/irc-logs / w3c / #testing / 2014-03-26 / end

Options:

1. # Session Start: Wed Mar 26 00:00:00 2014
2. # Session Ident: #testing
3. # [00:03] * Quits: lukeis (~Adium@public.cloak) ("Leaving.")
4. # [00:06] * heycam is now known as heycam|away
5. # [00:09] * Quits: mdas (~mdas@public.cloak) (Client closed connection)
6. # [00:14] * heycam|away is now known as heycam
7. # [00:39] * Quits: rhauck (~Adium@public.cloak) ("Leaving.")
8. # [00:41] * Joins: rhauck (~Adium@public.cloak)
9. # [01:08] * Quits: edoyle (~edoyle@public.cloak) ("")
10. # [01:24] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
11. # [01:30] * Joins: ptressel (~chatzilla@public.cloak)
12. # [01:34] * Joins: MK (kawamori@public.cloak)
13. # [01:39] * Quits: rhauck (~Adium@public.cloak) ("Leaving.")
14. # [01:39] * Joins: rhauck (~Adium@public.cloak)
15. # [01:42] * Quits: lmclister (~lmclister@public.cloak) ("")
16. # [01:44] * heycam is now known as heycam|away
17. # [01:46] * Quits: rhauck (~Adium@public.cloak) (Ping timeout: 180 seconds)
18. # [02:18] * Joins: yankhates_cb (~yankhatescb@public.cloak)
19. # [02:26] * Quits: yankhates_cb (~yankhatescb@public.cloak) (Ping timeout: 180 seconds)
20. # [02:38] * Joins: gitbot (~gitbot@public.cloak)
21. # [02:38] -gitbot:#testing- [web-platform-tests] onlimii closed pull request #728: Add iframe support for test runner (master...iframe-for-runner) https://github.com/w3c/web-platform-tests/pull/728
22. # [02:38] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
23. # [03:22] * Joins: mdas (~mdas@public.cloak)
24. # [03:26] * Quits: mdas (~mdas@public.cloak) (Client closed connection)
25. # [03:26] * Joins: mdas (~mdas@public.cloak)
26. # [03:30] * Quits: mdas (~mdas@public.cloak) (Client closed connection)
27. # [04:25] * Joins: rhauck (~Adium@public.cloak)
28. # [04:58] * Quits: ptressel (~chatzilla@public.cloak) ("restart ff")
29. # [05:05] * Joins: mdas (~mdas@public.cloak)
30. # [05:08] * Quits: mdas (~mdas@public.cloak) (Client closed connection)
31. # [05:18] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
32. # [05:30] * Joins: yankhates_cb (~yankhatescb@public.cloak)
33. # [05:40] * Quits: mdyck (~mdyck@public.cloak) ("Leaving")
34. # [05:48] * Quits: rhauck (~Adium@public.cloak) ("Leaving.")
35. # [06:06] * Joins: MK (kawamori@public.cloak)
36. # [06:27] * Quits: yankhates_cb (~yankhatescb@public.cloak) (Ping timeout: 180 seconds)
37. # [06:36] * Joins: yankhates_cb (~yankhatescb@public.cloak)
38. # [06:37] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
39. # [07:06] * Joins: zqzhang (~zqzhang@public.cloak)
40. # [07:14] <zqzhang> odinho: I want to apply the https://github.com/w3c/web-platform-tests/pull/516 with issues in https://critic.hoppipolla.co.uk/showcomments?review=558&filter=all addressed; seems I have to resolve conflicts before continuing. Any tips?
41. # [07:18] <zqzhang> BTW, I want to review and cleanup the 3 open pull requests for Indexed DB in next few days; then tests in IndexedDB/submissions/Opera/
42. # [07:33] <MikeSmith> zqzhang: odinho is on CET time so I reckon you might not hear back from him until another 2 hours or so
43. # [07:35] <zqzhang> MikeSmith: OK, thanks.
44. # [07:43] <MikeSmith> zqzhang: btw sorry I've not sent a message to the mailing list about the w3c-test.org migration yet.
45. # [07:48] <zqzhang> MikeSmith: no problem. I understand that
46. # [07:50] <zqzhang> BTW, I like the w3c-test.org migration :)
47. # [08:11] * Quits: yankhates_cb (~yankhatescb@public.cloak) (Ping timeout: 180 seconds)
48. # [08:20] * Quits: zqzhang (~zqzhang@public.cloak) ("Page closed")
49. # [08:30] * Joins: zcorpan (~zcorpan@public.cloak)
50. # [09:10] * Joins: gitbot (~gitbot@public.cloak)
51. # [09:10] -gitbot:#testing- [web-platform-tests] sideshowbarker pushed 1 new commit to submission/sideshowbarker/dnd: https://github.com/w3c/web-platform-tests/commit/4fbb6456922a930665eb480c1bc5340ecfcefc0f
52. # [09:10] -gitbot:#testing- web-platform-tests/submission/sideshowbarker/dnd 4fbb645 Michael[tm] Smith: Reorganize dnd X-doc test to fit with conventions.
