[05:34] Saviq: no, no aware, let's try to find what it is about [08:28] Saviq: you've made test failures be failures instaead of unstable? https://unity8-jenkins.ubuntu.com/job/test-0-autopkgtest/15/ [08:53] tsdgeos, otherwise we'd end up in the make exit code situation potentially [08:53] ok, that's a bit of a step back but not terrible i guess [08:54] tsdgeos, you mean to not see the difference between unstable and failed? [08:54] yep [08:56] tsdgeos, I could interpret adt-run exit status http://manpages.ubuntu.com/manpages/precise/man1/adt-run.1.html [08:56] i guess it's fine, i mean it's just a few extra clicks [08:56] so if 2,4,6,8, exit 0 [08:57] tsdgeos, no I get what you mean, when we get reports on MPs it's important to know the difference [08:57] hopefully I can still mark it unstable if 2,4,6,8 [08:57] so we don't miss make exit [09:00] yeah easy enough to do [09:01] ok :) [09:30] d'oh, looks like we'll build up a dummy .xml file after all :P [09:57] mzanetti: Saviq: we're going to need a decision on whether we want https://code.launchpad.net/~aacid/unity8/fallback_for_empty/+merge/282002 or not [10:00] tsdgeos, mzanetti, you were arguing that it might be desirable to pass empty, I somewhat agree that fallback was meant for the case where the scope can't do anything about it, because they don't know if we successfully retrieve the image [10:00] when sending empty, they might as well send the fallback url again [10:00] s/again/instead/ [10:00] that's our thinking yes [10:01] but dobey strongly disagrees [10:01] my guess is mostly by the "fallback" name than by real disagreeing [10:02] ultimately it's the API guys' decision [10:03] because they're signing the contract with the scope developer [10:03] s [10:04] fair enough. [10:09] tsdgeos, https://unity8-jenkins.ubuntu.com/job/test-0-autopkgtest/16/ should come up unstable now instead of failed (/me should have ran the parallel qmluitests instead, cuts the time by half) [10:17] * Saviq starts thinking Jenkaas might be a good thing for us in the end [10:18] looks like I'll be able to cut the CI turnaround by a significant margin [10:20] but we seem to have "lost" almost twenty tests somehow [10:21] :? [10:23] tsdgeos, and btw, it was enough to make them run in parallel https://code.launchpad.net/~saviq/unity8/parallel-qmluitests/+merge/282603 ;) [10:23] tsdgeos, I mean https://code.launchpad.net/~saviq/unity8/parallel-qmluitests/+merge/282603/comments/717308 [10:24] tsdgeos, well, ok, not 20, but 7 https://unity8-jenkins.ubuntu.com/job/test-0-autopkgtest/13/testReport/ vs. https://unity8-jenkins.ubuntu.com/job/test-0-autopkgtest/13/testReport/ [10:24] Saviq: that's the same link twice [10:24] grr [10:25] tsdgeos, https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-vivid/1774/testReport/%28root%29/ vs. https://unity8-jenkins.ubuntu.com/job/test-0-autopkgtest/13/label=amd64,release=vivid+overlay/testReport/%28root%29/ [10:26] meld to the rescue [10:26] weird [10:27] tsdgeos, that's the diff http://pastebin.ubuntu.com/14503845/ [10:29] ah I wonder if it put the .xml file somewhere [10:30] or or [10:30] same name [10:31] or not part of xvfballtests [10:31] 'cause I can't see them at all in the adt-run log [10:31] tsdgeos, oh oh, and... good news and bad news https://requests.ci-train.ubuntu.com/#/ticket/877 (check out "Automated test results") [10:32] aren't they part of make test? [10:32] tsdgeos, I'll find them, nw [10:33] hey, got a pass for amd64 [10:33] still some failures though :/ [10:34] tsdgeos, the i386 fail on vivid is a time out, something got stuck (3h between start and timeout) [10:34] armhf is like a swiss cheese :/ [10:35] yeah :D [10:35] i'll have a look [10:35] i think most are timing related [10:35] but still it'd be great if we could get them to pass [10:35] 123 failures [10:36] we could use the .xml out of that [10:37] xenial i386 is just two fails in launcher that didn't get pulled out, so some improvement needed there, too [10:37] * Saviq worried to make them pass, they will shout Regression! at us all the time after that [10:38] ah, wonder if it's the parallel that made them stuck [10:38] hmm no, xvfballtests still went -j1 [10:39] only 52 failures on xenial armhf :) [10:42] he he [11:23] Mirv, hey, so yeah lp:ubuntu-settings-components cross-builds on vivid+overlay but not on xenial: https://unity8-jenkins.ubuntu.com/job/build-2-binpkg/arch=cross-armhf,release=xenial/163/console [11:28] Mirv, same for lp:unity-api https://unity8-jenkins.ubuntu.com/job/build-2-binpkg/164/ [11:28] owait that's actually a different - g++ - issue [11:31] Saviq, hey, I'm sure I saw a bug report coming by lately for apps crashing if you launch them immediately after closing them [11:32] can't find it any more, neither in my inbox nor on LP [11:32] you happen to know about it? [11:32] mzanetti, bug #1527737 [11:32] bug 1527737 in qtmir (Ubuntu) "Apps do not start if restarted quickly after closing" [High,In progress] https://launchpad.net/bugs/1527737 [11:32] faenil, ^ [11:32] thanks Saviq [11:32] ah finally :) thanks [11:33] we almost landed a fix, but found a bad regression that sometimes stopped you from launching apps altogether... [11:35] I see [11:35] left a comment there with my current experience [11:43] Mirv, *or* my chroots are borked, can't seem to reproduce locally [11:43] ah [11:43] proposed?? [11:46] Saviq: hmm [11:46] Saviq: maybe some gcc proposed update indeed [11:46] Mirv, that