[06:44] <pitti> elopio, dobey: there is no need to pass -v to autopilot tests, that's the default now; it also produces a subunit stream in the --output-dir
[08:08] <pitti> fgimenez: also, I think we should discuss that in the public channel
[08:09] <pitti> fgimenez: so, http://paste.ubuntu.com/10070375/ is the output of autopkgtest
[08:09] <pitti> fgimenez: so you didn't call adt-run yourself, but it is called through some wrapper?
[08:09] <fgimenez> pitti: that's right, python3 -m ubuntu_sanity_tests.command_line from the trunk of ubuntu-sanity-tests
[08:10] <pitti> wow, wrappers around adt-run :)
[08:10] <fgimenez> pitti: with this diff applied http://paste.ubuntu.com/10068247/
[08:10] <pitti> so that one does --setup-commands 'mount -o remount,rw /; apt-get update; mount -o remount,ro /'
[08:10] <pitti> fgimenez: I suppose the second mount (for ro) fails then
[08:11] <pitti> fgimenez: not sure why, perhaps an additional 'sync' or 'sleep 5' or so works there? this needs checking for some running processes (lsof perhaps) what keeps / busy
[08:12] <pitti> (I just ran adt-run directly, without the apt-get update)
[08:14] <fgimenez> pitti: ok, now trying sync after apt-get update and before the ro mount in the wrapper
[09:55] <fgimenez> pitti: yes, the sync solves the busy / issue, but it is still unable to install the upgrade, as you know http://paste.ubuntu.com/10088869/
[10:55] <pitti> fgimenez: yeah, I figure that's the issue that elopio brought up on the ML
[10:55] <pitti> fgimenez: probably better to entirely remove that --setup-commands (see my reply)
[10:58] <jibel> fgimenez, I replied to elopio's email too. For sanity tests the image must not be modified at all, it would make the test invalid.
[11:07] <fgimenez> pitti: ok thanks
[11:33] <rhuddie> pitti, hey. I seem to have a problem with adt-run
[11:33] <rhuddie> pitti, it seems that when I run tests for a click, not all the required dependencies are getting installed
[11:34] <rhuddie> running tests for the camera app with: adt-run --click com.ubuntu.camera --- ssh -s adb
[11:34] <rhuddie> I get: ImportError: No module named 'wand'
[11:36] <rhuddie> python3-wand and python3-mediainfodll are both defined in the debian/control file for camera-app-autopilot
[11:36] <rhuddie> so when I install tests directly on device and run, it works fine
[11:37] <rhuddie> but I get the import error when I try and use adt-run
[11:37] <rhuddie> (reason for this is that the dashboard is reporting this failure)
[11:38] <pitti> rhuddie: could you please pastebin the complete command and log?
[11:38] <pitti> there might be some information there
[11:39] <pitti> rhuddie: you might just as well open a bug with the log, as we have a bug *somewhere* :)
[11:39] <rhuddie> pitti: http://pastebin.ubuntu.com/10090310/
[11:39] <rhuddie> pitti, will do, thanks
[11:40] <pitti> rhuddie: indeed, no python3-wand installed
[11:40] <pitti> rhuddie: I suppose the reason is that there is no camera-app-autopilot test dependency
[11:40] <pitti> rhuddie: so it looks like its manifest is simply missing that?
[11:41] <pitti> rhuddie: note that you can now use --override-control /path/to/fixed/manifest if you want to do a test run with a fixed manifest
[11:41] <rhuddie> pitti, oh thanks. let me give that a try
[11:44] <pitti> jibel: hm, http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#gcc-4.9 has a bunch of "in progress" tests which finished long ago; state files getting out of sync again?
[11:44] <pitti> jibel: how can this be fixed on the britney host?
[12:24] <rhuddie> pitti, my connection just dropped, i might have missed your reply
[12:24] <pitti> rhuddie: I didn't write anything further to you after your "let me give that a try" (and didn't see a further response)
[12:25] <rhuddie> pitti, oh strange!
[12:25] <rhuddie> pitti, I was looking at some other click manifests, but can't see any references to the autopilot dependencies in those http://pastebin.ubuntu.com/10090759/
[12:25] <rhuddie> am I looking in the right place?