53. # [09:10] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
54. # [10:05] <odinho> jgraham: So, the ttwf critic instance doesn't really support rebasing, or?
55. # [10:05] <odinho> jgraham: How to update to new master without loosing the review?
56. # [10:06] <odinho> Hmm. It does probably with the tarcking, just create a new branch, and point to that.
57. # [10:11] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
58. # [10:14] <MikeSmith> odinho: the critic instance doesn't support rebasing, in my experience
59. # [10:15] <MikeSmith> from several times of having rebasesed a review branch and having it bork up critic and then jgraham had to manually on-bork it on the server side
60. # [10:16] <odinho> M-m. I'm used to doing it all day to our internal one.
61. # [10:26] <jgraham> No no
62. # [10:26] <jgraham> It does support rebasing
63. # [10:26] <jgraham> It's just that the workflow is different
64. # [10:26] <jgraham> odinho: You push the rebased branch (without any new commits)
65. # [10:27] <jgraham> then you go to "Rebase Review"
66. # [10:27] <jgraham> And make sure that upstream is the actual sha1 of the new parent of the branch
67. # [10:27] <jgraham> Because the default value often gives the wrong result
68. # [10:28] <jgraham> Then press fetch branch and check that you only get the commits you expect in the list below
69. # [10:28] <jgraham> Then whichever the button is that actually does the rebase
70. # [10:28] <odinho> So it's the same work-flow for tracking reviews, but not for the push-based ones.
71. # [10:28] <jgraham> Then you reenable tracking
72. # [10:29] <jgraham> Yeah
73. # [10:29] <jgraham> But you don't
74. # [10:29] <odinho> OK, I think I know how to do those now, although I've fscked up a few times before :]
75. # [10:29] <jgraham> "create a new branch"
76. # [10:30] <odinho> OK? Not a -2? That's how we do it, always have branch/1 and push branch/2 and then point to that in the tracking rebase update.
77. # [10:31] <jgraham> No, because that will upset github
78. # [10:31] <odinho> okay :)
79. # [10:31] <jgraham> PRs can't change branch
80. # [10:31] <odinho> ahh, true.
81. # [10:32] <odinho> And when I push the in-place moved branch, github won't send that to critic, confusing it? I don't have to disable tracking first?
82. # [10:34] * Joins: Lachy (~Lachy@public.cloak)
83. # [10:43] * Joins: dom (dom@public.cloak)
84. # [10:50] <jgraham> Github will send it but critic won't like the non-ff update
85. # [11:06] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
86. # [11:06] * Joins: zcorpan (~zcorpan@public.cloak)
87. # [11:19] * Joins: abarsto (~abarsto@public.cloak)
88. # [11:19] * abarsto is now known as ArtB
89. # [12:03] * Joins: ptressel (~chatzilla@public.cloak)
90. # [12:05] * Joins: darobin (rberjon@public.cloak)
91. # [12:25] * Joins: gitbot (~gitbot@public.cloak)
92. # [12:25] -gitbot:#testing- [web-platform-tests] AFBarstow opened pull request #801: Afbarstow/warning (master...AFBarstow/WARNING) https://github.com/w3c/web-platform-tests/pull/801
93. # [12:25] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
94. # [12:37] <jgraham> ArtB: YOu need to create a seperate branch for each PR. That one has ended up with a commit from a different PR in it
95. # [12:38] <jgraham> Also I think that WARNING file sort of misses the point
96. # [12:38] <jgraham> The heart of the issue is that w3c-test.org only serves the HEAD of the testsuite repo
97. # [12:39] <jgraham> For an implementation report you want to link to a specific revision of the testsuite against which the report was made
98. # [12:40] <jgraham> w3c-test.org *could* be taught to allow that (I assume this is the solution darobin had in mind), but it would have substantial associated costs
99. # [12:40] <ArtB> hmmm; I thought I did create a new branch
100. # [12:41] <jgraham> Possibly you did, but got the wrong base for the branch
101. # [12:42] <jgraham> The foolproof way of doing things is git fetch origin; git checkout -b new_branch_name origin/master
102. # [12:42] <jgraham> (assuming origin is the w3c github repo)
103. # [12:46] <ArtB> what's the magic to delete a PR?
