[00:00]  * veebers reboots
[00:15] <veebers> thomi: we may have an issue with AP 1.5 & filemanager (http://q-jenkins.ubuntu-ci:8080/job/autopilot-release-gatekeeper/128/) I'm re-re-re-running with the ppa enabled this time in case there was a recent fix or something
[00:15] <veebers> thomi: to clarify, the linked job was without the ppa 100% passing
[00:16] <veebers> trying now with the ppa to get (more) resutls
[00:16] <veebers> results*
[00:16] <thomi> hmm, interesting
[00:16] <thomi> I'll start flashing my device
[00:17] <veebers> thomi: now that mine is actually charged I'll do the same
[02:16] <veebers> thomi: I see that you fired off another gatekeeper for filemanager, but it seems to have had troubles
[02:16] <thomi> ugh, what happened?
[02:16]  * veebers looks futher
[02:16] <thomi> I probably messed it up somehow
[02:17] <thomi> veebers: BTW, you need to manually install python3-autopilot for now
[02:17] <veebers> thomi: hmm, yeah "filemanager_app.phablet-test-run"
[02:17] <veebers> it should be just filemanager
[02:17] <veebers> (no -app)
[02:17] <thomi> ugh
[02:17] <thomi> can you re-kick off please/
[02:17] <thomi> ?
[02:17] <veebers> thomi: oh? how come? (re: manually installing)
[02:17] <veebers> yeah sure
[02:18] <thomi> veebers: because otherwise it doesn't install 1.5
[02:18] <veebers> fyi: http://q-jenkins.ubuntu-ci:8080/job/autopilot-release-gatekeeper/131/
[02:18] <thomi> because autopilot-touch is already in the seed
[02:19] <thomi> so the current installed python3-autopilot satisfies the dependency
[02:19] <veebers> thomi: ugh, click. damn, you meant in the params for the job
[02:20] <veebers> this is the actual one: http://q-jenkins.ubuntu-ci:8080/job/autopilot-release-gatekeeper/132/
[02:20] <thomi> hmm? yeah
[02:20] <thomi> sorry :)
[02:24] <veebers> nw, killed and restarted that job. Luckily I clicked now and not after the run :-)
[02:38] <veebers> ugh, thomi this is the 2nd time I've seen this failure (but I had a successful run after the first time I saw it): http://q-jenkins.ubuntu-ci:8080/job/autopilot-release-gatekeeper/132/label=daily-mako/console
[02:38] <veebers> namely the error; I: Unlock failed, script output: 'initctl: unable to determine sessions
[02:38] <thomi> veebers: kick it off again I guess, and maybe file a bug with the infrastructure people
[02:38] <veebers> (I pinged the CI guys about this earlier but haven't heard back, I fear they're all EOD)
[02:38] <veebers> thomi: aye, that's teh plan :-)
[02:38] <veebers> the*
[02:40] <veebers> hmm, perhaps I borked it. I forgot to add the ppa (after all that ruse to add python3-autopilot to it)
[02:40] <thomi> veebers: may as well change the job defaults for the python3-autopilot thing
[02:40] <thomi> no need to install python-autopilot any more, so just change it
[02:41] <veebers> thomi: ok, that shouldn't effect other people that may use that job? (namely Saviq which I understand does sometimes)
[02:41] <thomi> shouldn't
[02:41] <thomi> if it does, they can get their own :P
[02:42] <veebers> heh, aight, changing now
[04:08] <thomi> veebers: any luck with the latest run?
[04:08] <veebers> thomi: good question, I'll check now
[04:08] <veebers> thomi: no, now there are 3 failures: http://q-jenkins.ubuntu-ci:8080/job/autopilot-release-gatekeeper/133/testReport/
[04:09] <thomi> veebers: can you investigate please?
[04:10] <veebers> thomi: I'll just purge the ap testing ppa I added and add the silo. I'll bother you and pick your brain if it appears to be something with the new xpathselect stuff :-)
[04:10] <thomi> what new xpathselect stuff?
[04:10] <thomi> oh
[04:10] <thomi> you mean in autopilot
[04:10] <veebers> thomi: sorry, yes
[04:11] <thomi> hang on, I think I can see the roblem already
[04:11] <thomi> look at the error in: http://q-jenkins.ubuntu-ci:8080/job/autopilot-release-gatekeeper/133/label=daily-mako/testReport/filemanager.tests.test_filemanager/TestFolderListPage/test_cancel_rename_file_with_touch_/
[04:11] <thomi> veebers: ^
[04:11]  * veebers looks
[04:12] <veebers> thomi: you'll have to elaborate
[04:12] <thomi> see anything odd there?
