape)
- # [11:53] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [12:02] <ato> Great, so MS is pushing for NodeJS in wpt.
- # [12:02] * ato goes to rewrite all his code
- # [12:03] <gsnedders> MS are V8 fanboys now? :)
- # [12:10] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [12:13] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [12:24] <MikeSmith> ato: where are they asking? w3c list?
- # [12:25] <ato> public-browser-tools-testing@ specifically.
- # [12:25] <gsnedders> wait, we have *another* ML now?
- # [12:25] * Joins: abarsto (~abarsto@public.cloak)
- # [12:25] * abarsto is now known as ArtB
- # [12:27] <wilhelm> gsnedders: That's for the WebDriver work. And has been for several years.
- # [12:27] <gsnedders> oh
- # [12:27] <gsnedders> did I used to be on this, before I got unsubscribed from everything by changing to no affiliation?
- # [12:30] <MikeSmith> gsnedders: maybe
- # [12:31] <MikeSmith> gsnedders: I can re-subscribe you if you want
- # [12:31] <gsnedders> don't bother
- # [12:31] <MikeSmith> k
- # [12:31] <gsnedders> I haven't even bothered trying to get myself back on public-html
- # [12:31] <wilhelm> Why would you want to subscribe to that? :D
- # [12:33] * Quits: scott_gonzalez (~scott_gonzalez@public.cloak) (scott_gonzalez)
- # [12:33] <MikeSmith> for the GT
- # [12:33] <gsnedders> to tell them they're doing it wrong when appropriate
- # [12:34] <gsnedders> but mostly to ignore it like I do all mailing lists
- # [12:38] <MikeSmith> funny thing you probably didn't know, when you go to confession at church, priests can assign you to read mailing lists as penance
- # [12:38] <MikeSmith> instead of repeating hail marys or our fathers
- # [12:38] <jgraham> MikeSmith: Wow, you must have done something *terrible*
- # [12:39] <MikeSmith> jgraham: no for that they assign you to read IETF mailing lists
- # [12:39] <wilhelm> I just got the prayer thing. I lied about my sin, though. Maybe that would have qualified for the mailing list penance.
- # [12:40] <gsnedders> MikeSmith: Thankfully, i wasn't brought up Catholic
- # [12:42] <MikeSmith> I heard that so far only some Google PMs have done anything bad enough to merit the you-must-read-an-IETF-mailing-list penance assignment
- # [12:42] <MikeSmith> but that's just a rumore
- # [12:44] <MikeSmith> wilhelm: speaking of penance we are now officially re-bound to the Browser Tools and Testing WG yoke through December 2015
- # [12:45] <MikeSmith> https://lists.w3.org/Archives/Member/w3c-ac-members/2014AprJun/0058.html
- # [12:45] <MikeSmith> "The Browser Testing and Tools Working Group charter [1] is hereby extended through December 2015."
- # [12:45] <wilhelm> MikeSmith: I just forwarded that one to our list.
- # [12:46] <wilhelm> A little bondage is good for you.
- # [12:46] <MikeSmith> wilhelm: thanks
- # [12:46] <MikeSmith> wilhelm: btw about the "The first version is expected to be completed by March 2015" that is non-binding of course
- # [12:46] <wilhelm> Yes, but worth a shot. (c;
- # [12:46] <MikeSmith> yeah
- # [12:54] * Joins: Lachy (~Lachy@public.cloak)
- # [12:58] <jgraham> Ms2ger: https://critic.hoppipolla.co.uk/r/1915 if you have time to review
- # [13:07] * Joins: scott_gonzalez (~scott_gonzalez@public.cloak)
- # [13:07] <Ms2ger> jgraham, perhaps later
- # [13:19] * Joins: MK (kawamori@public.cloak)
- # [13:30] <jgraham> Ms2ger: np
- # [13:31] <Ms2ger> Also, planning to do some parsing+serialization work this weekend
- # [13:32] <jgraham> Nice
- # [13:38] * Joins: zcorpan (~zcorpan@public.cloak)
- # [13:43] * Quits: zcorpan (~zcorpan@public.cloak) ("Leaving...")
