/irc-logs / w3c / #testing / 2014-06-02 / end

Options:

# Session Start: Mon Jun 02 00:00:00 2014
# Session Ident: #testing
# [00:11] * Joins: ptressel (~chatzilla@public.cloak)
# [00:22] * heycam is now known as heycam|away
# [00:25] * Quits: lmclister (~lmclister@public.cloak) ("")
# [00:30] * Joins: lmclister (~lmclister@public.cloak)
# [00:35] * heycam|away is now known as heycam
# [00:48] * Quits: lmclister (~lmclister@public.cloak) ("")
# [00:57] * Joins: hank (~hank@public.cloak)
# [01:07] * heycam is now known as heycam|away
# [01:07] * Joins: lmclister (~lmclister@public.cloak)
# [01:26] * Quits: hank (~hank@public.cloak) (Ping timeout: 180 seconds)
# [01:28] * Joins: abarsto (~abarsto@public.cloak)
# [01:28] * abarsto is now known as ArtB
# [01:31] * Joins: hank (~hank@public.cloak)
# [01:34] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
# [01:44] * heycam|away is now known as heycam
# [01:45] * Quits: hank (~hank@public.cloak) (Ping timeout: 180 seconds)
# [01:47] * Quits: ptressel (~chatzilla@public.cloak) (Client closed connection)
# [02:56] * Quits: lmclister (~lmclister@public.cloak) ("")
# [02:57] * Joins: lmclister (~lmclister@public.cloak)
# [02:57] * Quits: lmclister (~lmclister@public.cloak) (lmclister)
# [02:59] * Joins: ptressel (~chatzilla@public.cloak)
# [03:08] * Joins: lmclister (~lmclister@public.cloak)
# [03:08] * Quits: lmclister (~lmclister@public.cloak) (lmclister)
# [03:08] * Joins: lmclister (~lmclister@public.cloak)
# [03:13] * Quits: MarkS (msadecki@public.cloak) (Ping timeout: 180 seconds)
# [04:26] * Quits: lmclister (~lmclister@public.cloak) ("")
# [04:37] * heycam is now known as heycam|away
# [05:13] * Quits: ptressel (~chatzilla@public.cloak) (Client closed connection)
# [05:24] * Joins: kawai (~kawai@public.cloak)
# [05:24] * Quits: kawai (~kawai@public.cloak) (Client closed connection)
# [06:32] * Joins: ptressel (~chatzilla@public.cloak)
# [06:49] * Joins: gitbot (~gitbot@public.cloak)
# [06:49] -gitbot:#testing- [web-platform-tests] sideshowbarker pushed 13 new commits to master: https://github.com/w3c/web-platform-tests/compare/d4e784f484bb...18f10b524090
# [06:49] -gitbot:#testing- web-platform-tests/master 8d00379 Tomoyuki SHIMIZU: TestTWF Tokyo Meetup: needs review
# [06:49] -gitbot:#testing- web-platform-tests/master ac9ab91 Tomoyuki SHIMIZU: tests for Meter Element created
# [06:49] -gitbot:#testing- web-platform-tests/master 30127dc Tomoyuki SHIMIZU: Merge remote-tracking branch 'upstream/master' into submissions/tomoyukilabs
# [06:49] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
# [06:49] * Joins: gitbot (~gitbot@public.cloak)
# [06:49] -gitbot:#testing- [web-platform-tests] sideshowbarker closed pull request #330: added tests for the Meter Element and the Progress Element (master...submissions/tomoyukilabs) https://github.com/w3c/web-platform-tests/pull/330
# [06:49] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
# [06:56] * Joins: gitbot (~gitbot@public.cloak)
# [06:56] -gitbot:#testing- [web-platform-tests] sideshowbarker closed pull request #469: second bunch of custom elements tests (master...submission/unipro-custom-elements-2) https://github.com/w3c/web-platform-tests/pull/469