104. # [12:48] <jgraham> Just hit "close" on the github page
105. # [12:48] * Joins: gitbot (~gitbot@public.cloak)
106. # [12:48] -gitbot:#testing- [web-platform-tests] AFBarstow closed pull request #801: Afbarstow/warning (master...AFBarstow/WARNING) https://github.com/w3c/web-platform-tests/pull/801
107. # [12:48] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
108. # [12:48] <zcorpan> jgraham: /html/infrastructure/urls/resolving-urls/query-encoding/ is not stable, how surprising :-)
109. # [12:50] <jgraham> :)
110. # [12:50] <MikeSmith> ArtB: I will work on fixing the 404s I caused with the w3c-test.org migration. We can't practically set up a rewrite mechanism on w3c-test.org now so it amounts to me copying the stuff to somewhere else and going through the TRs and fixing the links. I think I won't be able to do it til tomorrow. I had been planning on doing it yesterday but had a family emergency and was afk for pretty much the entire day.
111. # [12:52] <zcorpan> seems like a lot of tests in that list are tests i've written :-(
112. # [12:52] <jgraham> zcorpan: I think you've just written a lot of tests :)
113. # [12:53] <jgraham> (also for things that are unusually likely to be problematic like workers)
114. # [13:01] <jgraham> MikeSmith: Did you check that your changes to the DnD test in r/1084 actually work?
115. # [13:02] * Joins: plh (plehegar@public.cloak)
116. # [13:02] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
117. # [13:03] <MikeSmith> jgraham: yeah, produce exactly the same results
118. # [13:03] <darobin> I don't think that w3c-test.org should keep the backlog of all commits in an addressable manner
119. # [13:04] <darobin> that would just be too costly
120. # [13:04] <darobin> that said, I reckon git can make it relatively cheap, but still
121. # [13:05] <darobin> zcorpan: I would think that testing things that cause instability is a badge of honour rather than a cause for ":-("
122. # [13:06] <jgraham> darobin: I agree that it shouldn't. But I assume that's what you had in mind with "there are other options, but they are heavy-handed"
123. # [13:07] <jgraham> darobin: Depends a bit on why the test is unstable. But it's never a good thing for the CI system, even if it's a flaw in the product under test
124. # [13:07] <darobin> jgraham: yeah, not giving anyone the idea that it was technically possible was my reason for sweeping those behind a cryptic sentence :)
125. # [13:08] <darobin> jgraham: oh definitely, I just meant that I certainly don't think that zcorpan should sadface for having worked on that stuff
126. # [13:09] * Joins: gitbot (~gitbot@public.cloak)
127. # [13:09] -gitbot:#testing- [web-platform-tests] jgraham closed pull request #796: Added -manual filename suffix to some manual dnd test files. (master...submission/sideshowbarker/dnd) https://github.com/w3c/web-platform-tests/pull/796
128. # [13:09] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
129. # [13:27] <zcorpan> darobin: it's hard to know why a test is unstable. but obviously i like to think that there are bugs in gecko that the tests are exposing
130. # [13:27] <zcorpan> but i think i have a history of making mistakes, too
131. # [13:30] <darobin> zcorpan: I would be surprised if some of those tests weren't bugs in the engine
132. # [13:31] <darobin> of course I would also be surprised if they were all bugs in the engine; it's pretty easy to find race conditions on the web
133. # [13:33] * Joins: anssik (~uid10742@public.cloak)
134. # [13:35] * Joins: Ms2ger (~Ms2ger@public.cloak)
135. # [13:36] * Joins: jimevans (~jimevans@public.cloak)
136. # [13:36] * Quits: jimevans (~jimevans@public.cloak) ("Colloquy for iPhone - http://colloquy.mobi")
137. # [14:35] * Joins: gitbot (~gitbot@public.cloak)
138. # [14:35] -gitbot:#testing- [web-platform-tests] deniak closed pull request #783: tests on template element (master...template-element) https://github.com/w3c/web-platform-tests/pull/783
139. # [14:35] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
140. # [14:37] <MikeSmith> jgraham: darobin: at some point we should move the tests in the /html-templates into /html/semantics/scripting-1/the-template-element
141. # [14:38] <darobin> good point
142. # [14:38] <darobin> I don't see a reason not to just do it now
143. # [14:39] * Joins: gitbot (~gitbot@public.cloak)
144. # [14:39] -gitbot:#testing- [web-platform-tests] sideshowbarker deleted submission/sideshowbarker/dnd at 4fbb645: https://github.com/w3c/web-platform-tests/commit/4fbb645
145. # [14:39] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
146. # [14:39] <darobin> mmmm, some rewriting needed
147. # [14:39] <darobin> I'll do it
148. # [14:39] <MikeSmith> ok
149. # [14:44] * Joins: gitbot (~gitbot@public.cloak)
150. # [14:44] -gitbot:#testing- [web-platform-tests] deniak closed pull request #781: test on document.open() (master...opening-input-stream) https://github.com/w3c/web-platform-tests/pull/781
