=== nudtrobert1 is now known as nudtrobert === maclin1 is now known as maclin [09:04] pitti, can we force adt-run to install a specific package in the system path rather than the tmp location? [09:04] pitti, we hit a new issue with confined apps and libqttestability [09:05] brendand: if the root fs is r/o, we can't install packages there [09:05] brendand: you could of course make the test check if it's running on a phone, then remount-rw, call apt-get install, remount-ro [09:05] but -- eww :) [09:06] pitti, oh yeah as a --setup-command [09:06] or that (if you have control over that) [09:06] pitti, it is ew. but the only other thing i could think to fix it would be to hack the apparmor profile to allow access to the /tmp/ path [09:07] pitti, if it's possible then it's still much more ewww than the former [09:07] pitti, basically the problem is the app can't load the testability library as it's in /tmp [09:07] brendand: OOI, what is the actual problem? the test needs libqttestability but that isn't already installed in the image? [09:08] ah [09:08] pitti, like: http://paste.ubuntu.com/12797190/ [09:14] brendand: perhaps it'd be better to install that into the images then? don't we need this for pretty much every test? is it very big? [09:14] brendand: or we could change the adb setup script to munge the apparmor rules further; it already does that, after all [09:14] brendand: perhaps that'd be the best solution for now? [09:20] pitti, oh it does? that would be ideal then. i guess it would be safe to add that very specific path [09:21] pitti, i should have known that, i've spent enough time waiting for apparmor profiles to update :) [09:24] brendand: http://anonscm.debian.org/cgit/autopkgtest/autopkgtest.git/tree/lib/adt_testbed.py#n694 [09:25] brendand: not sure if that's the correct spot, though [09:25] pitti, no it's nothing to do with clicks i don't think [09:27] pitti, do you have time to do a patch or would you prefer to leave it to us? it's pretty critical for us but might take a little bit longer to work it out [09:29] brendand: if you can work on it, that'd be good; I have too many "omgnow" issues to work on right now [09:30] pitti, i suppose http://anonscm.debian.org/cgit/autopkgtest/autopkgtest.git/tree/lib/adt_testbed.py#n716 is somewhat relevant [09:30] pitti, i should just do the same thing for LD_LIBRARY_PATH [09:30] ? [09:30] would be good to know i'm at least on the right track [09:31] brendand: yes, that's right [09:32] brendand: note that if you don't install a click, the "else" branch hits and allows introspection for all clicks [09:32] brendand: that's for the case when you test a deb source, and adt-run can't know in advance which clicks you want to talk to [09:32] brendand: so either way, def apparmor_click() should be relevant for you [09:33] brendand: feel free to add some adtlog.debug() calls (and run with -d) === qwebirc512034 is now known as slickymasterWork [12:33] pitti, i just did a patch but it appears not to be working... [12:34] pitti, i'm sure the path to the .so file was added though [12:39] brendand: you should see that in the debug output; and you could just run a mini-test which only does "env" to verify it all the way through? [12:40] ah wait, apparmor -- so run the test and run "dmesg" after it or so? [12:49] pitti, the rules update call looks like: http://paste.ubuntu.com/12798413/ [13:06] pitti, do you know how to check the apparmor rules, is there a tool for that? [13:06] brendand: I don't know further than just setting them and checking violations in dmesg [13:07] you can put the profile into complain mode to just get the logged violations but don't actually block them [13:07] but as you probably just need one rule that sholdn't matter much [13:07] pitti, i can see '/tmp/adt-run.zBvagg/deps/usr/lib/** r' in the set of rules clearly, but still get the violation [13:46] nuclearbob, hey! Can you tell me what output we get from the upgrade tests for the desktop? [13:50] nuclearbob, just curious for an update on how the flavors ubiquity AP tests are running also ;-) [13:51] willcooke: http://iso.qa.ubuntu.com/qatracker/milestones/340/builds === qwebirc719331 is now known as slickymasterWork [13:54] willcooke: I'm working on defining more of that as we overhaul those and move them to adt, actually. If you've got requests, it's a great time to figure that out. I'll be out for a medical appointment for a bit, so you can email me, or I'll ping you when I get back [13:55] balloons: I've been focused more on getting the auto upgrade tests into adt for the moment. The ubiquity tests are still all failing on my system, so I'll need to diagnose that. The results should be published to the canonistack instance if anyone else wants to take a look. If someone can say "hey, change this line and ubuntu passes" then adding flavors will be the next easy step [13:56] nuclearbob, thx. In a meeting with davmor2 and brendand atm, so perhaps a catch up next week? [13:57] willcooke: sounds good. That's the main thing I'm working on right now, so if you have any thoughts about what you'd like to see, send them over any time [13:58] nuclearbob, will do, thanks [13:59] nuclearbob, ack. If you don't mind sending that along to the mailing list again, I think it would be useful. Seeing the results; even if broken, would be useful. And we can get more eyes on what might be wrong [14:19] willcooke: https://bugs.launchpad.net/bugs/1499873 [14:19] Error: ubuntu bug 1499873 not found === willcooke_ is now known as willcooke [20:46] balloons: I'm checking on the job publishing, and it hasn't actually worked yet, so I'll try to get that backlog pushed correctly so other people can see the results. I'm running it now, and if there are further problems, I can look at them on monday [21:08] * flocculant has nuclearbob as a ping :) [21:08] just to see if flavours gets mentioned [21:08] :D [21:08] and I bet balloons has gone awol again