# [06:56] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
# [07:01] * Joins: lmclister (~lmclister@public.cloak)
# [07:24] * heycam|away is now known as heycam
# [07:58] * Quits: lmclister (~lmclister@public.cloak) ("")
# [09:10] * Joins: Ms2ger (~Ms2ger@public.cloak)
# [09:38] * Joins: zcorpan (~zcorpan@public.cloak)
# [10:30] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
# [10:30] * heycam is now known as heycam|away
# [11:02] <jgraham> Oops
# [11:02] <jgraham> It seems I just committed to testtwf-website rather than opening a PR
# [11:02] * jgraham blames the github UI
# [11:03] <jgraham> tobie_: r? on the latest commit?
# [11:03] <jgraham> (it's only a small change)
# [11:05] * Joins: Lachy (~Lachy@public.cloak)
# [11:12] <jgraham> There just aren't words for how terrible GitHub seems to be. If you have commit access to a repository it seems like there is no way of using the inline editing feature and submitting a pull request rather than committing directly. So, you think, obviously I'm supposed to fork, commit to my fork, and then create a pull request or something. So you click fork only to discover that you already had a fork, but it is not up to date with the origin. Now the[CUT]
# [11:12] <jgraham> ... fetch the latest commits from origin and start a branch based on the latest commit. So they spent all the time to make an inline editing UI but it's actually totally useless and you have to do everything via the command line / your local machine.
# [11:14] <jgraham> Why doesn't someone create a new code hosting thing with better functionality and github compatibility (via the api) so that people can migrate onto a better solution but still accept some interactions from github
# [11:15] <jgraham> Code hosting just doesn't seem like it should have strong enough network effects or stickiness that we get stuck with a site that gave up improving the core product
# [11:15] <jgraham> (I don't know if anyone is actually *better* than GitHub yet, but it's clear that you could be)
# [11:19] <Ms2ger> Oh hi, wall of text
# [11:26] <tobie_> hi
# [11:26] <jgraham> Ms2ger: You know, talking to the text itself is a bit odd
# [11:27] <tobie_> jgraham: it's not only network effect but also branding
# [11:29] <jgraham> Isn't the branding just part of the (weak) network effect?
# [11:29] <jgraham> Kind of "no one ever got sacked for choosing GitHub"
# [11:33] <jgraham> tobie_: Have a review request
# [11:33] <tobie_> the utf8 thing?
# [11:34] <tobie_> jgraham: ^
# [11:37] <jgraham> tobie_: https://github.com/w3c/testtwf-website/pull/250
# [11:37] <jgraham> A followup
# [11:39] <tobie_> Just curious what the rational for the minimal template is?
# [11:40] <tobie_> Or more precisely, why do we bother with the rest of the cruft if all we need is what's in the minimal template.
# [11:40] <tobie_> :)
# [11:40] <jgraham> tobie_: The metadata stuff is basically optional, and it much more closely resembles something I think I can sell to Mozilla developers as a good idea
# [11:41] <tobie_> Feels like author metadata can be obtained from the versioning tool anyway.
# [11:41] <zcorpan> jgraham: "Authors must use the utf-8 encoding and must use the ASCII case-insensitive "utf-8" label to identify it."
# [11:41] <jgraham> tobie_: All the metadata stuff has been driven by people's desire to either try and measure coverage via metadata or do whatever it is that the CSS WG do
# [11:41] <jgraham> zcorpan: Bah. I thought we didn't like extra bytes :p
# [11:42] <jgraham> s/we/google.com/
# [11:42] <jgraham> tobie_: I agree about the author metadata
# [11:42] <zcorpan> jgraham: yeah if it was up to me i'd allow <!doctypehtml> also :-P
# [11:42] <jgraham> That's basically the easiest case to work out from the VCS
# [11:43] <tobie_> So, my suggestion would be replace the existing template with the minimal one.
# [11:43] <tobie_> But I'm happy to approve the current PR as such.
# [11:45] <tobie_> Also, for ttwf-website, I've taken a more liberal approach to PR, if you have commit access and you feel your commit is non-controversial, feel free to merge it.
# [11:45] <jgraham> Pushed some fixups based on zcorpan's complaint and the fact that div id=log isn't really needed anymore
# [11:45] <tobie_> didn't know re div id=log (that's awesome)
# [11:46] <jgraham> Personally I'm pretty happy to kill the template stuff, but I think I would rather just merge this now
# [11:46] <jgraham> s/template/metadata/
# [11:46] <tobie_> Just go ahead.
# [11:47] <jgraham> Thaks
# [11:47] <jgraham> +n
# [11:47] <tobie_> (also see my above comment)
# [11:47] <tobie_> (for ttwf-website, I've taken a more liberal approach to PR, if you have commit access and you feel your commit is non-controversial, feel free to merge it.)