151. # [14:44] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
152. # [15:12] <plh> hu, I get a weird 404? error on http://w3c-test.org/navigation-timing/idlharness.html with chrome but not with firefox
153. # [15:12] <plh> I don't get the error with my local up-to-date copy
154. # [15:14] * Joins: mdas (~mdas@public.cloak)
155. # [15:17] <darobin> plh: I get a 404 with both
156. # [15:17] <darobin> and the server is extremely slow to respond
157. # [15:17] <plh> well, the good news is that I don't need the server to run the test :)
158. # [15:18] <plh> and it's probably on the w3c-test.org side more than wptserve side. if the server had a leak or something, james would have probably ran into it by now
159. # [15:20] <darobin> I wouldn't suspect a leak as the first issue indeed
160. # [15:20] * plh tries to remember w3c-test.org real name
161. # [15:22] <darobin> grumple
162. # [15:22] <darobin> I see no excessive load
163. # [15:22] <plh> 99% of the CPU is from python
164. # [15:23] <darobin> mmmm, that's not reported in uptime
165. # [15:23] <darobin> woah
166. # [15:23] <darobin> it was at 6% a minute ago
167. # [15:23] <darobin> and now all the pages are 404ing
168. # [15:24] <plh> let's keep looking for a cause before killing it
169. # [15:25] <tobie__> Re: ArtB's WARNING doc, I suggest instead installing a git Web interface (e.g. http://gitlist.org/) and pointing git.testthewebforward.org to it.
170. # [15:25] <tobie__> Then using that as canonical URLs
171. # [15:26] <plh> darobin, are you able to decrypt /var/log/wptserve.log ?
172. # [15:26] <tobie__> FYI I ran some websockets tests earlier and the runner was behaving weirdly
173. # [15:27] <tobie__> (Tests kept on running despite having stopped them, had to navigate await from the runner to get it to stop)
174. # [15:30] * Joins: MK (kawamori@public.cloak)
175. # [15:30] <jgraham> tobie__++ on the git idea
176. # [15:31] <darobin> https://gist.github.com/anonymous/f55907000bc1780ac103
177. # [15:32] <darobin> too many open files
178. # [15:32] <darobin> jgraham: is it possible that wptserve isn't releasing open files?
179. # [15:33] <darobin> also, I notice that it says "DEBUG:wptserve:/u/www.w3c-test.org/web-platform-tests/master/" as it were doing it three times
180. # [15:35] * Joins: JohnJansen (~JohnJansen@public.cloak)
181. # [15:35] <jgraham> darobin: It's possible of course
182. # [15:36] <jgraham> I doubt anyone has had it running for this long before
183. # [15:36] <darobin> jgraham: by that I meant more precisely "do you have an idea of where it could come from" :)
184. # [15:36] <darobin> I know it's possible
185. # [15:36] <jgraham> :p
186. # [15:38] <jgraham> Well
187. # [15:39] <plh> darobin, at this point, can I restart the ptyhon server btw?
188. # [15:39] <darobin> plh: yeah go ahead
189. # [15:39] <darobin> I reckon I've gather as much information from the logs as I will
190. # [15:39] <jgraham> There aren't that many places that explicitly open() files and they should all close the files when they are done, unless the request hangs somehow
191. # [15:40] <jgraham> It could be that one of the .py scripts is opening a file and then hanging.
192. # [15:40] <darobin> which I guess could happen if we're indeed seeing multiple instances try to answer
193. # [15:40] <plh> we had a lot of TCP connections open btw
194. # [15:41] <plh> opened
195. # [15:41] <darobin> mmmmm
196. # [15:41] * Quits: JohnJansen (~JohnJansen@public.cloak) ("")
197. # [15:41] <plh> netstat gives a long list
198. # [15:43] <jgraham> Hmm. So I think the three lines in the log might be misleading
199. # [15:43] <plh> could it be that wptserve isn't closing the connections?
200. # [15:43] <jgraham> That's just logged whenever something tries to get the filesystem path corresponding to a request
201. # [15:44] <darobin> jgraham: on my local copy it just logs one line
202. # [15:44] <darobin> most of those open sockets are HTTP indeed
203. # [15:45] <plh> yep, and got closed after I restarted the server
204. # [15:45] <jgraham> Right, so there is at least one situation in which the connection might not get closed
205. # [15:45] <jgraham> Which is that if we had a handler function that did something that took a long time
206. # [15:46] <jgraham> like while True: pass
207. # [15:46] * Joins: gitbot (~gitbot@public.cloak)
208. # [15:46] -gitbot:#testing- [web-platform-tests] deniak opened pull request #802: import tests on document.open() from presto-testo (master...open-input-stream) https://github.com/w3c/web-platform-tests/pull/802
209. # [15:46] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
210. # [15:46] <darobin> if we're getting a handler that's stuck with an open file it would also get stuck with the open socket
211. # [15:46] <jgraham> then there isn't a mechanism to kill that handler
212. # [15:46] <jgraham> Well an open socket is an open file, right?
