[00:06] Oh cool! I can use Virtualbox to test 14.10? [00:08] bbl [01:07] Hi again, did I miss anything? [01:10] Guess not :\ [01:18] Nothing_Much, yes you can use vbox [01:19] Nothing_Much, an install you will use is best. Be it vm or real hardware. Real hardware is preferred but vm works too [01:21] Awesome === chihchun_afk is now known as chihchun [04:58] Good morning [07:50] pitti, I'm trying to produce a stacktrace for https://errors.ubuntu.com/oops/9531bd08-10a5-11e4-8591-fa163e78b027 [07:50] pitti, I unpacked the crash file and ran: gdb --batch --ex 'file /usr/lib/arm-linux-gnueabihf/indicator-location/indicator-location-service' --ex 'core-file CoreDump' --ex 'bt' [07:50] jibel: you run this on the phone? [07:50] pitti, which fails with a corrupt stack. Is it correct [07:50] ? [07:50] pitti, on the phone [07:51] jibel: it looks correct; to be sure, you can re-run it interactively [07:51] gdb /usr/lib/arm-linux-gnueabihf/indicator-location/indicator-location-service CoreDump [07:51] without ddebs I'll have unresolved symbols but not a corrupt stack [07:51] and then "bt full"; that might show more warnings/errors [07:52] but you should still see the gdb errors/warnings with your command as well; do you get any? [07:53] pitti, same result http://paste.ubuntu.com/7829037/ [07:54] warning: core file may not match specified executable file. [07:54] jibel: so, might be image mismatch; but still, same old gdb problem otherwise :( [07:54] ah, the phone has been upgraded [07:55] let me crash it again :) [08:05] ooh, too bad, it crashed :( [08:07] jibel: isn't that what you wanted? [08:07] pitti, it is [08:07] it's just too easy [08:08] and now system-settings uses 100% CPU but is not running in foreground [08:49] pitti, so same old problem with a fresh crash http://paste.ubuntu.com/7829204/ [08:51] jibel: yeah, and no bin mismatches now :/ === Ursinha-afk is now known as Ursinha [12:11] pitti, did you see the failure in umockdev http://paste.ubuntu.com/7829954/ [12:11] http://d-jenkins.ubuntu-ci:8080/job/utopic-adt-umockdev/30/ARCH=amd64,label=adt/console === _salem is now known as salem_ [12:35] jibel: just back from lunch, I'm looking at the recent failures [12:36] pitti, they are all the same [12:37] jibel: whoa -- fixing now [12:37] curious that this didn't come up in my testing, when I rolled out the new autopkgtest libpng etc. succeeded [12:37] and none of the tests caught it either *sigh* [12:40] jibel: ah, it affects --apt-source with "build-needed" only; sorry about that [12:46] jibel: fixed in git, rolled out, umockdev/tgt/colord retried [12:46] jibel: systemd is an "honest" failure, I'll look at this [12:47] pitti, thanks! [12:48] ah, sd_login_monitor_new() would fail in a chroot, indeed [12:48] it worked locally as I have an actual session running, I'll rewrite this === jpds_ is now known as jpds [14:40] balloons: so after "Tests running... [14:40] balloons: ... in http://paste.ubuntu.com/7815272/ nothing further happens? [14:42] pitti, yes exactly [14:42] it's quite bizarre. I thought it was perhaps just me but both jenkins and jibel report the same thing [14:43] * pitti runs here [14:45] ohh really? [14:45] what version of the image? [14:45] balloons: no, I mean "I" am running this now and try to reproduce [14:45] not "it" runs here [14:47] pitti, lol.. whew.. sorry reading into what you said :-) [14:47] balloons: right, I see the same here; "python3 -m autopilot.run run reminders" is running [14:47] but nothing visible [14:52] balloons: FYI, with autopkgtest 3.2 you can use --override-control to point to a local manifest (to avoid rebuilding the click) [14:53] pitti, ahh yes, thanks for the reminder. I actually built it a few ways so I just pass the clicks I built before [14:53] balloons: sure, just pointing out as you were asking about it [14:54] yep, glad it's possible.. much nicer than rebuilding a big armhf click [14:56] balloons: curious, very curious; that's my current output with -v: http://paste.ubuntu.com/7830651/ [14:56] balloons: (still running and progressing) [14:56] why don't I see anything?? [14:57] pitti, you overrode AP to give you back verbose output eh? [14:57] balloons: yes; put that in the manifest into the "module" value, or use the undocumented cheesy ADT_AUTOPILOT_MODULE="-v reminders" [14:57] so, my thought with why this was happening is because of how the tests work. They create a fake online account via python before running the tests [14:58] i'll have to look at the tests again, but there is a small wait in there. [14:59] The ubuntuuitoolkit.emulators module is deprecated. Import the autopilot helpers from the top-level ubuntuuitoolkit module. [14:59] balloons: coudl that have something to do with it? [15:00] pitti, nope, that's just a warning from the toolkit.. I have a bug to convert everything, but it has no affect on the helpers atm [15:01] balloons: ok; also, I notice that reminder's .desktop is broken [15:01] that sounds more of an issue [15:01] * balloons peeks at reminders tests [15:02] Exec=reminders [15:02] that's what ubuntu-app-launch would try to run, no? [15:05] meh, adb is such a terrible terminal [15:05] pitti, phablet-shell if you are using it [15:06] pitti, I'm not sure actually how much of a role the .desktop file plays [15:06] balloons: yeah, I think ubuntu-app-launch is doing some black magic [15:06] .cache/ubuntu-app-launch/desktop/com.ubuntu.reminders_reminders_0.5.184.desktop [15:07] it creates that, but it doesn't actually seem to use any information from that for running an app [15:07] maybe it searches the click dir for and directly runs that or os [15:07] so [15:09] brendand, elopio I see my mp for the sdk is still pending. https://code.launchpad.net/~nskaggs/ubuntu-ui-toolkit/fix-1328600/+merge/227394. What do you think of it? [15:11] balloons: looks good, but you still have to update one test. [15:11] thanks for this, you will make the team happy. [15:12] elopio, sure what test did I miss? I see one failure, I guess it's from me eh? [15:12] balloons: ubuntuuitoolkit.tests.custom_proxy_objects.test_date_picker.PickTimeFromDatePickerTestCase.test_pick_time(lower second) [15:12] testtools.matchers._impl.MismatchError: datetime.time(12, 39, 20) != datetime.time(12, 30, 20) [15:12] elopio, ahh ok, ty.. I'll look and fix [15:13] balloons: I did some further comments to the MP, but I'm afraid I'm running out of ideas here :/ [15:18] balloons: OOI, does that work with phablet-test-run? (reflashing my device ATM, can try later) [15:19] pitti, yes it should run with phablet-test-run.., Just make sure you push everything so it has it's depends [15:19] balloons: ah, you mean apt-get installing its test deps? [15:20] pitti, no. phablet-test-run will look inside /home/phablet/autopilot and load all the python modules in there [15:20] so just push them to that folder [15:20] balloons: does phablet-test-prepare-something do that, or did you do that manually? [15:20] so something like python3-dbus is sitting in there and can be used by any test [15:21] pitti, manually [15:21] balloons: apt-get download and dpkg-deb -x? [15:21] pitti, I usually just adb push from my own installation because I'm lazy, but yes :-) [15:21] (how does CI run these?) [15:22] very rarely will you find a compiled python module, but I have encountered them [15:22] so apt-get download (armhf) is the safer option [15:22] pitti, CI doesn't run reminders (and can't without adt for this reason) [15:22] ah [15:23] and for calendar, the phablet-setup stuff pulls it's dateutil depends and unpacks in /home/phablet/autopilot [15:25] in other words mock, dateutil and the toolkit are the "default" depends pulled by phablet-click-test-setup [15:34] pitti, while i still have you, is notes working for you now? https://code.launchpad.net/~nskaggs/notes-app/ap-setup-update/+merge/225992 [15:35] thanks for the notes on reminders. I'll give it a go again this afternoon. The extra debugging info should help [15:35] balloons: still fighting with phablet-test-run [15:35] (for reminders) [15:35] can look at notes tomorrow morning [15:36] pitti, ohh.. I saw the note about reminders not running via phablet-test-run. I haven't run it on the latest images but no reason to believe it's broken [15:39] ImportError: No module named 'requests' [15:39] balloons: I get that with ./phablet-test-run reminders [15:39] balloons: missing dep? [15:39] pitti, indeed [15:39] pitti, push everything in debian/control for the -autopilot package [15:40] I did [15:40] I have ~/autopilot/requests_oauthlib [15:40] python3 verison? [15:40] yes [15:40] python3-oauthlib_0.6.1-1_all.deb and python3-requests-oauthlib_0.4.0-1ubuntu1_all.deb [15:40] * balloons looks in his autopilot folder [15:41] File "/home/phablet/autopilot/requests_oauthlib/oauth1_session.py", line 10, in [15:41] import requests [15:41] bah, it's not even running for me [15:41] balloons: so if you could do "find -name 'requests*' [15:42] pitti, I have requests and requests_oauthlib [15:42] where does requests come from? [15:42] requests_oauthlib [15:42] adb push /usr/lib/python3/dist-packages/requests/ /home/phablet/autopilot/requests [15:43] ah, python3-requests [15:43] :-) [15:43] ImportError: cannot import name 'certs' [15:43] well, that's progress I suppose :) [15:44] ok, certs [15:44] whereever that comes from [15:44] * balloons is running reminders tests [15:45] I get your error about TypeError: Argument 1 does not allow None as a value, running the last thing I pushed onto the phone.. I should push trunk [15:46] https://packages.debian.org/search?searchon=contents&keywords=dist-packages%2Fcerts&mode=filename&suite=unstable&arch=any [15:46] there is no "certs" module [15:46] as far as certs.. I don't see the module in my ~/autopilot [15:46] nor is it in reminders [15:47] oh, it's requests/certs.py [15:47] * balloons wonders why find didn't see it [15:48] and there's even a __pycache__/certs.cpython-34.pyc [15:48] WTH [15:48] lol.. I didn't make it * [15:48] yea, I see that now [15:48] aah [15:48] phablet@ubuntu-phablet:~/autopilot$ PYTHONPATH=. python3 -c 'import requests' [15:48] ImportError: No module named 'urllib3' [15:48] python3-urllib3 [15:49] that's what I get with apt-get download, no dependencies.. [15:49] I have that in my notes.. I wonder if we should specify it [15:49] adt-run does install dependencies [15:49] Tests running... [15:50] exactly the same [15:50] pitti, my guess is it's hanging due to the syntax error you found. At the least I should solve that [15:51] balloons: yeah [15:51] balloons: followed up again [15:51] balloons: yep, I see that TypeError now [15:51] ohh wait! I know what that is [15:52] if we see that TypeError, it means the fake account failed to be created [15:52] hence, it's none.. we should catch it and display a better failure [15:52] anyway, if phablet-test-run and adt-run fail in exactly the same way, I suppose it's an SEP for me :) [15:52] still.. the point is, it should be working [15:52] pitti, yes, I think you are in the clear on this one for now [15:53] and now you see why adt is better than the old method [15:53] clearly :) [15:53] balloons: trying notes-app again then [15:53] no one could figure out how to run reminders without it :-) [15:57] balloons: so we should kill phablet-click-test-setup and rewrite phablet-test-run to invoke adt-run with the necessary magic? [15:57] balloons: ah, I suppose I still need to implement automagic source retrieval [16:00] balloons: oh, "badpkg: got 3 lines of results from extract where 5 expected" -> that's most likely you forgetting to specify --click [16:01] balloons: I'll see to improve the error message (it tries to apt-get download com.ubuntu.notes, which fails bitterly) === chihchun is now