[00:31] I'm off for dinner, bbl [01:58] Kaleo: well, when we get mirvfb, the ubuntu-touch-session setup script can probably do all the stuff much nicer. if you're using adt-run run locally under qemu, you can use it to get an xvfb setup to run the tests under. [02:05] === IMAGE 131 building (started: 20150313-02:05) === [03:15] === IMAGE 131 DONE (finished: 20150313-03:15) === [03:15] === changelog: http://people.canonical.com/~ogra/touch-image-stats/131.changes === === chihchun_afk is now known as chihchun === marcusto_ is now known as marcustomlinson [05:23] I will not allow myself be turned into framebuffer, even though lately I've seen increasingly highlights about that [05:26] Mirv: I think you'd make a great framebuffer! [06:06] don't vote for me, vote err.. for example sil2100fb sounds good! [09:32] huh [09:32] Authentication issues [10:04] sil2100, we had an unexplainable failure with the emulator with the last image build, i'll kick another vivid image to see if it happens again [10:05] ogra_: ok, thanks [10:15] sil2100: I don't find ww13 bugs that would be clearly in Qt (although of course still possible that 5.4.1 would help). From our LP bugs Qt 5.4.1 PPA fixes bug #1430337 (which makes greyback_ happy) and Device Pixel Ratio improvements that mzanetti needs for his SVG work in progress, and then of course any of the crasher/regression/leak/networkdisconnection fixes from https://qt.gitorious.org/qt/qtbase/source/dist/changes-5.4.1 that affect us, whet [10:15] bug 1430337 in qtdeclarative-opensource-src (Ubuntu) "Segfault with QSG_VISUALIZE=overdraw" [Undecided,In progress] https://launchpad.net/bugs/1430337 [10:16] And then Kubuntu wants a certain text rendering fix, and XCB related fixes. [10:16] that's about the summary "what it's needed for" [10:19] oh right and the Qt Wayland support. [10:20] === IMAGE 132 building (started: 20150313-10:20) === [10:22] rvr, I confirm that I received the notification from gmail but only the indicator changed, no sound, vibration or anything else [10:22] jibel: This is the bug I opened https://bugs.launchpad.net/webapps-core/+bug/1431517 [10:22] Ubuntu bug 1431517 in The Webapps-core project "[Gmail] No notification displayed, no sound, no vibration" [Undecided,New] [10:23] Mirv: ok, well, let's try to convince Pat about that it's relatively safe then [10:24] Although for that I would prefer us to have a QA sign-off already [10:24] dbarth_, can someone havea look at bug 1431517, it's a potential regression in rtm/256 [10:24] bug 1431517 in The Webapps-core project "[Gmail] No notification displayed, no sound, no vibration" [Undecided,New] https://launchpad.net/bugs/1431517 [10:25] cihelp, hey, any idea why errors.u.c shows only one thread in "Thread Stacktrace" https://errors.ubuntu.com/bucket/?id=/usr/bin/unity8-dash%3A7%3AtestAndSetRelaxed%3AtestAndSetAcquire%3AtestAndSetAcquire%3AfastTryLock%3AQMutex%3A%3Alock ? [10:26] Saviq: so normally that wouldn't be a cihelp ping, but you're in luck that I'm the vanguard today. ;) [10:26] Backtrace stopped: previous frame identical to this frame (corrupt stack?) [10:26] ev, ;) [10:26] ev, well, yeah, but there's only one thread, which seems unlikely? [10:27] so it looks like the core file the retracer got was busted. You could argue that maybe if we got one from a different user from the same problem it may well work, so I'd suggest filing a bug against daisy and letting bdmurray know. [10:27] well I'm inclined to not trust anything it says when there's corruption [10:28] ev, there's like 160 reports in https://errors.ubuntu.com/problem/57e495ad9b12ea8e78d0aed997ba70a6e2213dd2 for example, and same situation - just one thread reported [10:28] though given that this is a single instance of the crash on armhf, I wonder if we're having problems with gdb again [10:28] * Saviq files then [10:28] yeah, that one is more convincing :) [10:29] Saviq: so we had problems a while back with gdb breaking on armhf. It could be a regression (and frustratingly not something we test for, I think) [10:29] kk [10:35] cihelp can someone fix jenkins building for the core apps ppa, it's pulling the old terminal and not the new one... https://launchpad.net/~ubuntu-touch-coreapps-drivers/+archive/ubuntu/daily - new is 0.7, old is 0.5 [10:36] popey: looking [10:40] jibel: ok, on it [10:41] dbarth_, it doesn't seem specific to gmail [10:41] dbarth_, thx [10:41] jibel: ah but i doubt it's anything webapp related though [10:41] i would mostly triage and ask saviq to take a look, as it seems more like a shell problem [10:41] sil2100, who can look at this notification issue? [10:41] jibel, dbarth_: all in all it's a bit worrying I must say [10:42] dbarth_, np, investigation must start somewhere [10:42] Saviq: related to that bug 1431517, what could cause a push notification not to be displayed by the shell? [10:42] bug 1431517 in The Webapps-core project "[Gmail] No notification displayed, no sound, no vibration" [Undecided,New] https://launchpad.net/bugs/1431517 [10:42] dbarth_: could you take a look if it's indeed not anything on your side? I guess Saviq could help identifying what parts of the shell deal with notificatinons [10:43] jibel: right, i looked into this one with rvr yesterday, and there is already a bit of a diagnostic [10:43] dbarth_, it's the push service that wakes the device and vibrates [10:44] dbarth_, so you'd have to talk to Chipaca [10:44] sil2100: i doubt it is, because the push-helper living in the webapp package is actually called and returns a valid reply to ask for vibration, and display of the notification [10:44] right, Chipaca is a good person to annoy at this stage :) [10:45] I don't see anything landing from ubuntu-push for this release [10:45] Chipaca: ping === chihchun is now known as chihchun_afk [10:56] sil2100: pong [10:56] hello [10:59] dbarth_: read the bug report. missing the push client logs, i can only guess as to what's going on. [11:00] dbarth_: but if the bug report is accurate, push client sent the notification to the messaging indicator, and the messaging indicator believes it's showing it (so much so it turns the led on) [11:00] dbarth_: if you can reproduce the issue, turn on debug logs in the push client, reproduce the issue, and get me the logs [11:00] Chipaca, attached [11:02] sil2100, hmm, https://ci-train.ubuntu.com/job/ubuntu-landing-019-0-reconfigure/29/console why is it complaining, is your "reconfigure" different from my "reconfigure" in any way ? [11:03] jibel: can you reproduce the issue? [11:03] Chipaca: ok i will do that [11:03] Chipaca, yes [11:03] Chipaca: i can reproduce the issue as well [11:03] Chipaca, with telegram or gmail [11:04] jibel: with *telegram*? [11:04] nice [11:04] jibel: turn on debug logs, please, reproduce, get me that [11:04] but if the led really goes on, it's the indicator, somehow [11:04] Chipaca, how do you turn on debug logs? [11:04] jibel: turn on debug logs => either edit /etc/xdg/ubuntu-push-client/config.json, or copy that directory to ~/.config and edit it there [11:05] jibel: the edit is basically “sed -i -e 's/info/debug/' /the/file” [11:05] popey: sorry this is taking so long - I needed to sort super admin powers on that jenkins [11:05] ev: np [11:06] ogra_, I don't think you can use the reconfigure job to add new components, they use the prepare job for that I think? [11:06] ouch, ok ... the message should make this distinction then :) [11:07] telling me that reconfigure failed but telling me that someone else needs to reconfigure is confusin [11:07] g [11:09] jibel: oh, forgot to mention you need to restart ubuntu-push-client for the change to take effect [11:09] Chipaca, that's fine, I rebooted the phone [11:10] jibel: that'll do :) [11:12] ogra_: looking [11:12] ogra_: yeah [11:12] ok [11:12] As Saviq mentioned, right now it's forbidden - let me reconf for you [11:12] thanks [11:12] Chipaca, https://bugs.launchpad.net/webapps-core/+bug/1431517/+attachment/4344170/+files/ubuntu-push-client.debug [11:12] Ubuntu bug 1431517 in The Webapps-core project "[Gmail] No notification displayed, no sound, no vibration" [Undecided,New] [11:15] jibel: so, yeah, push is doing is asking for the notification centre entries (and being told they're there) [11:22] ogra_: ^ done [11:22] thanks !! [11:23] sil2100 if it is not a problem with the apps and with push client, who can look at the notification? [11:23] jibel: I'm looking at the changes again, hm [11:24] sil2100, changes from last promotion http://paste.ubuntu.com/10590324/ === MacSlow is now known as MacSlow|lunch [11:26] Chipaca: so all is good on the push side? [11:26] This is absurd [11:26] sil2100: from those logs, all seems correct [11:27] sil2100: spotted an unrelated issue wrt account-polld and individual message actions, but that's nothing to do with this :) [11:27] I think we need to get all hands on deck [11:27] davmor2, did you flash stable to confirm the regression with notifications or should I? [11:28] Saviq: I see some notification-based changes in unity8 - do you think this might cause some problems as we're seeing right now? [11:28] jibel: next job I think and then I can run the upgrade to 256 then [11:28] davmor2, OK [11:28] sil2100, you mean in silo 6 or? [11:29] Saviq: no, in here: paste.ubuntu.com/10590324/ [11:29] In rtm [11:30] === IMAGE 132 DONE (finished: 20150313-11:30) === [11:30] === changelog: http://people.canonical.com/~ogra/touch-image-stats/132.changes === [11:30] Saviq: we need someone with knowledge of the whole notification chain helping out triaging this bug [11:30] As it's a blocker for OTA-2 promotion [11:30] sil2100, i suspect we'll need more image rebuilds to fix this issue ... [11:30] ogra_: ACK [11:31] sil2100, if you're not getting vibration or power, that's not something unity8 is dealing with [11:31] * ogra_ starts another one with a more verbose livecd-rootfs [11:31] sil2100, I can test that we're displaying all the notifications as they arrive, did that when landing this change, too [11:32] sil2100, but currently (I know, bad arch) unity8 does not wake screen up or vibrate on notification [11:32] popey: it looks like you guys haven't updated debian/changelog to say 0.7 is the new version [11:33] in r13 of lp:ubuntu-terminal-app you bumped to 0.7 [11:33] but debian/changelog is still at 0.5 [11:33] hm [11:33] you can see this in action at http://91.189.93.70:8080/job/generic-land/2942/console [11:33] Saviq: if push is working fine, who do you think we should poke about the notifications not happening? [11:33] ev: well, sorry about that. I'll take a look, thanks. [11:33] sil2100, and since notification center and notification bubbles are separate still, whatever is putting the msg in the notification center seems to not be doing the legwork to wake up, vibrate and send the notification [11:34] sil2100, define "push is working fine" [11:34] Chipaca mentioned that from the logs push seems to be doing everything it's supposed to be doing [11:34] Chipaca: ^ [11:34] (from the logs) [11:34] Chipaca, it's push that wakes the screen up and vibrates, right? [11:35] === IMAGE 133 building (started: 20150313-11:35) === [11:35] unity8 is not doing it, that's for sure, but maybe there's another component that should, but isn't [11:35] Those are the questions I'd like to know the answers for [11:35] sil2100: who is on the image building spree [11:36] ogra_, ↑ ; [11:36] davmor2, me, trying to fix th eemulator builds that fail since yesterday [11:37] Chipaca, in the log there is "playing sound /usr/share/sounds/ubuntu/notifications/Slick.ogg