213. # [15:47] <plh> I believe so
214. # [15:47] <darobin> yes, but while there were lots of open sockets there weren't *that* many
215. # [15:47] <darobin> it was like 2-3 screenfuls, not thousands
216. # [15:47] <darobin> which is what I expect the order of magnitude for open files is
217. # [15:48] <darobin> 98576
218. # [15:48] <darobin> there weren't 100k open sockets there
219. # [15:48] <jgraham> Is that the ulimit -n?
220. # [15:48] <MikeSmith> tobie__: on what host would we install the git web interface?
221. # [15:49] <jgraham> It can be smaller than the theoretical maximum
222. # [15:49] <plh> ulimit -n returns 1024
223. # [15:49] <jgraham> Sounds like normal for Linux
224. # [15:49] <jgraham> (on OSX it's like 256 or something tiny by default)
225. # [15:50] <darobin> oh, you're right, it's ulimit
226. # [15:50] <darobin> but it's more than 1024
227. # [15:50] <darobin> the hard limit is 4096
228. # [15:50] <darobin> there weren't 4096 (or even 1024) open sockets in the netstat output I saw
229. # [15:50] <darobin> mmmm, which makes me note that wptserve runs as root :)
230. # [15:51] <jgraham> Yeah, I guess it ought to do wahtever the magic is to open the socket as root and then drop permissions
231. # [15:52] <plh> we were well under 500 according to netstat
232. # [15:53] <jgraham> (or there seems to be something called authbind that you could use to allow the specific user permissions to access port 80 without running as root)
233. # [15:54] <darobin> I see os.setuid() in the docs
234. # [15:54] <jgraham> http://stackoverflow.com/questions/2699907/dropping-root-permissions-in-python
235. # [15:55] <jgraham> Anyway, the only thing that I can think is that somehow some request threads are getting stuck and holding open resources longer than they should
236. # [15:55] <jgraham> I'm not entirely sure how easy this is to debug
237. # [15:57] <plh> one easy solution would be to restart the server at regular interval
238. # [15:57] <plh> to keep the server as clean as possible
239. # [15:57] <plh> since some handlers might go crazy
240. # [15:58] <plh> but it might remove from us an ooprtunity to catch those
241. # [15:59] <plh> I suspect most of the connections were crawlers btw
242. # [16:02] <jgraham> So https://gist.github.com/jgraham/2ee49bdb1c7c83ba9b43 is a script I cribbed off the internet to detect what each thread in a python app is doing
243. # [16:02] <jgraham> You could try running with that enabled for a while and look for evidence of long-lived threads
244. # [16:03] * Joins: yankhates_cb (~yankhatescb@public.cloak)
245. # [16:07] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
246. # [16:07] <tobie__> MikeSmith: not sure. Whatever works best between an MIT server, a random host (e.g. Dreamhost), the w3c-test server.
247. # [16:12] * Joins: MK (kawamori@public.cloak)
248. # [16:18] * Quits: yankhates_cb (~yankhatescb@public.cloak) (Ping timeout: 180 seconds)
249. # [16:23] <odinho> But the git interface won't be able to run the tests. What is it the specs actually want to reference? The code itself, or the test you could run? If the former, then something like cgit (or the php-based github-ripoff suggested :P) would work very well.
250. # [16:24] <odinho> But yea. ... What would be the use of running old tests anyway.
251. # [16:24] <odinho> I can see some limited value, but it's very limited. In most cases it'd just be wrong.
252. # [16:25] <jgraham> Yeach, but on the other hand there is value in saying "these were the tests that we actually ran"
253. # [16:26] <odinho> Yes. Though it would be quite easy to do that with git and wptserve now.
254. # [16:26] <odinho> You just have to do it locally.