[04:12] <thomi> read the error message
[04:13] <veebers> this? name '<class 'ubuntuuitoolkit._custom_proxy_objects.popups.ActionSelectionPopover'>'
[04:13] <thomi> right
[04:13] <thomi> WTF
[04:13] <thomi> that's not a class name
[04:13] <veebers> no, it's not
[04:14] <thomi> but I can't seem to reproduce that in ipython
[04:14] <thomi> I know python 3 does something with fully qualified names
[04:15] <veebers> odd, I'll see if I can step through and see better what's happening.
[04:15] <thomi> veebers: but I'd say that's what the issue is
[04:15] <veebers> It's really odd that there are now 3 failing but 2 befpre
[04:15] <veebers> before*
[04:17] <thomi> odd that the ActionSelectionPopover selection doesn't appear int he log
[04:17] <thomi> veebers: maybe run it with -vv, so we can see the debug log
[04:17] <thomi> but yeah, the class isn't being stringified properly I'll bet
[04:17] <veebers> thomi: hmm, i don't think I can do that with the gatekeeper job, but I'll try on my locally flashed device
[04:18] <thomi> veebers: right, I meant locally
[04:18] <veebers> coolio, on it
[04:18] <thomi> I'm happy to look at this with you, if you want to share your screen in a G+ hangout
[04:18] <veebers> thomi: let me get the device ready and see if I can reproduce it, otherwise sounds good
[04:57] <veebers> thomi: fyi running the filemanager tests locally on my desktop (on Trusty) all tests pass, still running on the device
[04:57] <thomi> you're running v1.5 & py3 on the desktop?
[04:58] <veebers> thomi: yes, just trying to get a 'reproducable' quicker
[04:58] <thomi> sure
[04:58] <thomi> hmmm
[04:59] <veebers> thomi: phone just finished, I get the failure, seeing if I can narrow it down to running just one test
[05:00] <thomi> veebers: crap, I just realised that my instinct was wrong
[05:01] <thomi> veebers: the fully qualified name is a red herring I'm afraid
[05:01] <veebers> thomi: hmm, ok :-\ I'll keep poking
[05:02] <thomi> but I'll fix that log message anyway
[05:02] <veebers> coolio
[05:07] <thomi> veebers: for your consideration, when you get the time: https://code.launchpad.net/~thomir/autopilot/temp-dev-fix-class-name/+merge/219286
[05:08] <thomi> Personally, I think it should go in the release - happy to leave that decision to you though
[05:08] <veebers> thomi: cool,I should be able to look at that tonight
[05:09] <veebers> thomi: that code looks familiar, didn't I propose a MP that did this the other day?
[05:10]  * veebers is confused
[05:10] <thomi> veebers: that was for a different case
[05:10] <thomi> veebers: that was for a log message - this is for SNFE
[05:10] <thomi> veebers: but year, same thinko
[05:10] <thomi> I'm surprised we didn't catch this case when you filed your original MP, but oh well
[05:10] <veebers> thomi: ah ok, no longer confused
[05:11] <thomi> glad to hear it :)
[05:11] <veebers> well, perhaps 'less confused' is a more accurate statement
[05:13] <veebers> thomi: hmm, just running the failing test now passes for me on the device >:-| I'll poke around and see if I can narrow it down to a single failing or at least a combination that fails
[05:13] <thomi> hmmm
[05:13] <thomi> are you running with -vv?
[05:13] <veebers> thomi: I am now, I modified phablet-test-run to use -vv
[05:14] <thomi> would be really interesting to see the queries that are being run
[05:14] <thomi> in both cases
[05:14] <veebers> thomi: yeah, I'm running this 1.5 now, then I'll purge and run again with 1.4 + -vv and compare against the failing etc.
[05:19] <veebers> thomi: ugh awesome, this time around I get 1 test fail with SNFE  1 x ValueError error & 1 x AssertionError failure :-\
[05:19] <veebers> so it's inconsistent
[05:20] <thomi> veebers: pastebin?
[05:21] <veebers> thomi: http://paste.ubuntu.com/7455887/
[05:21] <thomi> ugh - we *really* need to fix that logging bug
[05:22] <veebers> heh. Gotta love the "Key pressed!" log message :-)
[05:24] <thomi> heh, and the "<skip>Can't do this properly on desktop or phablet</skip>"
[05:24] <veebers> heh :-P
[05:25] <veebers> thomi: how come I only see one instance of "'GetState Query(b'/comubuntufilemanager/QQuickView/MainView[id=2]//ActionSelectionPopover[objectName="fileActionsPopover"]')" before exception raised?
