15:04:08 RRSAgent has joined #htmlt
15:04:08 logging to http://www.w3.org/2011/10/04-htmlt-irc
15:05:32 Agenda http://lists.w3.org/Archives/Public/public-html-testsuite/2011Oct/0002.html
15:06:03 Agenda Item #1 Bugs on Approved Tests
15:06:05 http://www.w3.org/Bugs/Public/buglist.cgi?query_format=advanced&short_desc_type=allwordssubstr&short_desc=&product=HTML+WG&component=testsuite&long_desc_type=allwordssubstr&long_desc=&bug_file_loc_type=allwordssubstr&bug_file_loc=&status_whiteboard_type=allwordssubstr&status_whiteboard=&keywords_type=allwords&keywords=&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&emailtype1=substring&email1=&emailtype2=substring&email2=&bugidtype=include&bug_id=&votes=
15:07:11 bug 14240?
15:07:43 A few new bugs (canvas tests)
15:07:44 http://www.w3.org/Bugs/Public/show_bug.cgi?id=14240
15:08:57 Which wants to replace CanvasPixelArray with Uint8ClampedArray
15:09:58 http://www.w3.org/Bugs/Public/show_bug.cgi?id=14326
15:11:55 Which is an update due to webidl changes with rounding
15:12:45 http://www.w3.org/Bugs/Public/show_bug.cgi?id=14356
15:12:54 is not really a bug rather a question
15:13:51 If anyone has any comments on these bugs please speak up
15:16:19 no comments on my side (didn't look at the bugs)
15:16:54 I'll need to look at the webidl stuff and comment back next time if this is correct
15:17:25 lets move on to the next agenda item
15:17:44 Agenda Item #2 test re-use/license question
15:18:32 on bug 14240, I would note that http://dev.w3.org/html5/2dcontext/Overview.html#imagedata didn't change
15:18:55 so I suggest holding on it until the the canvas2d spec is changed
15:18:59 At our last meeting some conserns were raised about dat file re-use out side of the w3c
15:19:53 specifically if dat files start to appear in the w3c html5 tests for parser testing would they also be able to be leveraged by HTML5lib?
15:20:31 I guess I don't have enough background to understand the issue here
15:20:51 what's the best way to get up to speed?
15:21:00 keep listening :)
15:21:04 ok :)
15:21:36 I'm not very familiar with HTML5lib parser tests and where they are stored
15:21:43 Though it's not a w3c resource
15:22:21 Though the community has been using this for a common location for HTML5 parser tests
15:22:31 the parser came from an outside source indeed but, like the canvas tests, we should have a copy of them
15:23:09 I think that is fine, though it's about when someone only submits new parser tests to the w3c test suite
15:23:58 Could someone make a copy of the w3c tests and submit them to the HTML5lib location?
15:24:16 I would suggest to try to push the changes upstream indeed
15:24:56 and yes, I believe our license would authorize to make a copy
15:25:22 jgraham does this answer your question?
15:26:22 I guess James isn't around
15:26:49 We don't need an instant answer...
15:27:27 if we can't push the changes upstream, I guess we could have a separate directory for those, but we should try to avoid that if possible
15:27:45 which way is upstream?
15:27:56 from HTML5Lib -> W3C?
15:28:06 nope, the other way around
15:28:13 since the source is HTML5Lib
15:28:39 yep that is the question
15:28:56 if someone would want to push a set of tests upstream - could they do this?
15:29:26 Seems like the answer is yes as long as they keep the w3c test license
15:30:20 that could be a problem for html5lib, since I guess they use a different license
15:30:44 do we have a case of someone wanted to make changes to html5lib tests and not willing to push them upstream until the html5lib license?
15:31:30 No one wants to change the html5lib tests
15:31:45 so, we're only talking about additions?
15:31:50 correct
15:32:20 so, if those additions can't be put under the html5lib license, I suggest having a separate directory for those
15:33:29 So then you are suggesting that on the w3c side additional html5lib parser tests could be added in a separate directory with the html5lib license?
15:33:37 I'm not a lawyer :)
15:33:57 But this sounds like this would help keep these tests in sync
15:35:50 Plh does this give you enough background about the issue?
15:36:14 yes
15:36:23 Good
15:36:37 if the tests cannot be put under the html5lib license, those tests are not html5lib tests
15:36:46 thus a separate directory
15:37:27 thanks
15:38:12 Since this is not a simple issue, seems appropriate to have you confirm that this is what you want to happen
15:38:22 sure
15:38:35 that's my suggestion at least
15:38:54 sounds good
15:39:04 next agenda item new test submissions
15:40:43 see http://dvcs.w3.org/hg/html/shortlog
15:41:35 Note that base64.html changes/updates
15:42:33 Note no new class of test submissions or new submitters
15:43:21 Last Agenda Item - Test Review Period from Oct 15th -> Dec 15th
15:43:29 This is more of a FYI to the group
15:44:48 plh shall we adjourn?
15:45:19 RRSAgent, make logs public
15:45:41 Meeting adjourned
15:45:43 rrsagent, generate minutes
15:45:43 I have made the request to generate http://www.w3.org/2011/10/04-htmlt-minutes.html krisk
15:45:53 see you at TPAC
15:49:32 I totally was around and didn't norice the time or the channel change colour
15:54:52 So the conclusion is that it isn't possible to move tests from W3C to html5lib
15:55:08 And per last week microsoft won't submit to html5lib
15:55:15 So we are forking the tests
15:55:17 Yay
17:15:39 Zakim has left #htmlt
17:42:59 hi James
17:43:16 my understanding is that we're not forking the current tests
17:43:25 we're creating additional tests
19:38:18 plh: That is forking from my point of view
19:39:43 It creates pain when generating the testharness.js copies of the tests and creates an oppertunity for people to get an incomplete set of parsing tests
19:44:00 hum, I'm afraid I'm not understanding why it creates additional pain. the html test suite already has tests that don't come from html5lib. the test framework can easily with several tests, each running testharness.js, the fact we have two directories with different set of tests for the html5 parser doesn't add a lot of pain.
19:44:22 It does if they both use the .dat format
19:44:44 and expect to generate the testharness.js format using the existing script
19:45:10 That script assumes that the .dat files are in a html5lib checkout
19:45:29 ok, then I guess I don't understand the .dat format enough and some of the implications here
19:46:26 Basically the .dat format is just a reasonaby concise way of writing input/expected for tests that only depend on the parser
19:46:45 But to actually run the tests you need a way to feed the data into your parser
19:46:57 In our case that is the web browser
19:47:07 There are a number of ways to do this of course
19:47:35 One could load the .dat files using e.g. XHR and parse them in javascript for example
19:48:48 The approach I took, to avoid dependencies on too many other bits of the platform, was to generate static html/javascript files using a python script that reads the .dat files and spits out html
19:49:33 The script depends on html5lib somewhat, not least in the fact that it reuses the infrastructure for locating and loading the testcases
19:49:55 That dependency can be weakened or removed of course
19:50:36 But it is pain that seems wholely unneeded when it comes down to BSD vs MIT license issues