[00:50] you want another build? sure [00:50] flexiondotorg, ^^ [00:53] flexiondotorg, I added it. you should be able to setup the rest and add a build I believe === pleia2_ is now known as pleia2 [01:54] When powering up Testdrive to help test beta 1 I noticed it doesn't list Ubuntu MATE in the preferences. Should I file a bug about that? [01:54] Is Testdrive even recommended/used anymore? [01:56] I am not sure I mainly have found just zsyncing easier and had some bugs last time I treid to use testdrive [01:56] it seemed to support virtualbox better than kvm [02:05] ianorlin, hm, I guess I could try zsyncing them. I always just used VB anyway so I didn't find its integration into Testdrive to be an issue. Hopefully I'll be able to test on my spare partition soon. [02:06] I used Testdrive because it was convenient, but might as well do it the hard way :p [04:18] hmm, would something crashing *after* the install (on first boot) fail the testcase or pass with with a non critical bug? [05:55] Good morning [05:56] dobey: you can also put your branch pretty much anywhere (github, gitorious, people.canonical.com), or even easier, just attach git format-patch to a bug report [05:56] the latter is easiest [15:07] elopio, lots of things seem broken today [15:09] brendand: there is a branch for each one. [15:10] elopio, yeah it's just depressing [15:10] brendand: today, it's the price we are paying for not using upstream code. [15:11] elopio, btw the bug fgimenez mentioned, it's already reported and fixed in testtools [15:11] elopio, but testtools still hasn't been updated in debian [15:11] for almost six months [15:12] ok. [15:12] should we prepare an updated debian package and give it to them? Or do the skip earlier in our code? [15:12] elopio, or move the check out of the setup [15:13] oh wait [15:14] fgimenez, do file a bug actually, against the sanity test suite [15:14] fgimenez, i'll take a look [15:14] elopio, i don't see any branch about the uninstall test failing [15:15] elopio, i get 'autopilot.exceptions.StateNotFoundError: Object not found with name 'PreviewActionButton' and properties {'objectName': 'buttonconfirm_click'}.' consistently [15:15] brendand: is that running the install branch? [15:15] elopio, yes [15:15] I've just fixed it. The branch is not ready for review. [15:16] it was a mistake I did yesterday. [15:20] elopio, does your system settings branch fix 'AttributeError: 'module' object has no attribute 'MainWindow'' [15:20] brendand: yes. [15:21] brendand: ok, you should be able to reproduce it in the test_sim_pin test without sim inserted [15:21] fgimenez, ok [15:28] rhuddie: our policy is one card per branch. [15:28] so you can make a card for your avatar branch, and put it in the review column. [15:29] you also can ping renato to get it landed. [15:30] elopio, ah, i see. thanks. I have some helpers to add yet, but also wanted to get the .deb file from jenkins [15:34] rhuddie: got it. Is there a way to split that card in more pieces? I'm not sure what to do in the evening. [15:34] elopio, for that branch, do you think it best to just get that single change landed and then add helpers to the sanity-tests branch? or try and get the helpers in with that MP? [15:35] rhuddie: the helpers need the test peers. Is that easy to do? [15:35] elopio, hmmm, i'm not sure about that yet [15:35] I saw some of your branches in progress for that, I don't know how close are they to land. [15:35] elopio, well they are quite old now [15:36] rhuddie: yes, but they don't need a lot of updates. [15:36] not much has changed. [15:36] i did the merge from trunk, which seemed ok [15:36] rhuddie: I would say, make the helpers in sanity. If we have time at the end, make an MP for the address book with as many of them as possible. [15:37] yes. i think we should be able to get the object name landed easily, but if we start doing the helpers it could take much longer [15:38] rhuddie: yes, I agree. [15:39] elopio, should I wait for jenkins results before getting another review? [15:39] rhuddie: no need. There's no way jenkins can fail with that. [15:39] :) [15:40] rhuddie: you probably have to update the copyright year there. [15:40] you know? just because I can't go to bed if I don't leave a needs fixing :) [15:42] fgimenez: I want to learn c++ from scratch. Any recommendations? [15:44] elopio: buf, perhaps i'm not the indicated person to give you directions on this :) [15:44] elopio, updated copyright [15:45] fgimenez: I suppose I'm confused. For some reason, I thought you were a c++ hacker. [15:45] to many emails going on this days. [15:47] elopio: i've worked with it but there's a lot for me to learn [15:50] I suppose this coude be good: http://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-096-introduction-to-c-january-iap-2011/lecture-notes/ [17:31] elopio, I keep getting this error when running the