[05:25] <veebers> I would have assumed 10 lines in the log then eht exception
[05:25] <thomi> veebers: can you 'bzr blame' the source and see who's been touching the source?
[05:25] <thomi> veebers: PB line #?
[05:26] <thomi> veebers: it's just a select_single not a wait_select_single
[05:26] <thomi> oh wait
[05:26] <thomi> yeah
[05:26] <thomi> what I said
[05:27] <veebers> thomi: sorry I must be being slow, bzr blame which source?
[05:27] <thomi> the test source
[05:27] <veebers> ah good point
[05:27] <veebers> can do
[05:27] <thomi> You could try making it a wait_select_single and see if that improves things
[05:28] <veebers> thomi: hmm, I wonder if this version of AP is faster which is why 1.4 works currently
[05:28] <thomi> could be, although if it is, it won't be by much.
[05:28] <thomi> I haven't done any explicit optimisations
[05:29] <veebers> thomi: hmm ok. Hey, is there a nice why to 'blame' a whole directory?
[05:29] <thomi> don't think so - that doesn't really make sense
[05:30] <thomi> what would it show?
[05:30] <thomi> mode changes to the directory maybe, but I suspect that's not what you want :)
[05:30] <veebers> *shrug* split-pane with a tree of files and clicking them shows the diff/blame for that file
[05:30] <thomi> ahh, gotchya
[05:31] <thomi> bzr log does that then :)
[05:31] <thomi> even better - it does it with the author as the first index :)
[05:31] <thomi> bzr qlog tests/autopilot/
[05:31] <thomi> or whatever
[05:31] <veebers> thomi: ah, that's what I'm doing at the moment :-)
[05:34] <veebers> thomi: so might be of interest, looking at the logs balloons recently made changes to the tests, one being this http://pastebin.ubuntu.com/7455926/ (changing wait_select_single -> get_action_selection_popover)
[05:35] <veebers> where get_action_selection_popover is in the ubuntu ui toolkit ap and uses select_single
[05:35] <thomi> hmmm
[05:36] <thomi> what does that method use?
[05:36] <veebers> I'll try patching that to use wait_select_single and see
[05:36] <veebers> thomi: you mean get_action_selection_popover?
[05:37] <thomi> yeah
[05:37] <veebers> thomi: [17:35:12] <veebers> where get_action_selection_popover is in the ubuntu ui toolkit ap and uses select_single
[05:38] <veebers> I'm just about to re-run the test on the device with the uuitk patched to use wait_select_single
[05:48] <veebers> thomi: hmm, Without seeing the logs (still running) it seems that I saw a failure run past where the devices screen was blank white. It would be interesting to see the phones state (cpu usage etc.) during the test
[05:53] <thomi> veebers: I'm going to EOD now
[05:54] <thomi> veebers: can you please email whoever looks after that app with a link to the failures, our investigation, and ask them to look into it as a matter of urgency please?
[05:54] <veebers> thomi: ack. sure, can do
[05:54] <thomi> we *really* need to release AP tomorrow, which means whoever owns these tests needs to either allow us to release, or get them fixed :)
[05:54] <thomi> feel free to CC me
[05:54] <thomi> thanks man
[05:54] <thomi> talk to you later maybe.
[05:54] <veebers> thomi: If I can't find the dev for filemanager I may have to email the phone list
[05:55] <veebers> thomi: aye, night o./
[05:55] <thomi> veebers: why didn't bzr blame work?
[05:55] <thomi> I thought you identified balloons?
[05:55] <veebers> thomi: right, but should it be on his shoulders? I don't know. I can start with him and he can hopefully fwd if needed
[05:56] <thomi> veebers: right, but you can get a complete list of authors with bzr blame, right?
[05:56] <thomi> I don't understand why you'd need to use the ML
[05:56] <veebers> thomi: I'll also add Chrs/robotfuel  as it looks like both of them have looked at the flakiness of tests latetly
[05:56] <thomi> awesome
[05:56] <veebers> thomi: leave it to me I'll sort it out
[05:56] <thomi> schweet
[05:56] <thomi> going to have a beer and dinner
[05:57] <thomi> laters!@
[05:57] <robotfuel> veebers: what's flaky?