# [11:48] <jgraham> OK, I didn't know that
# [11:48] <tobie_> yeah, I feel it's a reasonable policy.
# [11:48] <tobie_> Don't you think?
# [11:49] <tobie_> Nothing's mission critical in there, and it's hard to break stuff.
# [11:49] <tobie_> (given it's all text)
# [11:50] <tobie_> jgraham: have been working on a tool to generate test stubs for specs, fill in the idl stuff, and keep in sync with changes in the spec. I'm testing it on SW right now and planning to make a PR on wpt soonish.
# [11:51] * Joins: dom (dom@public.cloak)
# [11:53] <tobie_> Would benefit from some way to mark a test as not implemented yet (e.g. through a todo() function or some such) which could get reported similarly to TIMEOUTS.
# [11:53] <tobie_> (in testharness.js, that is)
# [11:53] <tobie_> Wasn't going to bring that up yet, but mounir had a similar request on twitter yesterday evening.
# [11:54] <tobie_> https://twitter.com/mounirlamouri/status/473180951796011008
# [11:54] <tobie_> jgraham: is that something you'd consider for testharness.js?
# [11:56] * Joins: mounir (~mounir@public.cloak)
# [11:57] <tobie_> mounir: I was just mentioning your suggestion proposal minutes ago.
# [11:57] <mounir> tobie_: I hope you took the opportunity to say all the great things you think while I wasn't listening ;)
# [11:58] <jgraham> So it seems like that would mix UA-specific metadata in with the test file
# [11:58] <jgraham> wptrunner has seperate ini-ish files that are designed to hold that kind of data (although you might think it's a bit heavyweight)
# [11:58] <mounir> wptrunner?
# [11:59] <jgraham> https://github.com/w3c/wptrunner/tree/jgraham/initial
# [11:59] <jgraham> Tool for running the tests mainly in automation
# [12:00] <jgraham> Works with several Mozilla browsers and has somewhat undertested Chrome support based on WebDriver
# [12:00] <tobie_> jgraham: it might be that mounir and I aren't looking at the same thing.
# [12:00] <tobie_> jgraham: I'm interested in having a way to signal a test hasn't been written yet.
# [12:01] <tobie_> jgraham: I think you referring to a system to avoid running a test for which no implementation exists yet in the client.
# [12:01] <tobie_> I'm no longer sure what mounir refers to.
# [12:01] <mounir> tobie_: I was thinking kind of the two sides but with an implementer hat on
# [12:01] <jgraham> No, I was just referring to a system to avoid reporting a result as "FAIL" when you know it's going to FAIL
# [12:01] <mounir> so, it's pretty similar as far as implementers are concerned
# [12:02] <jgraham> My solution to the problem I had in mind looks like https://hg.mozilla.org/projects/cedar/file/99f8fa4add6d/testing/web-platform/meta/custom-elements/concepts/custom-elements-type-allowed-chars-first-char.html.ini
# [12:02] <jgraham> But let me reread the twitter thread
# [12:03] <tobie_> jgraham: that's really not what I'm interested in. :)
# [12:03] <tobie_> jgraham: but that might solve mounir's use case.
# [12:04] <mounir> tobie_: so, why wouldn't you write a test case?
# [12:04] <mounir> tobie_: I mean, if you can write the test, why not?
# [12:04] <jgraham> OK, I think I understand what you want, but I don't exactly know why, and I don't know what mounir wants
# [12:04] <tobie_> mounir: I've built a tool to generate test stubs
# [12:04] <tobie_> I'd like those to report a TODO status until someone actually writes the test for it
# [12:05] <tobie_> i.e.: test(function() { todo() });
# [12:05] <jgraham> would probably have to be this.todo()
# [12:06] <mounir> jgraham: https://github.com/w3c/wptrunner/blob/jgraham/initial/requirements.txt < I love how 'html5lib' is the only non-Mozilla project in the list
# [12:06] <tobie_> sure. whatever, it's generated code anyway. :)
# [12:06] <jgraham> mounir: Well technically selenium is also a requirement for running in Chrome, but I didn't want to make everyone install that
# [12:07] <jgraham> mounir: It turns out that Mozilla have built a number of cross-platform libs for writing test harnesses which are quite useful when writing a test harness :)
# [12:10] <mounir> jgraham: sure, might just be harder to use that in Blink
# [12:11] <jgraham> mounir: Yeah, so there are a few of those dependencies you can kill if you aren't planning to run in firefox, but a number of them are generic
# [12:13] <jgraham> (mozrunner, mozdevice, mozprofile and marionette_client aren't needed except for Firefox(OS), the others are I thik)