255. # [16:26] <jgraham> Sure
256. # [16:27] <jgraham> I think that's a fine solution for "I want to rerun all the tests that were run for this IR"
257. # [16:28] <jgraham> And I agree that's a sufficiently narrow use case that building out all the infrastructure needed to support it from w3c-test.org is a losing proposition
258. # [16:28] <plh> one use case that I have is to be able to link to a test so that others can run easily. for example, I sent a message to public-web-perf earlier, linking to a test in github, http://lists.w3.org/Archives/Public/public-web-perf/2014Mar/0030.html
259. # [16:28] <plh> but if I put a link to w3c-test.org, that link will be unstable
260. # [16:29] <jgraham> Yes
261. # [16:29] <plh> so I refrained to link to it, making it harder for others to simply try it
262. # [16:29] * Joins: lukeis (~Adium@public.cloak)
263. # [16:29] <jgraham> But if someone fixes a bug in that test you don't want them to run the old version anyway
264. # [16:29] <plh> agreed
265. # [16:30] <jgraham> So the only problematic case is where a test gets moved and we don't encounter that so often that it's worth building a complex solution just to deal with that
266. # [16:36] <darobin> jgraham: at any rate I can confirm that under normal operations on grumple wptserve only logs the file it's about to send once
267. # [16:36] <jgraham> darobin: Curious
268. # [16:36] <darobin> so clearly when it went wrong there were three parallel attempts at delivery for each request
269. # [16:37] <darobin> at least up to the point where that logging happens
270. # [16:37] <jgraham> Curiouser and curiouser
271. # [16:37] <darobin> somehow I thought wptserve was multiprocess rather than multithread
272. # [16:38] <darobin> jgraham: do you know if it's easy to add some form of thread ID to the log?
273. # [16:38] <jgraham> darobin: Yeah it should be
274. # [16:38] <darobin> that at least might help make the issue clearer
275. # [16:39] <darobin> re the git stuff, if all you want to record is what the tests were that you ran against, just include the commitish in the implementation report and call it a day
276. # [16:39] <darobin> no need for infrastructure
277. # [16:39] <darobin> whoever needs to repro can
278. # [16:41] <jgraham> http://docs.python.org/2/howto/logging.html#changing-the-format-of-displayed-messages and http://docs.python.org/2/library/logging.html#logrecord-attributes
279. # [16:41] <jgraham> So you basically need to add %(threadName)s to the format somewhere
280. # [16:42] <jgraham> (well it currently just uses the default format)
281. # [16:42] <jgraham> (it might also be possible to give some threads useful names, I'm not sure)
282. # [16:42] <darobin> oh fuck me
283. # [16:42] * jgraham declines
284. # [16:43] <darobin> IE still has problems producing the JSON results if the test suite is too big
285. # [16:43] <darobin> I thought I had content but it was broken
286. # [16:43] <jgraham> Welcome to the 2014 web
287. # [16:43] <jgraham> Where producing a few 10s of Mb of text is too hard
288. # [16:44] <jgraham> (maybe even less)
289. # [16:44] <darobin> jgraham: it's even less than that :)
290. # [16:48] * Joins: lmclister (~lmclister@public.cloak)
291. # [16:48] * Joins: kawabata2 (~kawabata@public.cloak)
292. # [16:54] * Joins: rhauck (~Adium@public.cloak)
293. # [16:58] * Joins: glenn (~gadams@public.cloak)
294. # [16:58] * astearns__ is now known as astearns
295. # [17:03] * Joins: abarsto (~abarsto@public.cloak)
296. # [17:03] * Quits: ArtB (~abarsto@public.cloak) (Client closed connection)
297. # [17:04] <darobin> plh: http://berjon.com/tmp/test-results.zip has a preliminary report (21MB) for the DOM TS
298. # [17:05] <darobin> I had to exclude IE11 since I still can't get working JSON for something this large — Erika is going to send me something from her end
299. # [17:05] <darobin> it's a long report though, hard to make much sense from it
300. # [17:05] <darobin> I want to add a filter to only get stuff that doesn't pass in two implementations
301. # [17:05] <darobin> (but that will make a lot more sense once I have IE results)
302. # [17:06] <plh> nice. and yes, I figured that most difficult part is to make of the results
303. # [17:06] <plh> s//sense/
304. # [17:07] <darobin> plh: I reckon a view that only shows tests that don't have at least two successes might do the trick
305. # [17:07] <darobin> for one it would make for a much smaller document :)
306. # [17:07] <jgraham> darobin: Did you use the report.py file in the tree?
307. # [17:07] <plh> that's a good start indeed
308. # [17:08] <darobin> jgraham: yes, I had started writing my own when I noticed that was there...