- # [14:10] * Quits: Ms2ger (~Ms2ger@public.cloak) ("bbl")
- # [14:33] * Quits: MarkS (msadecki@public.cloak) (Ping timeout: 180 seconds)
- # [14:40] * Joins: anssik (~uid10742@public.cloak)
- # [14:47] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
- # [15:07] * Joins: plh (plehegar@public.cloak)
- # [15:45] * Quits: plh (plehegar@public.cloak) ("Leaving")
- # [16:11] * Joins: gitbot (~gitbot@public.cloak)
- # [16:11] -gitbot:#testing- [web-platform-tests] dontcallmedom closed pull request #348: Import IDL tests for Geolocation API (master...submission/zqzhang/geo-idl-test-suite) https://github.com/w3c/web-platform-tests/pull/348
- # [16:11] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [16:23] * heycam is now known as heycam|away
- # [16:45] * Joins: glenn (~gadams@public.cloak)
- # [16:48] * Quits: botie (botie@public.cloak) (Client closed connection)
- # [16:49] * Joins: plh (plehegar@public.cloak)
- # [17:00] * Quits: plh (plehegar@public.cloak) ("Leaving")
- # [17:01] * Joins: plh (plehegar@public.cloak)
- # [17:17] * Joins: gitbot (~gitbot@public.cloak)
- # [17:17] -gitbot:#testing- [web-platform-tests] andreastt opened pull request #1080: Make tools/scripts/manifest.py executable (master...ato/manifest_executable) https://github.com/w3c/web-platform-tests/pull/1080
- # [17:17] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [17:18] <ato> Are there any docs on writing manifests anywhere?
- # [17:21] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [17:23] <jgraham> Writing what kind of manifest?
- # [17:27] <ato> I think I may have misunderstood. The manifest is deduced from filename?
- # [17:27] * Ms2ger curses
- # [17:27] <ato> So I need to patch manifest.py to treat .py files specially if I want to add webdriver tests to the generated manifest?
- # [17:28] * jgraham blesses
- # [17:29] <jgraham> ato: Yeah, you need something in manifest.py to detect .py files in the webdriver repo as WebDriverTests
- # [17:29] <jgraham> (you will also need to add WebDriver to the list of test types)
- # [17:29] <Ms2ger> jgraham, so there's a bunch of Philip's tests for which the generated code seems checked in, but the source is gone
- # [17:29] <Ms2ger> Hmm
- # [17:30] * Ms2ger checks something
- # [17:30] <ato> jgraham: Thanks
- # [17:30] <jgraham> http://canvex.lazyilluminati.com/misc/tests/ ?
- # [17:33] * Quits: dom (dom@public.cloak) ("")
- # [17:35] * Ms2ger tries and fails to undelete a file
- # [17:35] <ato> So I'm wondering if “webdriver tests” is the best name. My specific use case right now is that I want to include the webdriver spec tests in wptrunner, but presumably it will in the future be used for converting manual tests into automated, interactive tests.
- # [17:37] <Ms2ger> Okay, at least one is a dupe
- # [17:37] <Ms2ger> ato, webdriven tests?
- # [17:38] <ato> I just don't want to add something by the wrong name and sticking with it “because it's hard to change”. (-:
- # [17:39] <ato> Also I notice there are no manifest.py tests. )-:
- # [17:43] <jgraham> ato: WebdriverProtocolTest?
- # [17:49] <jgraham> ato: And yeah, manifest.py isn't my finest moment
- # [17:57] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
- # [17:58] * Quits: plh (plehegar@public.cloak) ("Leaving")
- # [17:58] * Joins: plh (plehegar@public.cloak)
- # [18:03] <Ms2ger> Hm
- # [18:03] <Ms2ger> What happened to the img.complete tests?
- # [18:04] <jgraham> 303 zcorpan
- # [18:06] <Ms2ger> !summon zcorpan
- # [18:08] <jgraham> 404 not found
- # [18:10] <Ms2ger> Dammit you
- # [18:10] <wilhelm> 412
- # [18:12] * Ms2ger fails wilhelm
- # [18:13] * jgraham thought consultants always responded with 402
- # [18:17] <Ms2ger> Bah
- # [18:17] * Ms2ger pokes plh
- # [18:17] * Joins: lmclister (~lmclister@public.cloak)
- # [18:20] <jgraham> 100
- # [18:24] * Ms2ger grumbles
- # [18:24] <Ms2ger> Surely plh should be up by now
- # [18:33] <plh> back
- # [18:34] <plh> I was having a bite :)
- # [18:35] <plh> I don't understand the problem yet
- # [18:37] <ato> Hah.