using paplay" if I use paplay from phablet user is it supposed to work? [11:37] ogra_: haha! [11:37] sadly it looks like they run fine til the very end ... but we end up with a 0 byte device tarball [11:37] popey: no worries at all; I got to learn a bit more about the internals of this thing [11:37] ogra_: why it does work anyway :P [11:37] so i'll have to do a few rebuilds with adding more and more verbosity to finally find the sisue [11:38] jibel: sounds on rtm aren't working, that's a separate bug [11:38] doesn't even [11:38] k [11:38] jibel: that's rsalveti [11:38] somehting about pulseaudio's mom coming to visit [11:43] dbarth_, sil2100, if you keep the screen on, is the notification bubble displayed, or is it just the envelope that goes green? [11:46] jibel: ^ ? [11:50] jibel, also, do SMS and incoming calls work fine? if so, it really must be somewhere between push and usensord/powerd [11:51] Saviq: just the enveloppe [11:52] I just tried SMS and calls, and is fine, vibration, power and bubble, so everything seems to be working fine on the receiving ends of this [11:53] someone needs to dig into push and what it's calling to trigger the above, why isn't it getting through [11:54] Saviq: are you testing on RTM? i have image #19 here and rvr was on rtm-proposed [11:56] dbarth_, rtm-proposed [11:56] r256 [11:56] Yes, SMS is fine in rtm proposed [11:58] so yeah, the broken bit is somewhere between push and usensord/powerd AFAICT [11:59] :| [12:04] Saviq: how usensord/powerd? [12:04] Saviq: not sure where that one comes in :) [12:04] Chipaca, you're talking to them to trigger vibration and screen, no? [12:04] "you" == ubuntu-push [12:04] Saviq: "screen"? you mean the led? [12:04] Chipaca, no, I mean screen, as in backlight [12:05] we don't touch the backlight, no [12:05] someone has to [12:05] Saviq: why? [12:05] the screen doesn't go on when a notification arrives [12:06] Chipaca, that's what people complain about :) [12:06] uh? [12:06] that there's no notification bubble or vibration on incoming email [12:06] jibel, dbarth_ ↑? [12:06] for emails, vibrations are explicitly not asked for [12:07] ditto for bubbles [12:07] notification centre entries should be there though [12:07] oh, let me check bubbles wrt emails [12:07] i know telegram does not ask for them [12:07] so, yes, gmail asks for bubbles [12:08] jibel, sil2100, dbarth_, the plot thickens ↑ [12:08] but bubbles does not mean 'turn the screen on' [12:08] however the bug also says the notification centre is empty [12:08] or did i misread? [12:08] Chipaca, there's no bubbles even if screen on apparently [12:09] Chipaca, that I'm not sure of, at least the LED comes on, so the envelope goes green [12:09] Uh oh! [12:09] which means something did happen to the notification centre, whether an item got added, not sure [12:10] * Saviq has no gmail account, and is not really even involved in this, other than improving communication apparently [12:10] Saviq: yeah, the notification led coming on but the notification centre being empty is why i think the problem lies in the indicator [12:11] but i had missed the thing about the bubble [12:11] jibel: poke [12:11] * Chipaca wants to confirm [12:11] Chipaca, yeah, if LED on and no item in centre, that's indicator [12:11] or well, could be unity8 too in theory, but SMS and missed calls work, so... [12:12] Saviq: also because in the debug log, push *thinks* it's asking for (and getting) everything it's been told to [12:21] Let me try upgrading and check that [12:23] sil2100: is this on rtm-proposed on krillin? [12:31] Chipaca: yes [12:32] sil2100: any risk to me flashing it? [12:33] No ;) It's our promotion candidate, so it's good in overall [12:33] * sil2100 tries to add his google account [12:33] sil2100: that's still ubuntu-touch/ubuntu-rtm/14.09-proposed yes? [12:33] hm, I'm unable to do that apparently [12:34] Chipaca: yes [12:34] sil2100: the problem seems to not be only gmail, so you could install poke [12:34] jibel, davmor2: I apparently can't setup my gmail account - the google account page ends up white all the time [12:35] sil2100: congratulations I've managed to set it up across the board :) [12:36] Oh, I think I just have connection issues [12:41] davmor2: how do the gmail notifications work? If I'm logged in the gmail account and have it running in the BG, should it automatically inform me of new e-mail? [12:42] sil2100: yes [12:42] sil2100: you need to enable gmail in the google account and login once I think then you should get notifications [12:42] sil2100: and you can prod it to poll sooner over dbus [12:42] Chipaca: ok, I got a notification popup [12:42] phew :) [12:42] No sound, but the bubble appeared [12:42] sound doesn't work in rtm [12:43] Not sure about wibration, it might have vibrated but I'm not sure ;p But it was on a screen that's turned on [12:43] Should it work when the screen is turned off? [12:44] sil2100: yes, it should, but if it isn't vibrating, and sound isn't working, how would you know :) [12:44] sil2100: is it appearing in the notification centre? [12:44] Chipaca: I saw it on the notification indicator [12:45] hm, on screen turned off I didn't seem to get a notification [12:45] no sound in rtm is bug #1413818, fwiw [12:45] bug 1413818 in pulseaudio (Ubuntu RTM) "Notification sounds aren't being played back" [Undecided,New] https://launchpad.net/bugs/1413818 [12:45] Chipaca, sil2100: so in image 16 I get blue flashy light and an item in indicator but no bubble for gmail and nothing for telegram, in 19 I get blue flashy light an item in indicator and no bubble for gmail and a vibrate osd and item in indicator for telegram [12:45] Saviq: ^ [12:46] davmor2: gmail is not asking for vibration, so you wont' get one [12:46] Chipaca, sil2100: ofcourse I blame Saviq