[12:28] <pitti> rhuddie: yeah, those don't have any additional test deps
[12:29] <pitti> rhuddie: you can just add a "depends" field like in https://people.debian.org/~mpitt/autopkgtest/README.click-tests.html
[12:29] <rhuddie> pitti, right. thanks for clarification.
[13:29] <elopio> good morning.
[13:50] <rhuddie> pitti, unfortunately I am seeing another problem similar to the one I saw with imagemagick previously: http://pastebin.ubuntu.com/10091944/
[13:51] <rhuddie> but this time: ImportError: MagickWand shared library not found
[13:51] <rhuddie> I tried adding this directly to the list of dependencies in the manifest, but it didn't make a difference
[13:52] <rhuddie> I have autopkgtest 3.9.4
[13:59] <pitti> rhuddie: ah, that looks like something in the test clobbers LD_LIBRARY_PATH, but a bit hard to say just from that log
[14:00] <pitti> rhuddie: so, bug report with -d (debug) output appreciated; I'll look into it, maybe it's possible to hack around this
[14:01] <rhuddie> pitti, yes. I'll put all the details into a bug. thank you :)
[14:01] <pitti> and meh -- folks, don't expect a gazillion *apt* packages for *click* package tests!
[14:01] <pitti> the unpack/env var trick only works that far..
[14:23] <pitti> jibel: I removed all the *.state files now and re-ran the tests; hope that helps
[15:00] <dobey> pitti: where does the output stuff go when using adt-run with qemu? afaict, the VM gets destroyed at the end, and all the data is gone with it
[15:23] <brendand> elopio, from your uninstall_click branch? http://paste.ubuntu.com/10093282/
[15:23] <elopio> brendand: that's because you are on utopic.
[15:23] <elopio> brendand: fixed here: https://code.launchpad.net/~canonical-platform-qa/ubuntu-sanity-tests/remove_tests_import/+merge/248817
[15:24] <brendand> elopio, ok top-approved that one
[15:32] <brendand> elopio, i have this feeling that a lot of our test code is a bit messy and inconsistent
[15:32] <brendand> elopio, like different tests doing different things in the setUp compared to the test itself
[15:32] <brendand> elopio, lots of duplicated code actually
[15:33] <brendand> elopio, i think we might take some time to refactor it in the next sprint if you think it's worthwhile
[15:34] <brendand> elopio, normally i'm not a fan of refactoring code for the sake of it, but i think we might start causing ourselves problems if we don't address it now
[15:36] <elopio> brendand: sounds ok. Are you talking about the self tests or the autopilot tests?
[15:37] <brendand> elopio, no the selftests are fine
[15:37] <brendand> elopio, the sanity tests - i.e. the main purpose of the project :) are looking a little slapdash
[15:38] <brendand> elopio, for example there are a number of steps which any test that launches an app will do
[15:38] <elopio> brendand: ok. How would you like to do it? report a bug for each issue that you see?
[15:38] <brendand> elopio, there should be a class derived from BaseTestCase to do those steps in the setup
[15:38] <brendand> elopio, yeah i can raise a bug for each issue
[15:39] <brendand> elopio, something for next week at this stage
[15:40] <elopio> brendand: ok. Take into account the argument that veebers and thomi always make of not hiding steps in the set up.
[15:40] <elopio> but I suppose we can improve things with fixtures or just common nicely named methods.
[15:44] <brendand> elopio, yep
[16:46] <brendand> elopio, did you push the branch mentioned in https://code.launchpad.net/~canonical-platform-qa/ubuntu-sanity-tests/remove_update/+merge/248853
[16:46] <elopio> brendand: not yet. I was looking at the bq phones.
[16:46] <brendand> elopio, :)
[16:46] <elopio> I want one white with headphones.
[16:47] <elopio> brendand: sorry, I'll do it now.
[16:47] <brendand> elopio, you've already got one :)
[16:51] <elopio> brendand: yes, but it's black. I have three phones for testing now, one for each hand and one on the ear. I need another to test with the other ear.
[16:51] <brendand> elopio, i know, phones everywhere
[17:36] <elopio> brendand: does veebers approval counts for the timeout branch? should I top approve?
[18:23] <elfy> balloons: ping
[20:10] <balloons> elfy, pong
[20:18] <elfy> often