- # [18:38] <ato> So I patched manifest.py but the Python test name rule I put in place didn't match a single file in the repo.
- # [18:43] * Joins: Lachy (~Lachy@public.cloak)
- # [18:45] <Ms2ger> plh, my problem is that cloning https://dvcs.w3.org/hg/html/ asks for a password
- # [18:45] <plh> hu
- # [18:45] * plh looks
- # [18:46] * Quits: anssik (~uid10742@public.cloak) ("Connection closed for inactivity")
- # [18:49] <plh> ms2ger, I don't know why there is a password yet but but isn't that the repo we used to have for html related tests? We don't use it anymore
- # [18:49] <Ms2ger> plh, yeah, but I want the history :)
- # [18:50] <Ms2ger> Because zcorpan isn't around to tell me where Philip's img.complete tests went
- # [18:50] <plh> ok
- # [18:52] <plh> I'll need to one of our system guys. I get a password as well
- # [18:52] <plh> s//ask/
- # [18:52] <plh> give me 30 minutes
- # [18:53] <Ms2ger> Thanks!
- # [19:03] <plh> ah, found a way to access that repo in the meantime. I made a zip of it: http://jay.w3.org/~plehegar/htmlhg.zip
- # [19:03] <plh> I'll work on why that stuff is being a password
- # [19:05] <plh> s/being/behind/
- # [19:08] <ato> jgraham: Should the manifest be an exhaustive list of child tests also if possible?
- # [19:08] <Ms2ger> plh, thanks :)
- # [19:08] <jgraham> ato: No, no child tests
- # [19:08] <Ms2ger> (I wonder, should we add a lint for references to obsolete drafts?)
- # [19:09] <jgraham> Or just "You appear to be adding metadata. Statistically this metadat will be out of date. Please stop"
- # [19:10] <Ms2ger> Do spec links often get out of date?
- # [19:10] <Ms2ger> I think most of the ones in dom/ are accurate, at least
- # [19:10] <jgraham> No idea. Ask me again in 5 years
- # [19:10] <Ms2ger> Perhaps not terribly useful, but not wrong
- # [19:11] <jgraham> Certianly for some specs (HTML) I wouldn't rely on them
- # [19:11] <Ms2ger> Yeah, I guess HTML moves things around more than average
- # [19:12] <ato> What happens if a test file contains no tests?
- # [19:12] <jgraham> ato: Nothing much
- # [19:12] <Ms2ger> Is it still a test?
- # [19:12] <Ms2ger> How about if it falls in the forest?
- # [19:12] <jgraham> *fails
- # [19:13] <Ms2ger> Ha
- # [19:13] <Ms2ger> Man, the old html tests repo is hard to navigaet
- # [19:13] <Ms2ger> gate
- # [19:14] <jgraham> ato: Specifically in that case I would expect that you would get a test status of OK and no subtest statuses
- # [19:15] <ato> Is there a test results document that I can look at on the WWW?
- # [19:16] <ato> Showing results from wptrunner.
- # [19:16] <jgraham> ato: What kind of results?
- # [19:16] <jgraham> I mean it just outputs structured logs
- # [19:16] <jgraham> I can point you at one
- # [19:16] <ato> Yeah, that works.
- # [19:17] <jgraham> http://mozilla-releng-blobs.s3.amazonaws.com/blobs/cedar/sha512/196fefc04c933117d3723d45a6b23a0b1cdf7e66bdef1ff32fdedbcc7289ee4669978082858310987c7106553b439892538c764ff54d810b139cb4c088fd132b
- # [19:17] <jgraham> That file is a little large, be warned
- # [19:17] <ato> Thanks, that's perfect.
- # [19:18] * Joins: glenn_ (~gadams@public.cloak)
- # [19:20] <ato> So a parent test in wpt is constituted by _one file_, right? And it would be problematic if a test type used something other than that, say a Python module's full name for parent test.
- # [19:20] <jgraham> Well
- # [19:20] <jgraham> I suppose it's possible to use anything
- # [19:21] <jgraham> So far testharness tests use the url and reftests use (url, type, ref_url)
- # [19:22] <jgraham> So actually nothing *strictly* uses filename
- # [19:22] <jgraham> *But*
- # [19:22] <jgraham> The expected data stuff does assume a mapping between a testname and a path which might not work so well if you don't use filenames.