[05:57] <veebers> robotfuel: isn't is super late for you? Perhaps you're clock is flaky ;-)
[05:58] <veebers> robotfuel: thomi and I are investigating some failures in the filemanager autopilot tests when using the to-be-released autopilot 1.5
[06:00] <robotfuel> yes, I have some ideas. I was too busy with other stuff to fix it today, I am going to sleep so I can fix it in the morning.
[06:00] <robotfuel> veebers: can you email me instructions for using ap 1.5?
[06:00] <robotfuel> like which ppa I use?
[06:00] <veebers> robotfuel: awesome, ppa is: ci-train-ppa-service/landing-003
[06:00] <veebers> robotfuel: should I email you and balloons re: pretty much what we've discussed here?
[06:01] <robotfuel> sure
[06:01] <veebers> robotfuel: I'll link to the gatekeeper job that we've been using to run it too
[06:02] <robotfuel> veebers: thanks again, good night
[14:24] <tscheck> Hi, is there are reason, why the qt and gtk modules in autopilot/introspection have no api doc pages?
[14:24] <tscheck> See: http://unity.ubuntu.com/autopilot/_modules/autopilot/introspection/
[14:25] <tscheck> When you use the Quick search of the online help, you can't find any info on signals.
[14:27] <tscheck> Hm. There is another online help without qt and gtk modules at: http://developer.ubuntu.com/api/devel/ubuntu-14.04/autopilot/_modules/autopilot/introspection/
[14:27] <tscheck> Strange.
[14:27] <tscheck> Shall I file a bug?
[14:45] <balloons> tscheck, those are strange links
[14:46] <balloons> I don't believe the pages you found are meant to be published atm.. meaning, http://unity.ubuntu.com/autopilot/_modules/autopilot/introspection/gtk.html for instance
[14:46] <balloons> looks quite old
[15:12] <tscheck> balloons: I see. But the first link are the official api docs, right? Any info about why the qt and gtk modules are not created?
[15:14] <balloons> tscheck, not offhand. some devs will be around in about 5 hours and they should have an answer for you
[15:26] <balloons> elopio, did you ever make an optionselector emulator?
[15:38] <elopio> balloons: robotfuel did. It's already published.
[15:39] <balloons> elopio, I remember it, I just don't see it
[15:39] <balloons> and I need it now :-)
[15:40] <elopio> balloons: http://bazaar.launchpad.net/~ubuntu-sdk-team/ubuntu-ui-toolkit/trunk/view/head:/tests/autopilot/ubuntuuitoolkit/_custom_proxy_objects/_optionselector.py
[15:40] <elopio> ubuntuuitoolkit.OptionSelector, that's the new namespace.
[15:47] <balloons> elopio, is trusty missing this?
[15:47] <elopio> balloons: I don't know.
[15:47] <elopio> that's a question for bzoltan.
[15:47] <balloons> seems to be the case..
[15:48] <balloons> ok, I finally found what you are talking about and I'll give it a try now. Looks nice
[15:48] <balloons> but yea, the package in trusty doesn't have it afaict
[16:14] <elopio> balloons: we should probably backport it.
[16:15] <elopio> it's not going to be hard to have the autopilot package of the toolkit working on trusty and on utopic
[16:21] <balloons> elopio, turns out, problem solved. zoltan said sdk team pushes updates to there ppa. ppa:ubuntu-sdk-team/ppa. I've asked francis to add that ppa to the jenkins builders so we should be covered
[16:22] <elopio> oh, well, maybe a little hard because they add features like the new header
[16:22] <elopio> balloons: nice, thanks.
[20:19] <thomi> barry: did you get a chance to look at my packaging stuff?
[20:20] <barry> thomi: not yet.  will soon
[20:20] <thomi> cool - thanks :)
[20:21] <veebers> morning
[22:17] <thomi> nuclearbob: cgoldberg: quick thing:
[22:18] <thomi> we said in a meeting that you guys were going to attend the qa/ci meeting and talk about something, was it the subunit integration work?
[22:18] <cgoldberg> thomi, yea
[22:18] <thomi> in any case, we need to talk about that between ourselves first, so you have all the facts :)
[22:18] <cgoldberg> thomi, can you bring it up tomorrow in the call?  im about to EOD
[22:18] <thomi> Do you think we can do that in the TnT standup tomorrow, or should we make a separate call?
[22:19] <thomi> that answers that question :)
[22:19] <cgoldberg> :) night