# [12:13] <jgraham> *think
# [12:16] <tobie_> mounir: do we want the same thing?
# [12:16] <tobie_> wrt todo?
# [12:17] <tobie_> Or is your use case met by wpt's manifest files?
# [12:17] <mounir> tobie_: I think something like what jgraham proposed is more what I'm looking for
# [12:17] <mounir> I'm not a big fan of using a manifest though
# [12:17] <mounir> It's painful
# [12:17] <mounir> but I understand why it is the prefered reasons
# [12:18] <mounir> ... given that testharness is made to send the test to the w3c
# [12:18] <tobie_> well, if it's UA-specific , then it has to be outside of the test file
# [12:18] <mounir> tobie_: it's for UA-specific tests too...
# [12:18] <mounir> tobie_: one UC I have is also to mark that a test should be done
# [12:19] <mounir> ... but that it should be done later because it's not doable now
# [12:19] <mounir> tobie_: I think we have slightly different UC here
# [12:19] <mounir> conclusion: twitter is a poor medium to discuss serious stuff :)
# [12:19] <tobie_> heh
# [12:20] <Ms2ger> Well, it could've been github...
# [12:20] <tobie_> by marking that a tesh should be done, what precisely do you mean?
# [12:20] * jgraham awards mounir +1 wisdom
# [12:20] <tobie_> do you have a bot actually recording that?
# [12:20] <tobie_> like w3c's karma bot they have on internal channels?
# [12:21] <jgraham> Haha, no I should
# [12:21] * tobie_ lunch
# [12:21] <Ms2ger> tobie_, I guess "this should be tested, but I'm not going to write the test now"
# [12:23] * Joins: darobin (rberjon@public.cloak)
# [12:24] <mounir> tobie_: a good example is form controls, the tests for the foo attribute might require special tests with type='bar' but it's not implemented
# [12:24] <mounir> tobie_: what I used to do was a todo() that will check that bar isn't implemented
# [12:24] <mounir> tobie_: so when bar was implemented, it would break the todo, make the test fail and hopefully get the developer to update the test file with bar's tests
# [12:24] <mounir> tobie_: I assume that the manifest solution should make that UC working too
# [12:28] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
# [12:29] * Joins: abarsto (~abarsto@public.cloak)
# [12:29] * abarsto is now known as ArtB
# [12:54] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
# [13:11] * Joins: zcorpan (~zcorpan@public.cloak)
# [13:49] * Joins: scott_gonzalez (~scott_gonzalez@public.cloak)
# [13:50] * Joins: darobin (rberjon@public.cloak)
# [13:54] <tobie_> jgraham: while on the topic of th.js enhancements, have you considered cjs assert support? See https://groups.google.com/d/topic/mozilla.dev.platform/vdyLJBJTdYY/discussion
# [13:56] <jgraham> tobie_: Yeah I read that ans thought "I wonder how long before someone asks for that"
# [13:58] <tobie_> heh
# [14:01] <jgraham> I pretty much hate the names (no consistency, hard to tell what they do) and the sematics aren't clear to me (the specification seems to assume that I already know how it works)
# [14:07] <tobie_> Yeah. I hate the names too. :)
# [14:09] <tobie_> is that what you're looking at: http://wiki.commonjs.org/wiki/Unit_Testing/1.0
# [14:10] <jgraham> I was looking at 1.1 which had a broken link to 1.0
# [14:10] <zcorpan> jgraham: do i need to make sure to clean up stash from tests?
# [14:11] <tobie_> yeah, I don't think anyone ever implemented 1.1
# [14:11] <tobie_> Think 1 is the standard
# [14:11] <tobie_> checkout what node.js does
# [14:11] <tobie_> and qunit
# [14:12] <jgraham> zcorpan: ?
# [14:12] <jgraham> tobie_: So istm that ok(), equal and notEqual() are harmful because they are type converting
# [14:13] <zcorpan> jgraham: if a test sets stash then fails an assertion, is that ok or should it do add_cleanup(take_stash) ?