309. # [17:08] <jgraham> The one in the tree does the "list of tests without two passes" thing
310. # [17:08] <jgraham> iirc
311. # [17:08] <darobin> oh
312. # [17:09] <darobin> jgraham: I don't see that
313. # [17:09] <darobin> I'm in master and up to date
314. # [17:09] <darobin> it tells me the tests that fail in all UAs
315. # [17:09] <darobin> (the vast majority of those being interface tests)
316. # [17:09] <jgraham> Ah, right
317. # [17:10] <jgraham> It wasn't has "two passes" it was "doesn't have any passes"
318. # [17:10] <darobin> yeah
319. # [17:11] <darobin> I was thinking of modifying that code, but then I would actually like to produce output that only has that information
320. # [17:11] <darobin> (more manageable)
321. # [17:12] <jgraham> Well
322. # [17:13] <jgraham> I'm not sure it's a good way of looking at things even from an IR point of view
323. # [17:13] <jgraham> Because it hides the fact that you might have very good interop by showing only the bad cases
324. # [17:13] <jgraham> So people think that they have to drive the bad cases to 0
325. # [17:13] <darobin> I don't want to use that as the IR
326. # [17:13] <darobin> I just want to use that as information to know what might need to be dropped from the snapshot
327. # [17:14] <darobin> or what needs some weasel^Winteresting explanation for the Director
328. # [17:14] <jgraham> Oh well if you are already doing silly things I guess doing more silly things to facilitate the existing silliness isn't so silly
329. # [17:14] <darobin> (e.g. the WebIDL stuff I reckon we can make a case)
330. # [17:14] <darobin> jgraham: you have captured the very essence of my existence
331. # [17:15] <dom> s/my existence/life
332. # [17:16] <darobin> heh
333. # [17:16] <darobin> oh, this is amusing
334. # [17:16] <darobin> so I managed to lock up IE with this huge piece of JSON
335. # [17:16] <darobin> it asked me if I wanted to "Recover the page", so I said hell yeah, why not
336. # [17:16] <darobin> it then showed me a dialog for ten seconds saying "Windows is looking for a solution to the problem"
337. # [17:17] <darobin> then it reloaded to page and showed it to
338. # [17:17] <darobin> me
339. # [17:18] <jgraham> Well it obviously found a solution to the problem
340. # [17:19] <darobin> I guess I can't complain
341. # [17:29] * Quits: ptressel (~chatzilla@public.cloak) (Client closed connection)
342. # [17:31] * Joins: gitbot (~gitbot@public.cloak)
343. # [17:31] -gitbot:#testing- [web-platform-tests] darobin created dump-json (+1 new commit): https://github.com/w3c/web-platform-tests/commit/8aaa8eb60e69
344. # [17:31] -gitbot:#testing- web-platform-tests/dump-json 8aaa8eb Robin Berjon: add option to just dump JSON to a textarea
345. # [17:31] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
346. # [17:32] * Joins: gitbot (~gitbot@public.cloak)
347. # [17:32] -gitbot:#testing- [web-platform-tests] darobin opened pull request #803: add option to just dump JSON to a textarea (master...dump-json) https://github.com/w3c/web-platform-tests/pull/803
348. # [17:32] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
349. # [17:32] * Quits: kawabata2 (~kawabata@public.cloak) (Client closed connection)
350. # [17:33] * Quits: mdas (~mdas@public.cloak) (Client closed connection)
351. # [17:39] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
352. # [17:40] * Joins: rhauck1 (~Adium@public.cloak)
353. # [17:46] * Quits: rhauck (~Adium@public.cloak) (Ping timeout: 180 seconds)
354. # [17:54] * Joins: gitbot (~gitbot@public.cloak)
355. # [17:54] -gitbot:#testing- [web-platform-tests] darobin pushed 1 new commit to dump-json: https://github.com/w3c/web-platform-tests/commit/51a6a7e7753a11acf9166f9aea7007e5fbef11a9
356. # [17:54] -gitbot:#testing- web-platform-tests/dump-json 51a6a7e Robin Berjon: remove leftover value attribute; put ugly braces
357. # [17:54] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
358. # [17:55] * Joins: gitbot (~gitbot@public.cloak)
359. # [17:55] -gitbot:#testing- [web-platform-tests] foolip closed pull request #729: Test when the muted IDL attribute is updated from the content attribute (master...muted) https://github.com/w3c/web-platform-tests/pull/729
360. # [17:55] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
361. # [18:01] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
362. # [18:02] * Joins: zcorpan (~zcorpan@public.cloak)
363. # [18:06] * Joins: MK_ (kawamori@public.cloak)
364. # [18:09] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
365. # [18:11] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
366. # [18:14] * Quits: dom (dom@public.cloak) ("")
367. # [18:21] <darobin> progress: I now have some JSON
368. # [18:21] <darobin> problem: I am still getting hard to track errors with it
369. # [18:23] <jgraham> What kind of errors?