known as chihchun_afk [16:05] pitti, yes forgetting to specify --click is a bit offputting [16:06] balloons: I filed a bug for it, I'll fix that [16:06] pitti, umm, tweaking how the phablet-tools work I suppose isn't a bad idea [16:06] i. e. give a proper error message [16:06] pitti, any thoughts on the apt-get index issues? [16:06] I think that [16:06] might be the biggest issue atm [16:09] balloons: followed up to https://code.launchpad.net/~nskaggs/notes-app/ap-setup-update/+merge/225992 [16:09] with the phone being read-only, it's not a simple thing to be able to update the indexes [16:09] pitti, ty [16:10] balloons: archive.u.c. keeps old versions for 24 hours I think, so that'll mostly hit on older images [16:10] balloons: one workaround would be to mount r/w, apt-get update, remount r/o [16:11] (in the adb setup scripts) [16:11] another to use a local $APT_ROOT with copying the entire config and running apt-get update as user with temp cache dirs or so [16:11] but yeah, the eternal conflict between "we want to use apt" -- "no, we don't, really" :/ [16:13] balloons: I'm still stuck with ddeb-retriever, but once that's done I can spend more time again on adt-run [16:16] brendand: I replied to your review on the unity8 branch. You are probably running the installed unity. [16:16] elopio, right, probably. i should build and install the one from your branch then [16:17] brendand: you can follow the readme file on the branch [16:17] I think it's called coding [16:17] elopio, bzr bd usually does the trick [16:18] brendand: oh yes, I just don't want to pollute my installed packages. [16:18] but that will work. [16:26] pitti, sounds like you at least have ideas.. I'm guessing avoiding enabling r/w mode is preferred? but perhaps not. Toggeling r/w mode, updating indexes and toggling off is probably the simplest way === Ursinha is now known as Ursinha-afk === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr === Ursinha-afk is now known as Ursinha [19:35] balloons, hi [19:35] Letozaf_, hello [19:36] balloons, howzit ? [19:36] busy busy eh? :-) [19:37] balloons, :) have you got a second for me ? [19:37] of course! [19:37] balloons, I need help on running the clock app test I wrote on device [19:38] balloons, I get stuck after click-buddy --dir . --provision [19:38] Letozaf_, have you tried out using autopkgtest? [19:38] balloons, no [19:38] Letozaf_, give it a try. It should handle all that manual stuff we used to do in the past [19:38] balloons, ok thanks I will === Ursinha is now known as Ursinha-afk [19:43] Letozaf_, give it a whirl now and let me now how it goes.. I can walk you through it if needed [19:44] basically you just build the click and then run the command.. it should all just work [19:44] balloons, ok thanks, I am reading what you wrote on theorangenotebook now :) I will let you know [19:50] * Letozaf_ is updating her device with the latest image [19:52] :-) [19:57] balloons, but do you just run adt-run... in the root directory of the app, don't you need anything else ? [19:58] balloons, well autopkgtest of course is needed :) [20:01] Letozaf_, yes autopkgtest is needed [20:01] you actually run it outside the app directory [20:01] and pass the directory as the first argument [20:02] Letozaf_, you can then choose to build a click package or not [20:02] if you choose to use the click package installed on the phone to run with, you specify it with click=com.ubuntu.clock [20:02] otherwise, the second argument is the click package you built locally [20:06] balloons, ok, so I ran it outside the app directory specifying click=com.ubuntu.clock [20:06] balloons, but I get an error :( I just branched a new clock app just to be sure it's not my code thats wrong [20:07] balloons, : failure: ['/usr/share/autopkgtest/ssh-setup/adb', 'open'] failed (exit status 255) [20:07] adt-run [22:06:00]: ERROR: testbed failed: cannot send to testbed: ['BrokenPipeError: [Errno 32] Broken pipe\n'] [20:07] Letozaf_, can you copy the line you are using to run? [20:07] the syntax is a little funny tricky [20:07] balloons, adt-run ubuntu-clock-app --click=com.ubuntu.clock --- ssh -s /usr/share/autopkgtest/ssh-setup/adb [20:08] balloons, that --- ssh sounds strange to me :P [20:10] Letozaf_, yea, I'd run like this [20:10] adt-run ubuntu-clock-app/ click=com.ubuntu.clock --- ssh -s /usr/share/autopkgtest/ssh-setup/adb [20:10] Letozaf_, if your shell is acting funny you can escape the ssh line with a / in front [20:11] * Letozaf_ is trying [20:12] like \ssh -s /usr/share/autopkgtest/ssh-setup/adb [20:13] balloons, :( looks like it does not like that / after ubuntu-clock-app [20:13] balloons, I get: adt-run: error: click=com.ubuntu.clock: unsupported action argument [20:14] balloons, no even without the / I get the same error [20:15] Letozaf_, so following what I laid out in the post doesn't work for you? [20:16] balloons, let met try again, but up to now it hasn't worked for me === Ursinha-afk is now known as Ursinha [20:18] Letozaf_, I mean, literally can you replicate the example.. I think I used calendar app as the example [20:18] balloons, ok let me try with calendar [20:19] hehe, not that it matters honestly [20:19] perhaps just a paste of the full output of what your doing would help me [20:19] you can also run with -d for more debugging outpout [20:19] so adt-run -d .... [20:22] balloons, http://paste.ubuntu.com/7832261/ [20:25] Letozaf_, and you have a phone connected right? And just one phone? [20:25] balloons, just one nexus 4 [20:25] it's failing to connect.. [20:25] can you run phablet-shell and get acesss? [20:26] balloons, no, I suppose I have something wrong :( [20:26] letozaf@letozaf-PC:~/autopilot-tests$ phablet-shell [20:26] start: Job is already running: ssh [20:26] ssh-keygen: /home/letozaf/.ssh/known_hosts: Permission denied [20:27] Letozaf_, :-) there we go [20:27] balloons, maybe I should just reboot device and pc [20:27] sanity has return [20:27] balloons, I have been fidling to much [20:27] just rebooting the device is enough I'd say [20:27] balloons, ok [20:32] * Letozaf_ is rebooting pc [20:41] balloons, looks like it's working now [20:42] balloons, yes ! it's working thanks! [20:43] balloons, as soon as it ends I will try it on clock app [20:44] Letozaf_, awesome! [20:50] Letozaf_, so you should be able to run any app this way.. the only exception is reminders, which still needs a tweak to merge for it to work [20:51] balloons, ok, I am still waiting for calendar tests to end :P [20:51] Letozaf_, lol... you can stop them if you wish [20:51] balloons, ok I was wondering If I could :p [20:58] balloons, but if I branch my tests, and rename the branch ubuntu-clock-app, I get the old clock app tests running non the new one [20:58] balloons, but it's getting late now, maybe I will take a look better tomorrow night [20:59] Letozaf_, lol, just pass the directory name as the first argument [20:59] it matters not what it is called [20:59] Letozaf_, good night! [20:59] balloons, it's what I did first, but I will try again tomorrow and let you know [20:59] yep forsure [20:59] balloons, thanks and good night :-) === salem_ is now known as _salem [22:45] balloons, ping [23:08] knome, pong [23:12] ahh [23:12] so, [23:12] there are a few templates in the ubuntu wiki under QATeam and the old subdir Testing [23:12] can we remove the unused ones? [23:34] sorry I just realized my error from earlier with Carla [23:34] knome, yes no reason I could think of to not keep it clean [23:34] you probably know better what's used and what not [23:34] or is any template used? [23:35] I personally don't use any [23:36] okay [23:36] i'll see if there are any recent ones not edited by you [23:41] ok, the templates are cleaned up [23:41] you involved with any loco? [23:41] us-fl [23:42] ok, no templates for you ;)