add contact test: http://pastebin.ubuntu.com/10411508/ [17:32] it seems like it should work, so not sure if you have any ideas? [17:33] rhuddie: did you remove the Address book helpers from sanity? [17:33] elopio, no [17:33] rhuddie: you should use now the ones upstream. There's an address_book_app.AddressBookApp [17:34] elopio, yes. i tried using that in the call to get_proxy_object_for_existing_process(), but it gave the same result [17:35] i feel like i must be missing something [17:36] rhuddie: if you didn't remove the one from helpers, it might be in the object registry. [17:37] elopio, so html5 and scopes. Are we comfortable yet in recommending things for testing? Presumably selenium and phantomjs [17:37] oh, no, crap. [17:37] the webbrowser typo got into trunk. [17:37] that's the problem. [17:38] elopio, oh! [17:38] balloons: for scope testing, pete is working on a tool to test them without ui. [17:38] for html5, autopilot + selenium + oxide. [17:38] elopio, did it? [17:41] I wonder how the tests pass. [17:43] elopio, ack, ty [17:43] elopio, ohh, for the 'tool' is there source or a name? [17:44] rhuddie: can you try with this? [17:44] https://code.launchpad.net/~canonical-platform-qa/address-book-app/fix_app_proxy/+merge/250969 [17:44] balloons: I haven't seen it yet. [17:46] elopio, with ubuntu-sanity-tests? how do i make it install that version on the device? [17:46] rhuddie: I would make the device rw. [17:46] install this and unity8-autopilot [17:47] adb push the ubuntu_sanity_directory your are using. [17:47] and autopilot3 run from phablet-shell === alex_abreu is now known as alex-abreu === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr [19:55] bdmurray, do you indeed intend to backport apport-noui to trusty? === seth_ is now known as sethj [20:09] nvm I see it's already there, awesome === roadmr is now known as roadmr_afk [20:28] balloons: its there but it doesn't work well [20:28] bdmurray, so what are the tweaks you describe then? Do they eliminate all prompts? [20:30] balloons: yes by the following [20:30] * debian/apport-noui.upstart: remove early exit (LP: #1235436) [20:30] * debian/apport-noui.dirs: create /var/lib/apport (LP: #1235436) [20:30] * data/whoopsie-upload-all: backport vivid version of it [20:30] * apport-noui: make the package installation automatically enable [20:30] Launchpad bug 1235436 in apport (Ubuntu) "/etc/init/apport-noui.conf is non-functional on the phone" [Critical,Fix released] https://launchpad.net/bugs/1235436 [20:30] autosubmission, and update the package description accordingly. [20:30] LP: #1351137. [20:30] Launchpad bug 1351137 in apport (Ubuntu) "ubuntu-system-settings-wizard does not create /var/lib/apport/autoreport on first boot" [Critical,Fix released] https://launchpad.net/bugs/1351137 [20:31] ahh some additional bugs [20:31] balloons: then the whoopsie change just makes it so that the OOPS ID gets logged [20:31] right, so you only pulled bug 1084979 [20:31] bug 1084979 in apport (Ubuntu Utopic) "Submitting error report asks confounding questions" [Medium,Fix released] https://launchpad.net/bugs/1084979 [20:32] I think eliminating all prompts especially on the LTS would be very welcome [20:32] we don't gain any favors by exposing LTS users to it [20:33] The apport-noui change requires people to manually install that package and sends all crashes without prompting to errors. [20:33] I don't know how large the audience is for that. [20:44] whee, so i am hitting more interesting annoyances in autopilot :-/ [20:47] dobey: what are you experiencing? [20:48] veebers: well, apparently self.launch_test_application() calls "which" for each argument. and it always inserts "-testability" as the second argument on the command line, regardless what the first item is [20:49] so ["which", ""] fails, and then replacing the "" with "sh" fails because it tries to do ["/bin/sh", "-testability", "/usr/bin/qmlscene", ...] [20:50] dobey: right, so launch_app is designed to launch applications in which to introspect (and thus returns a proxy object) [20:50] "ubuntu-app-test" if use_mir else "", [20:50] trying to do that [20:50] because we have to do ubuntu-app-test qmlscene foo to run it under mir [20:51] dobey: you should use UpstartApplicationLauncher if you want to launch something using ual [20:52] veebers: no i can't [20:52] dobey: that's wrapped in TestCase.launch_upstart_application [20:52] veebers: this isn't an actual application [20:52] so ubuntu-app-launch won't work [20:53] otherwise, i'd just use self.launch_click_package() and it'd work and i wouldn't have had to override it to do what i need to do :) [20:53] dobey: ah I see sorry, I read "UbuntuAppLaunch.start_application_test" in the source and saw 'test' and 'ubuntuapp..' [20:54] in what source? [20:54] dobey: autopilot [20:54] oh [20:54] yeah, that would be fine if this was a normal app [20:54] but it's not [20:54] it's a trusted session thing [20:55] dobey: what's ubuntu-app-test? [20:56] veebers: it is a tool that will hopefully be available in vivid soon, to be able to launch a trusted session for special things like we are, under mir, so we can run our autopilot tests on a device/emulator [20:57] dobey: I wonder if perhaps there needs to be a feature added to autopilot to handly loading applications in this manner [20:57] dobey, nice! [20:58] * balloons worries about fragmentation of autopilot launching apps [20:58] veebers: well, i guess using ubuntu-app-test would be that feature, but for now i'm happy to get it working just in my test [20:58] veebers: for now, i think we're the only app that really needs this for autopilot, so our tests is at least a good place to hash out how it should work :) [20:59] dobey: ack :-) [21:01] dobey: so, if it's not being run under mir wouldn't the application be qmlscene (in this example) and not "" or "sh"? [21:02] veebers: we need qmlscene in both cases [21:02] veebers: but i only need to prefix the command with ubuntu-app-test if we're running under mir [21:02] Evening. [21:02] elfy, You about? [21:02] veebers: autopilot seems to be making some assumptions about what the arguments are though, rather than actually examining them and doing the right thing [21:03] flexiondotorg: ish [21:03] elfy, So, I've "signed" off one image so far. [21:03] elfy, I have a somewhat annoying bug in Beta 1. Not a show stopper and only affects i386 so far. [21:03] dobey: perhaps I dont' understand properly, under mir: self.launch_test_application('ubuntu_app_test', ['qmlscene', 'other args']) [21:04] elfy, Is it OK to process with known issues? [21:04] and not under mir: self.launch_test_application('qmlscene', ['other args']) [21:04] I see most other flavours have way more bugs than the Ubuntu MATE guys have encountered. [21:04] flexiondotorg: we intend to release with a big known issue if it doesn't get fixed in time [21:04] elfy, Cool. [21:04] just make sure your release notes detail anything you think your audience will need to know [21:04] Any idea what time tomorrow the Beta 1 will be released? [21:04] elfy, Working on those now. [21:05] veebers: i don't think that's right [21:05] flexiondotorg: excellent === roadmr_afk is now known as roadmr [21:05] flexiondotorg: when laney is happy and I've chased people for stuff :) [21:05] veebers: i guess i'll just have to have an if and make the call twice, once with ubuntu-app-test, and once without, rather than doing it in-line [21:06] Rough idea of what time? [21:06] veebers: but i'm still worried that autopilot will add the -testability in the wrong place in that case [21:06] flexiondotorg: well - both of us are using a proper timezone where for 6 months of the year UTC=the clock :) [21:06] same one as you I think :p [21:07] dobey: hmm, I think if you add '-testability' explicitly as an argument it shouldn't add it [21:07] flexiondotorg: early evening I would guess [21:07] so this (untested) might work: self.launch_test_application('ubuntu_app_test', ['qmlscene', '-testability', 'other args']) [21:07] veebers: hmm, i'll try that [21:07] dobey: let me know how it goes, we might end up filing some bugs :-) [21:08] yeah, hopefully adt-run will go fast this time [21:09] dobey: also, might be of use, if you supply app_type (as in, app_type='qt') it will use the qt testability env settings without trying to figure it out itself [21:13] elfy, PowerPC baby. It works 😃 [21:13] lol [21:53] jibel, pitti: hey, looking at the apparmor entry for http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html, I see the systemd adt run failed. Looking at https://jenkins.qa.ubuntu.com/job/vivid-adt-systemd/lastBuild/? it seems the i386 run failed due to archive problems/test setup failure. Is there any way that test can be rerun? [22:07] veebers: adding "-testability" by hand helped, but can't default to "sh" because of course it tries to read qmlscene as a shell script then which fails. so had to refactor a little and use an if, calling launch_test_application() in two different ways [22:09] dobey: ah ok, so it's currently working for you? (albeit the codes not as elegant as you would like?) [22:10] veebers: well i don't know if it "works" yet on the mir case, but i think i have it working ok for the qemu/x11 case (running again right now to verify) [22:11] dobey: ah ok, understood [22:11] but i think it works, yes :) [22:11] and hopefully we will be able to get ted's fix landed in vivid soon, so i can verify it works under mir as well [22:17] dobey: sweet, keep me in the loop as I'm interested in anyway that autopilot can be improved [22:17] veebers: don't worry, i'm good at complaining :D [22:17] ^_^ [22:29] veebers: so ('foo', ['qmlscene', 'foo', ...]) doesn't work because it tries to parse the list as a string [22:29] whee