370. # [18:26] <darobin> "Unexpected token"
371. # [18:26] <darobin> but it all looks nice and rosy
372. # [18:26] * Quits: anssik (~uid10742@public.cloak) ("Connection closed for inactivity")
373. # [18:26] <darobin> Now successfully killed in the process of getting the IE JSON report for the full DOM suite: IE, TextMate, Firefox, Chrome
374. # [18:32] <darobin> ok, I'm giving up — I'll find the solution tomorrow
375. # [18:32] * Joins: zcorpan (~zcorpan@public.cloak)
376. # [18:33] * Joins: gitbot (~gitbot@public.cloak)
377. # [18:33] -gitbot:#testing- [web-platform-tests] darobin pushed 1 new commit to dump-json: https://github.com/w3c/web-platform-tests/commit/44924dd61e559cecef2ff090cf4362402e3a1d2c
378. # [18:33] -gitbot:#testing- web-platform-tests/dump-json 44924dd Robin Berjon: make it clearer what this is for
379. # [18:33] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
380. # [18:34] <jgraham> darobin: ed won't let you down
381. # [18:36] * Joins: gitbot (~gitbot@public.cloak)
382. # [18:36] -gitbot:#testing- [web-platform-tests] jgraham pushed 1 new commit to master: https://github.com/w3c/web-platform-tests/commit/3c0abe638997ede640f5d88e22e69530112b03d1
383. # [18:36] -gitbot:#testing- web-platform-tests/master 3c0abe6 jgraham: Merge pull request #803 from w3c/dump-json...
384. # [18:36] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
385. # [18:39] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
386. # [18:43] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
387. # [18:53] * Quits: Lachy (~Lachy@public.cloak) ("My MacBook Pro has gone to sleep. ZZZzzz…")
388. # [19:20] * abarsto is now known as ArtB
389. # [19:33] * Joins: zcorpan (~zcorpan@public.cloak)
390. # [19:34] * Joins: rhauck (~Adium@public.cloak)
391. # [19:38] * Quits: rhauck1 (~Adium@public.cloak) (Ping timeout: 180 seconds)
392. # [19:39] * Joins: Lachy (~Lachy@public.cloak)
393. # [19:40] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
394. # [19:43] * Quits: rhauck (~Adium@public.cloak) ("Leaving.")
395. # [19:48] * Joins: rhauck (~Adium@public.cloak)
396. # [19:55] * Joins: rhauck1 (~Adium@public.cloak)
397. # [20:00] * Quits: rhauck (~Adium@public.cloak) (Ping timeout: 180 seconds)
398. # [20:05] * Joins: mdas (~mdas@public.cloak)
399. # [20:24] * Quits: rhauck1 (~Adium@public.cloak) (Client closed connection)
400. # [20:25] * Quits: ArtB (~abarsto@public.cloak) ("Leaving.")
401. # [20:29] * Joins: rhauck (~Adium@public.cloak)
402. # [20:34] * Joins: zcorpan (~zcorpan@public.cloak)
403. # [20:41] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
404. # [20:49] * Quits: MK_ (kawamori@public.cloak) (Ping timeout: 180 seconds)
405. # [20:54] * Joins: MK (kawamori@public.cloak)
406. # [21:07] * Quits: MK (kawamori@public.cloak) (Ping timeout: 180 seconds)
407. # [21:26] * Quits: rhauck (~Adium@public.cloak) ("Leaving.")
408. # [21:35] * Joins: zcorpan (~zcorpan@public.cloak)
409. # [21:38] <plh> anyone around who has chrome installed on android and confirm that they see the video at http://jay.w3.org/~plehegar/videos/ ?
410. # [21:38] <plh> it's a simple test video that works everywhere, except on my chrome in android (works fine in firefox and opera mobile)
411. # [21:40] <jgraham> zcorpan: Seems some tests for workers should set setup({allow_uncaught_exceptions: true}) because the worker error propogates to the top level window
412. # [21:42] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
413. # [21:49] * heycam|away is now known as heycam
414. # [21:57] <astearns> plh: works in chrome beta on android
415. # [21:57] <plh> thank you Alan!
416. # [21:57] <plh> I suspect my chrome is somehow broken
417. # [21:58] <plh> since it plays the sound but doesn't show the image
418. # [21:58] * Quits: lukeis (~Adium@public.cloak) ("Leaving.")
419. # [21:59] <astearns> and in the regular chrome - android 4.4.2
420. # [22:00] <plh> thank you again
421. # [22:03] * Joins: zcorpan (~zcorpan@public.cloak)
422. # [22:11] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
423. # [22:12] * Joins: zcorpan (~zcorpan@public.cloak)
424. # [22:19] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
425. # [22:42] * Joins: zcorpan (~zcorpan@public.cloak)
426. # [22:44] * Quits: zcorpan (~zcorpan@public.cloak) (Client closed connection)
427. # [22:44] * Joins: zcorpan (~zcorpan@public.cloak)
428. # [22:51] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
429. # [22:58] * Joins: lukeis (~Adium@public.cloak)
430. # [23:37] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
431. # [23:37] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
432. # [23:44] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Ping timeout: 180 seconds)
433. # [23:45] * Joins: zcorpan (~zcorpan@public.cloak)
434. # [23:52] * Quits: zcorpan (~zcorpan@public.cloak) (Ping timeout: 180 seconds)
435. # Session Close: Thu Mar 27 00:00:00 2014

The end :)