for all of this :) [12:46] davmor2: so it seems not to be a regression, right? [12:46] Chipaca: yeah but I don't get OSD on gmail either [12:46] davmor2: since I got an OSD on gmail on the latest -proposed [12:47] (when screen was on) [12:47] sil2100: ah I'm here with screen off I'll try again with screen on [12:48] communication shutdown [12:48] davmor2: how are you checking for bubbles? [12:48] davmor2: if the screen is off, how do you know you don't get a bubble? :) [12:48] (hint: you can't; the screen is off) [12:49] Chipaca: normally it wakes the screen or is that linked to the ping/vibrate? [12:49] davmor2: it doesn't normally wake the screen [12:49] i don't know where that impression comes from, but it's a false impression [12:50] Chipaca: SMS, calls and alarms that all trigger OSD bubbles [12:52] davmor2, all those wake the screen up explicitly indeed [12:54] davmor2: so it does. Push doesn't do that. [12:55] === IMAGE 133 DONE (finished: 20150313-12:55) === [12:55] davmor2: nor have we been asked to do that. Not sure we should do that or not :) [12:55] === changelog: http://people.canonical.com/~ogra/touch-image-stats/133.changes === [12:55] huh [12:55] in testing i just got a bubble for twitter, on rtm-proposed, fwiw [12:56] not sure what we're looking for any more. jibel? [12:56] Chipaca, Saviq, sil2100, jibel, rvr: okay so upgrade to 256 is all good, gmail does indeed display an OSD when the screen is on [12:56] i should make account-polld ask for vibrations [12:56] I think we need jibel and rvr to double-check the expectations here [12:56] and maybe find out whether we want to turn on the screen when showing a popup [12:57] Chipaca, yeah, it's all a bit confusing, is there a spec somewhere that describes the expected behaviour of the different notifications? [12:57] also look into the %!s(MISSING) bit of the log [12:57] Since the bug expects vibration, sound and screen notification, but that's not the case? [12:57] jibel: i don't think there is a spec as to what to expect from gmail notifications [12:58] jibel: i can point you at the developer docs that say what is available [12:58] jibel: but i don't think you're asking for that [12:58] Chipaca, no, I'd need a design doc [12:58] jibel: for gmail notifications [12:58] Chipaca, gmail and other notifications in general [12:58] jibel: but you're not asking about other notifications in general [12:59] i mean, other notifications in general vibrate, because they ask to vibrate [12:59] Chipaca, designing tests from implementation is a guaranteed failure :) [12:59] you're expecting notifications that don't ask to vibrate to vibrate :) [12:59] Chipaca, jibel, the notification system is kind-of a mess right now, we never had the time to converge it all properly, but design team is currently working on a notification refresh, which is when we should ask those questions :) [13:00] Chipaca: So I don't mind the OSD not waking the screen it saves the battery, but I think vibrate would be nice just so there is some sort of notification that there is a mail there maybe [13:00] davmor2: yes, i think account-polld should ask for vibrate [13:00] davmor2: sergiusens can turn them off if he dislikes them ;-p [13:00] Chipaca, I'm looking for a doc that says "Here is the beahaviour when I receive a notification from gmail, here is the behaviour when I receive a notification from app X, ..." [13:00] jibel: no such thing exists, and i doubt it will exist ever [13:00] * sil2100 sighs [13:01] jibel, it will always miss $my_favourite_service [13:01] We should get our designs straight [13:01] sil2100, shush [13:01] sil2100: that is being done, as Saviq pointed out :) [13:01] Saviq: look it's all your fault anyway ;) [13:01] yeah, Saviq, all your fault [13:01] Saviq, yeah, but there are probably classes of apps that share the same behaviour [13:01] you owe us all cake, now [13:02] anyway, we can as well remove these tests from the test suite [13:02] jibel, so yeah, you should've said "email" not "gmail" :) [13:02] jibel: that's closer to what i was trying to get at [13:02] jibel: what are you testing for? === alan_g is now known as alan_g|lunch [13:03] Saviq, jibel: it's things like do we treat telegram the same way as sms being as the two are basically a messaging app? [13:04] Chipaca, we are testing rtm for regression but some of the initial tests have been designed from the implementation, and without reference it's hard to say if it's a regression or an expected change in the implementation [13:04] davmor2, *I* would agree that it should be the same, it's all IM [13:04] jibel: sorry, i'm trying to get to concrete things we can fix. You say tests have been designed from the implementation, but you seem to be testing for things which have never worked [13:05] davmor2: Saviq: I hope, as part of implementing the new notifications spec, we can unify notification handling and not have n paths for it, which is at the root of dissimilar behaviours [13:06] jibel: e.g., never has the screen come on to show a gmail notification bubble [13:06] jibel: never have gmail notifications vibrated [13:06] jibel: etc [13:06] Chipaca, concretely sound/vibration on incoming telegram message or gmail worked and is not working anymore, not sure when it changed though [13:07] Chipaca, same here (hope) [13:07] jibel: ok. sound is not working in rtm (that's bug 1431517) [13:07] bug 1431517 in The Webapps-core project "[Gmail] No notification displayed, no sound, no vibration" [Undecided,New] https://launchpad.net/bugs/1431517 [13:07] Chipaca: why not enable everything and in the notifications page in setting allow people to select what they get for each app? Novel concept I know :) [13:07] um [13:08] jibel: sorry, wrong bug [13:08] jibel: bug 1413818 [13:08] bug 1413818 in pulseaudio (Ubuntu RTM) "Notification sounds aren't being played back" [Undecided,New] https://launchpad.net/bugs/1413818 [13:08] davmor2: notification setting page would be too complex [13:08] davmor2: although we come pretty close [13:09] davmor2: you can disable vibrations (for all notifications), or all notifications [13:09] it's something :) [13:09] davmor2, then you'll ask for setting per-app per-profile (silent, quiet, DnD etc.) :P [13:09] * Chipaca will have a D&D profile just to confuse people [13:10] iOS have that [13:10] You can configure which can of notifications are displayed [13:10] rvr: not saying it can't be done, saying it would take more design power than we currently can wield [13:10] jibel: so, what should we do in this case? [13:10] jibel: I'm not sure we still know what changed, and I'm not sure we even know what's the expectation here [13:11] I'm a bit worried it might have worked by 'accident' in the past or something [13:11] sil2100, davmor2 compared to 16 and 19 and apparently it is the same behaviour [13:11] sil2100: jibel: I was worried this problem was a regression, but seems it is not, because notifications are an important feature for users. [13:11] * Saviq thinks it actually never worked, but memory is flawed [13:11] rvr: good call, I would say it's important, but indeed it doesn't seem like a regression at least [13:12] I'm sure it worked because notifications during the beta program in January was driving me crazy :) [13:12] jibel: and have just confirmed that account-polld--generated notifications haven't had working vibrations for several months at least [13:12] although it seems that's unintentional :) [13:14] and gmail notifications used to work too, don't know when it stopped. [13:14] jibel: "work"? [13:14] Chipaca, I mean sound + vibration on incoming gmail message [13:15] jibel: outside of rtm, sound works; that might be confusing you? [13:15] Chipaca, maybe [13:16] jibel: vibration hasn't worked since august of 2014 at least [13:16] *for gmail* [13:16] rvr: so the bug is invalid per se, i guess; though not uninteresting, considering the discussion above [13:16] (and other account-polld sources) [13:17] i can use the bug to fix the vibration thing, which is a real bug and needs fixing (although not a regression) [13:17] or i can create a new bug for it :) [13:19] I marked our test as "To repair" and we'll figure that out after this release. At least not a regression compared to latest promotion. [13:19] sil2100, ^ [13:19] jibel: +1 [13:19] jibel: +1 [13:20] morning [13:22] ogra_: any news regarding the emulator image? [13:23] saw you built many images today [13:23] rsalveti, it just started working with the last build [13:23] and i dont know why [13:23] hm, alright [13:23] ogra_: what was the issue exactly? [13:23] we ended up with a 0byte generic_x86 file [13:24] so it failed to pull that from the builder to cdimage [13:24] but the file is fine inside the android package ... [13:24] what i did was to replace all cp commends in livecd-rootfs that cares for copying the android bits with "cp -v" to get more info in the log [13:25] and it just started working [13:25] *commands [13:26] so it is something to keepan eye one, but works for now [13:26] ogra_: haha, alright [13:26] yeah [13:26] it only started with tonights image ... might be the moon phase or so === MacSlow|lunch is now known as MacSlow === alan_g|lunch is now known as alan_g [14:33] trainguards: ^^ can i get a silo for line 52 please? thanks [14:33] dobey: on it! === chihchun_afk is now known as chihchun [14:56] dobey: one moment, some technical problems ;) [14:59] sure [15:13] sil2100: hmm, seems to be broken still? dashboard doesn't show anything, and trying to build silo 25 results in error about no config for it [15:13] huh? [15:14] oh [15:14] nevermind [15:14] somehow i was looking at rtm not ubuntu [15:14] dobey: looks ok here [15:14] :) [15:16] == Images krillin 256, mako 214, generic_x86 206 PROMOTED to RC == [15:47] jibel, davmor2: is everything more or less wrapped up on your side? Will you have some cycles to already resume vivid sign-off? [15:47] sil2100: no we are to busy partying like it's 1999 [15:48] hm, oh partying? Then maybe you want to skip the evening meeting once again to have time for more dancing, hmm? [15:48] HMM? [15:50] sil2100: what neetming hic you're not my dad hue carnt tell ME wata due hic [15:52] sil2100: not we are cracking on with sanity on mako and silos for vivid now [15:53] I'm actually once again thinking about the evening meeting, since in such a case like today I don't want to waste anyone's time if everything's clear [15:53] popey: do you have anything for us in the evening meeting? [15:53] sil2100, it's resumed already [15:54] I'm fine with having the meeting, but not if there's nothing new to discuss [15:56] sil2100: no [16:04] jibel: do you have anything for the meeting? [16:05] sil2100, bot really [16:05] not* [16:05] robru, ogra_, davmor2, rvr: anything from you guys? [16:05] nope [16:05] Nope [16:05] ogra_: how's the emulator? I heard it just fixed itself? [16:05] images build fine again [16:05] nope [16:06] sil2100, yeah, no idea what that was [16:06] hm, I'm starting to feel bad by cancelling yet another evening meeting [16:06] Should I feel bad about that? I wonder [16:06] * ogra_ thinks its is fine [16:07] canceling meeting when there is nothing on the agenda is not a bad thing [16:08] +1 [16:09] robru: skipping the meeting once again o/ [17:01] sil2100: sweet [17:21] bregma: landing 26 === alan_g is now known as alan_g|EOD [18:13] hmm [18:15] dobey: what's up? [18:16] robru: just wondering when ToyKeeper will be able to retest my pay-ui landing [18:17] dobey: I think her day usually starts in 3-4 hours. [18:17] dobey: The OTA image was already