would actually explain the boost issue as well [11:46] Saviq: there was a new gcc-5 upload 14 hours ago [11:46] at least [11:46] hmm I first saw the issues earlier than that [11:48] boost stuff from a week ago is in release pocket [11:58] * Saviq recreates the chroots [11:59] Mirv, meh, can't reproduce locally, will let you know what I find out [12:01] Saviq: ok, let's see later [12:27] tsdgeos, yay, unstable https://unity8-jenkins.ubuntu.com/view/unity8/ (second from the top) [12:28] Saviq: third? [12:28] maybe third by now [12:28] parallel one, no? [12:28] yes [12:28] yellow-ish [12:29] instead of red [12:29] need to fix the upstream job, too (thought I did already) [12:29] ah [12:29] now I did [12:30] food! === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [13:48] Saviq, how trustworthy are the unity8 CI bot results so far? [13:50] dandrader: have a sec? [13:52] tsdgeos, yeah [13:53] dandrader: i can't figure out why mouseMove on the tests is failing to turn containsMouse of a MouseArea to true [13:53] i thought it was the MouseTouchAdaptor [13:53] but i've set it to false and still fails [13:55] dandrader: i can push the branch somewhere if you want [13:55] tsdgeos, MouseTouchAdaptor should be completely transparent to tests [13:55] tsdgeos, your second is up [13:55] :) [13:55] ok [13:56] tsdgeos, does it happens only in tests? [13:56] tsdgeos, or in "make tryFoo" as well? [13:57] make tryFoo works if i uncheck the "mouse emulates touch" [13:57] checkbox [14:01] tsdgeos, add some wait(xxx) between commands. helps to see if there's some race/timing issue [14:01] yeah i did, didn't seem to help [14:01] will keep trying, don't worry [14:02] it was just if you had something quick in mind [14:12] mterry, by now should be really trustworthy, modulo autopilot, which don't run yet [14:12] and armhf, which we can't build [14:12] and I need to find 7 tests that disappeared :P [14:12] Saviq, hah we can't build armhf? Because of ci restrictions? that seems like a big gap :) [14:12] mterry, no [14:13] because of bug #1472186 (I wanted to cross-build) [14:13] bug 1472186 in indicator-network (Ubuntu) "Can't install libconnectivity-qt1-dev on multiarch" [High,Confirmed] https://launchpad.net/bugs/1472186 [14:13] where possible [14:13] Saviq, ah [14:13] will likely need an armhf node for non-cross-buildable stuff [14:13] and then we have a phone, but don't want to use that for building, only for ap [14:43] on that note [14:54] is there a way to change the keyboard layout of a bluetooth keyboard connected to a phone? [14:57] nm, found https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1412492 === alan_g is now known as alan_g|afk [14:57] Launchpad bug 1412492 in unity8 (Ubuntu) "can't change the keyboard layout" [Undecided,Confirmed] [14:58] mzanetti, not sure what failed in launcher-sizing autopkgtest, added some debugging and restarted [14:58] oSoMoN, nope, not supported yet [14:58] hmm [14:59] that’s unfortunate, there’s a lot of people out there who don’t use qwerty and want to type non latin1 characters [15:01] oSoMoN, we're thinking hard about it ;) [15:01] but [15:02] people not using qwerty are... well, you know what I mean :P [15:02] Saviq, no I don’t, what do you mean? [15:03] I mean it's their own fault :P [15:03] should be forced to use qwerty for the sake of humanity [15:03] Saviq, right but it's mainly about being able to input non latin1 stuff :) [15:04] I'm using qwerty and input non-latin1 stuff all the time [15:04] not to mention compose key [15:04] we've done away with the "polish" layout some time ago and everyone uses "polish (programmer's)" now ;) [15:04] with AltGr to add diacriticts [15:05] but yeah, /me has caps mapped to compose, much more useful [15:05] does Mir allow remapping keys? [15:07] not yet I don't think [15:10] Saviq, yeah but not in u8, Mir forces a "en_US" layout [15:10] ltinkl, I know, I know :P [15:12] Saviq, otherwise I'm using a cz+qwerty+AltGr as well [15:21] grr /me broke autopkgtest job, fixing [15:27] fixeded [17:12] Saviq: is it possible to install&run click apps on desktop unity8 yet? [17:12] I played a bit with click install + register [17:13] but didn't get anywhere [17:13] (note: it could have been that the click was corrupt as well) [17:13] mzanetti, you have a recipe for that ↑ ;) [17:14] faenil, should work, yes [17:15] faenil, need click install, yes. but then it should just work [17:15] faenil, store is broken. you can still click on them to download them. they'll end up in ~/.cache [17:16] faenil, installation with pkcon will fail, but click install should work [17:16] you need multiarch or x86 packages obviously [17:17] mzanetti: yeah, does store hand out x86 or arm packages? [17:17] and I used click install + register but it didn't work...must have been the click then [17:17] mzanetti: pkcon is because it has apt configured as backend right? [17:18] I had a quick look at how to swap pkcon's backend but didn't find any simple way [17:24] Saviq, ltinkl: I could use a review for https://code.launchpad.net/~osomon/qtubuntu/qkeyevent-native-virtual-key/+merge/282515 [17:45] Here we go! https://developer.ubuntu.com/en/blog/2016/01/15/announcing-ubuntu-scopes-showdown-2016/ [17:49] faenil, yes, and yes [17:49] faenil, core apps are packaged multiarch nowadays [17:49] faenil, simple webapps or qml only would work too [17:49] mzanetti: okay, cool [17:49] mzanetti: nah, clicks === alan_g is now known as alan_g|EOW