- # [19:23] <jgraham> Or rather urls
- # [19:23] <jgraham> So that may have to be updated
- # [19:23] <ato> My concern here is that using unittest imposes certain constraints on how its test runner is designed to run tests.
- # [19:24] * Quits: glenn (~gadams@public.cloak) (Ping timeout: 180 seconds)
- # [19:25] <jgraham> So can't you just take a path to a python file and tell unittest to run that path worth of tests?
- # [19:25] <ato> Not directly.
- # [19:26] <ato> You can with a few workarounds, so we *could* make it work.
- # [19:26] <jgraham> You can somehow because it supports that from the command line
- # [19:27] <ato> Yes, so what it does from command line is to go on a test discovery hunt through the path or file you give it.
- # [19:27] <ato> Which means traversal into sub directories, and then into files matching a pattern to look for classes extending unittest.TestCase.
- # [19:28] <ato> Then it takes what it finds and puts it in multiple unittest.TestSuite's (which the test runner later conflates into an array) which it passes to the TextTestRunner.
- # [19:29] <jgraham> But if you give it a full path to a .py file it will only look in that file, surely?
- # [19:31] <ato> Yes, if the path is a single file it will do only the latter part about inspecting the file's classes for certain special traits.
- # [19:32] <jgraham> Right, so one possible implementation strategy would be to build a list of .py files containing tests, call each of those a "test", run them in turn, and call all the results from each file a "subtest"
- # [19:32] <jgraham> (another strategy would be to not use unittest ;)
- # [19:35] <ato> So I agree with that approach. I think files are conceptually a lot easier for humans to reason about in this case.
- # [19:36] <ato> Although I'm taking the exact opposite approach with a test harness I'm developing at Mozilla.
- # [19:37] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
- # [19:51] * Joins: ptressel (~chatzilla@public.cloak)
- # [20:06] <Ms2ger> Aha
- # [20:10] * Ms2ger traces this test back to revision 5bfdbef872e3 (Wed Feb 23 18:39:20 2011 +0000)
- # [20:11] <Ms2ger> Oh wait
- # [20:11] <Ms2ger> That commit removed 2d.drawImage.incomplete.html
- # [20:13] <Ms2ger> Oh nooooooo
- # [20:13] <Ms2ger> approved/ strikes again
- # [20:15] <Ms2ger> Oh, and the reason I can't find 2d.drawImage.svg is that some Daniel Davis imitated Philip's code, but didn't add his test to the yaml
- # [20:23] * Quits: lmclister (~lmclister@public.cloak) ("")
- # [20:24] * Joins: lmclister (~lmclister@public.cloak)
- # [20:27] * Quits: glenn_ (~gadams@public.cloak) (Client closed connection)
- # [20:27] * Joins: glenn (~gadams@public.cloak)
- # [20:28] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [20:34] * Joins: rhauck (~Adium@public.cloak)
- # [20:34] * Joins: glenn (~gadams@public.cloak)
- # [20:35] * Quits: glenn (~gadams@public.cloak) (Client closed connection)
- # [20:40] <Ms2ger> Spec references: <a href="../annotated-spec/canvas.html#testrefs.2d.drawImage.svg">2d.drawImage.svg</a>
- # [20:40] <Ms2ger> Dead link.
- # [20:44] * Ms2ger kicks people wrapping tests in (function() { ... })();
- # [21:19] <plh> ms2ger, I don't think we ever had the annotated spec in that repo
- # [21:26] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [22:04] * Quits: ptressel (~chatzilla@public.cloak) ("bbl")
- # [22:22] * Quits: rhauck (~Adium@public.cloak) ("Leaving.")
- # [22:25] <AutomatedTester> Yves: did you ever get the REST api endpoint?
- # [22:26] <Ms2ger> plh, the canvas one? Sure did :)
- # [22:27] <Yves> not yet no :(
- # [23:17] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [23:26] * Quits: plh (plehegar@public.cloak) ("Leaving")
- # [23:48] * Joins: rhauck (~Adium@public.cloak)
- # [23:53] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
- # Session Close: Sat Jun 28 00:00:00 2014
The end :)