built and is almost done being tested... it's probably too late. [18:17] or she's up early ;-) [18:17] (was probably too late yesterday) [18:18] ToyKeeper: it's too late for the OTA image indeed. but we still want to get it in the store as an update (and then it will get pulled into the next image for ww13 or whatver) [18:18] oh [18:19] ToyKeeper: did you move to .nz or something? [18:19] No, I've pretty much always been on a NZ-like schedule. [18:19] Mirv: Mega-silo 12 is mega-failing. Doesn't pass the bq logo screen after installing the packages. [18:20] ToyKeeper: ah, ok [18:20] ToyKeeper: NZ-like? I thought you lived in NZ! [18:21] heh [18:21] I just woke up, but it looks like the OTA already got promoted about 3 hours ago. [18:22] dobey: So, I should be able to get back to silos since there's nothing left for the OTA. [18:22] ToyKeeper: ok, thanks [18:22] dobey: Is it already landed in vivid? [18:23] ToyKeeper: pay-ui is a click package only [18:23] Right, okay. I had been thinking it was targeted at one or the other. [18:24] I think this last OTA is the end of 14.09, so I'll try it out on vivid. [18:24] nope, just the store. but it is included in the images by default, so we wanted to get it into the ota image too [18:24] (if I understand correctly, anyway) [18:24] (but I'm not fully awake yet...) [18:25] ToyKeeper: well, pay-ui has to work on both, so testing on rtm would be best i think [18:25] (or both even) [18:27] I should remember not to attempt thinking until I've been up for at least half an hour. [18:28] sometimes you forget though :) [18:31] rvr: hey, don't test silo 12 for now [18:31] rvr: it's not something for a normal sign-off [18:31] It's a big one with 5.4.1 and required actually a separate image and almost a full regression suite ran [18:32] ugh google calendar doesn't show week numbers either [18:32] sil2100: We talked about installing the silo in our devices and do an informal check during the weekend [18:32] sil2100: (We = ops team) [18:33] sil2100: But after installing silo packages, the system doesn't boot [18:36] uh oh [18:36] Mirv: boing ^ [18:36] o/ [18:43] brbrunch [19:53] popey: Music app is good \o/ [19:57] davmor2: ooh! [19:57] davmor2: good as in "seems okay so far" or "clear to upload" [19:58] popey: clear to upload it does everything it's meant to as far as I can tell [19:58] thanks davmor2 [20:01] \o/ uploaded [20:01] ahayzen_: ^ [20:18] rvr: did you follow the instructions about cleaning QML cache etc? [20:19] rvr: and not using citrain tool [20:20] davmor2: are you on the ops team too? ^ [20:20] I am [20:21] Mirv: ^ [20:21] davmor2: did you upgrade and do you know if rvr (or others) maybe thought they can use normal citrain tool? [20:21] Mirv: probably [20:22] cihelp: i need help understanding how to get a project set up in CI train [20:22] Mirv: are you saying you can't? [20:22] davmor2: yes, like the instructions says [20:23] davmor2: could you move the card maybe back from Failed? [20:23] zyga: I think you need trainguards for that [20:23] dobey: (let's talk here) [20:23] josepht: I don't understand what that is [20:23] Mirv: which instructions I was about to install it here [20:23] ok [20:23] davmor2: https://wiki.ubuntu.com/Process/Merges/TestPlans/Qt "and instructions for updating phone or desktop at https://wiki.ubuntu.com/Touch/QtTesting - do not forget to remove the QML cache before rebooting! " [20:23] zyga: s/cihelp/trainguards/ :) [20:24] josepht: ah, I see [20:24] it's not a bot :) [20:24] josepht: i don't think trainguards manage the config for CI do they? [20:24] dobey: is the CI system specific to phablet/click packages? [20:24] zyga: no [20:24] dobey: specific to .deb packages? [20:24] zyga: the qt 5.4.x update goes through the ci train for example [20:24] dobey: if no, then I think it could be useful for checkbox [20:25] dobey: is that something we can use for per-merge-request CI? [20:25] * Mirv makes the instructions on the testplan page more bold [20:25] zyga: yes, there's jobs that run tests on MPs [20:25] dobey: no [20:25] davmor2: in short, citrain tool fails to cope when something in the silo has a new dependency that needs to be installed from normal archives, after which citrain proceeds to remove half of the rootfs [20:25] Mirv: nice one I'll try that now if I get it all up and running I'll move the card then does that sound fair :) [20:25] dobey: how can I define what kind of tests run? [20:25] zyga: see all the "PS Jenkins bot" comments on https://code.launchpad.net/~dobey/pay-ui/mir-app-test/+merge/251163 for example [20:26] zyga: are you looking for tests to be run on MPs? [20:26] davmor2: thanks! do you know if anyone else besides rvr needs to be pinged? [20:26] zyga: via debian/rules (for when building a deb), and via autopkgtest config for click packages, plus the config bits in the CI system itself [20:26] Mirv: no I'll add a note to the ticket [20:26] josepht: I'm not sure, I don't understand any of the UE CI stuff yet [20:27] josepht: zyga has checkbox currently being merged via a tarmac outside of CI train. i was suggesting it would be good to get that moved into the CI train [20:27] dobey: we're not building debs or clicks like that, we don't have packaging in our branches [20:27] popey, \o/ [20:27] josepht: I'm not super familiar with that, but maybe that's the autolanding cu2d-config job? [20:27] we have a script in the tree that discovers tests to run, sets up containers/vms and proceeds to run each test in each environment [20:28] then that either passes or fails the merge request [20:28] hmm [20:28] we can edit one file and say that we want to test on vivid now, or stop testing utopic [20:28] similarly we can add a new library and its dependencies [20:28] Ursinha: I think cu2d-config is what he's after [20:28] or stuff like that [20:29] I'm trying to understand how that translates to this system [20:29] we build a number of libraries and tools [20:29] some of which end up in debs later (and are packaged in debian and in a set of PPAs) [20:29] davmor2: ok, thanks! too bad if rvr is already gone so he doesn't get to test it [20:29] some of which are built as click packages though that process is manual for now (we run that script once a release is planned) [20:30] debian packages have DEP-8 tests and build-time tests [20:30] click packages have nothing as expected [20:30] how can I translate that over to the CI train (is that the right name?) [20:30] davmor2: also, it may be confusing that I still included the citrain commands on https://wiki.ubuntu.com/Touch/QtTesting even though I said not to use them, so I now put them as strikethrough so it's more clear to use the manual dist-upgrade instructions below [20:31] oh and we have like 30 branches monitored by one tarmac [20:31] what I [20:31] what I'm describing is just lp:checkbox [20:31] Mirv: commands run [20:31] but the mechanics is generic [20:31] zyga: we have a bot that monitors branches; when an MP is proposed it kicks off a jenkins job to run the appropriate tests (I'm not sure how customizable that bit is) and merges the branch (assuming the correct number of approvals are given and such) [20:31] zyga: i'm not sure what all exactly you're doing in checkbox. it sounds like maybe you're doing way more complex things than i was expecting though [20:32] dobey: it's an old project with lots of things to support :/ [20:32] dobey: we really want our 12.04-based tests to run for example [20:32] Mirv: looks good here I bet he just used citrain [20:32] dobey: as sometimes we introduce something that fails on python3.2 [20:32] josepht: where are the definitions of things that jenkins runs? [20:32] fginther has all the answers, if he's around he could clarify and we would all learn :) [20:32] zyga: right, i'm sure CI train can run tests on a precise node/vm [20:33] Mirv: I'll update the ticket now look out for an anouncement on Monday Landing meeting ;) [20:33] * fginther catches up [20:33] is the CI train capable of working with non-bzr branches (wink wink) [20:33] no [20:33] :-) [20:33] is that planned at least? [20:34] zyga: the configs are here: http://bazaar.launchpad.net/~cupstream2distro-maintainers/cupstream2distro-config/trunk/files [20:34] probably not before launchpad git support is finished [20:34] zyga, the CI infrastructure is completed geared toward launchpad projects and debian packages (and to a vers small extent click packages) [20:34] to be supported somewhere yes, we still don't know if in CI train as it is :) [20:34] davmor2: you managed to upgrade already? cool! [20:34] Mirv: yes I have Broadband ;) [20:34] davmor2: thanks. I also replied to rvr's e-mail in case he checks it still so we'd have two extra testers. [20:35] davmor2: oooh, I've heard of those! :) [20:35] but you can't use tarmac without bzr either :) [20:35] dobey: we don't love tarmac [20:35] :( [20:35] dobey: what we have can run without tarmac [20:35] dobey: in fact, it's something we can run from the tree at any time [20:35] dobey: so it's useful for reproducing issues [20:36] dobey: zyga: CI Train does not run tests beyond what get run when you 'debuild' in your source tree. the train just automates building of debs in a PPA, then you manually verify the debs, and if you approve them, then the train merges your merges for you. it's very rudimentary and very manual. [20:36] dobey: we'll be off bzr the day it's possible though that's a secondary topic [20:36] robru: so we cannot use it as we're not keeping packaging in trunk [20:36] well, sure. tarmac is designed to run things from the tree the same as the would be run by the developer [20:36] robru: packaging is in debian and in a number of packaging overlays [20:36] zyga: yeah that's the other thing, you need debian/ in trunk [20:36] robru: oh, then this is not going to work for us [20:37] robru: we keep 10+ projects in one tree as this is much easier to work with (it's one project but it has libraries and plugins and stuff and packaging is pretty complex) [20:37] robru: where project == something we release and package as separate thing [20:37] ick :) [20:37] zyga: yeah sounds like a really poor fit. train enforces lp project == debian package [20:37] robru: because we didn't want to invent anything that can combine sensibly $N trees or have to have $N trees around and synchronized [20:37] robru: I see [20:37] zyga, yeah, that is something that the CI MP testing and ci-train were not designed to handle [20:37] robru: ok, that settles it for now then :/ [20:37] you're going to be one of those people who uses submodules when you move to git, aren't you? [20:38] dobey: maybe, we're not sure [20:38] dobey: one of the reasons is that we don't have a stable api by design [20:38] dobey: more like linux + drivers [20:38] dobey: though we've started a process where that changes [20:38] dobey: and we have 5+ new projects that have stable api and are handled separately of lp:checkbox [20:38] dobey: so over time checkbox will become smaller [20:38] dobey: but that's a process [20:39] fginther: so about the way it works though [20:39] fginther: can the CI system support a project that wants to build and test for precise, trusty, utopic and vivid from one tree? [20:39] zyga: yeah if all you need is "a bot that runs tests when an MP is proposed" then lp:cupstream2distro-config might be able to sort you out [20:39] but that's different than the train [20:40] * zyga looks [20:40] well, all I want is a git hook [20:40] and we'll script the rest via bits and pieces [20:40] just lp has no hooks [20:40] so we improvise