# [14:13] <jgraham> deepEqual is something that I think we didn't implement in testharness.js because it's quite hard to do right and produce anything like a useful error message
# [14:14] <jgraham> notDeepEqual seems useless
# [14:14] * Joins: plh (plehegar@public.cloak)
# [14:15] <jgraham> strictEqual and notStrictEqual seem like bad names because they are the form that you want but they are hard to type
# [14:15] <jgraham> throws seems fine
# [14:15] <jgraham> So I am not very impressed by this
# [14:15] <jgraham> zcorpan: You should do, although it only theoretically makes a difference
# [14:17] <zcorpan> jgraham: so why should it do it?
# [14:19] <jgraham> zcorpan: Well if you clean it up then it is deleted on the server side. If you don't clean it up then everything is fine as long as there is no key clash, which there shouldn't be but it's impossible to enforce. Also it leaks a small amount of memory (assuming that your stashed object was small).
# [14:20] <zcorpan> ok, thx
# [15:20] <zcorpan> jgraham: it seems .sub.html {{GET[foo]}} changes & into &. feature or bug?
# [15:23] <jgraham> zcorpan: Feature with a comment in the code that it might be a bug in some cases
# [15:26] <zcorpan> jgraham: i tried using it in a <script> which didn't do what i wanted, but it's easy to fix and actually i don't need this pipe at all, i just thought it would be convenient
# [15:28] <jgraham> Yeah, so I could possibly create some syntax like {{GET[foo]|script}} to specify the type of escaping that you want
# [15:32] <zcorpan> that would backslash-escape " and ' and </script and <!-- or some such? :-)
# [15:33] <jgraham> Something like that
# [15:38] <zcorpan> would be nice
# [15:39] <zcorpan> and \ i guess
# [15:43] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
# [16:34] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
# [16:34] * Joins: zcorpan (~zcorpan@public.cloak)
# [16:41] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
# [16:54] * Joins: zcorpan (~zcorpan@public.cloak)
# [17:03] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
# [17:03] * Joins: zcorpan (~zcorpan@public.cloak)
# [17:10] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
# [17:38] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
# [17:43] * Joins: lmclister (~lmclister@public.cloak)
# [17:48] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
# [18:48] * Joins: Lachy (~Lachy@public.cloak)
# [18:59] * Quits: ptressel (~chatzilla@public.cloak) (Client closed connection)
# [19:00] * Joins: ptressel (~chatzilla@public.cloak)
# [19:04] <tobie_> jgraham: interop or nice APIs: pick one.
# [19:49] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
# [19:50] * Joins: MarkS (msadecki@public.cloak)
# [19:59] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
# [20:42] * Joins: darobin (rberjon@public.cloak)
# [21:03] * Joins: zcorpan (~zcorpan@public.cloak)
# [21:10] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
# [21:28] * Joins: zcorpan (~zcorpan@public.cloak)
# [21:29] * Joins: gitbot (~gitbot@public.cloak)
# [21:29] -gitbot:#testing- [web-platform-tests] pendragon-andyh opened pull request #1036: WebAudio: WaveShaperNode curve tests (master...master) https://github.com/w3c/web-platform-tests/pull/1036
# [21:29] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
# [21:35] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
# [22:03] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
# [22:05] * Joins: Lachy (~Lachy@public.cloak)
# [22:11] * Joins: zcorpan (~zcorpan@public.cloak)
# [22:14] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
# [22:14] * Joins: darobin (rberjon@public.cloak)
# [22:16] * Quits: plh (plehegar@public.cloak) ("Leaving")
# [22:18] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
# [22:19] * Joins: plh (plehegar@public.cloak)
# [22:21] * Quits: darobin (rberjon@public.cloak) (Ping timeout: 180 seconds)
# [22:25] * Quits: ptressel (~chatzilla@public.cloak) (Client closed connection)
# [22:26] * Joins: ptressel (~chatzilla@public.cloak)
# [22:28] * Quits: ptressel (~chatzilla@public.cloak) ("bbs")
# [22:29] * Joins: ptressel (~chatzilla@public.cloak)
# [22:35] * Joins: darobin (rberjon@public.cloak)
# [22:36] * Quits: darobin (rberjon@public.cloak) ("Leaving...")
# [23:13] * Quits: dom (dom@public.cloak) ("")
# [23:15] * Joins: rhauck (~Adium@public.cloak)
# Session Close: Tue Jun 03 00:00:00 2014

The end :)