with tarmac [20:41] * zyga still wishes for a travis like system for canonical [20:41] zyga: right, lp:cupstream2distro-config has the setup that allows you to configure what tests to run and when, based on scanning your trunk for MPs against it. it's not "a hook" but it can make things happen shortly after MPs get proposed. [20:41] where one can say "I want to support those configs" [20:41] "this is how you provision a config" [20:41] "this is how you test" [20:41] do the rest [20:41] * zyga reads that code [20:42] zyga, sorta, the CI system can build the same source package against multiple releases, but post build testing is very limited. Someday it will properly support dep8 testing, but that's not there today [20:42] fginther: dep8 is different [20:42] fginther: not everything works with dep8 in general [20:43] fginther: many tets require things that are only in the source tree [20:43] (I love dep-8 but that's the reality) [20:43] zyga: dep-8 can do that [20:43] dobey: oh, no it cannot, how could it [20:43] dobey: it can only install packages, it cannot see the build tree anymore [20:43] dobey: it runs after the build tree is gone [20:43] dobey: at least in debian [20:43] zyga: uhm, there is a flag to require a built source tree [20:43] dobey: you can package that stuff [20:44] zyga: oh yes he can, he get the source and run from there [20:44] zyga: where the dep-8 tests will grab the source and build it [20:44] ah, I see [20:44] I didn't know that [20:44] though I also meant that sometimes upstreams don't put stuff into their tarballs [20:44] it's just in the VCS [20:44] zyga: that's pretty much all some of the ubuntuone dep-8 tests did, as it would grab the source and build/run the tests from the tree, to ensure that any dependencies didn't break it [20:45] dobey: yeah, we're using dep-8 whenever we can [20:45] zyga: we caught many pep8 and twisted changes before they landed in distro, by doing that [20:45] dobey: it's a great thing [20:46] zyga: well, anything you need to run the tests should be in the tarballs. if it's not in the tarball, the tarball is broken, i'd say [20:46] dobey: though it started with debian, it's really useful for breaking api and changing dependencies without being careful [20:46] dobey: I know that now, I package some things where upstream doesn't realease tests in the tarball because it's big and they don't want to [20:46] dobey: all the stuff we're writing is dep-8 friendly [20:47] btw, what's the source-rebuild flag/ [20:47] I could use that for providers that don't yet have a cli tool to test after installation [20:47] that could catch job clashes when providers that share a namespace do smething nasty [20:47] Restrictions: build-needed [20:47] in debian/tests/control [20:48] ah, nice, I have to try that! [20:49] dobey, thanks for all the tips, I've learned a lot! [20:49] sure [20:50] Mirv: Ah, I see! Thanks, I'll check! [20:53] rvr: o/ :) [20:54] dobey: quick question, is there anything related to webapps that's using this system? [20:54] dobey: checkbox has a sever backend that's pretty big and totally not packaged [20:54] dobey: it's also using this system that I've described [20:54] dobey: do you know what other webapps are doing by any chance? [20:55] zyga: i think the main webapps are using CI train, but i'm not 100% sure [20:55] * Mirv feels it way past bedtime, -> sleep [20:55] they're all pretty much just .desktop files which run "webapp-container $URL" after all [20:56] dobey: I'm not talking about click webapps [20:56] dobey: think launchpad [20:56] dobey: not a click that wraps launchpad.net link [20:56] dobey: I'm talking about the actual website with message queues, databases and stuff [20:56] no, things that aren't packaged aren't using it afaik [20:57] do you know what they are using? [20:57] i don't recall what lp is using exactly [20:57] zyga: webhooks are on the list for LP git, though not fully specced out yet [20:58] i know the app store, login.ubuntu.com, stuff, and such are a tarmac though [20:58] Launchpad itself? It uses PQM and buildbot [20:58] cjwatson: yeah, I know, I'd love to be able to influence that [20:58] Very 2005 [20:58] cjwatson: I don't know if my recommendation reached you (it was done via several layers0 [20:58] zyga: We had webhooks coming from two teams in the stakeholder process, so it'll be high on the agenda once we're far enough through dependencies [20:59] zyga: There was no point having it on the agenda for this dev cycle since it's too far down the stack [20:59] cjwatson: understood [21:00] cjwatson: github hooks are different than git hooks, whatever we build must be enough to support travis.ci supporting launchpad down the road [21:00] even if we have to send patches down them to convince them (I suspect there are more things thay need apart from the hook itself) [21:01] cjwatson: not supporting travis would be a mistake, it measn that hooks are useless for the average person that dones't have their own infrastructure to hook into [21:01] * zyga talks too much on 22:01 on Friday [21:01] zyga: travis.ci was the specific example given, we understand that [21:01] travis-ci, rather [21:02] zyga: though it's not entirely within our control, it will involve negotiating with travis-ci since they'd need to add Launchpad support too [21:02] cjwatson: yeah [21:02] cjwatson: it's open source [21:02] cjwatson: so we could do that if we convince them to merge it back [21:02] sure, but who cares if it's not deployed on travis-ci, right? [21:02] cjwatson: they also need basic identity management and repo discover [21:02] I don't know their politics [21:02] cjwatson: yes, hence the convincing part [21:03] and yes I am well aware that github hooks are different from git hooks, I'm capable of googling for the spec ;) [21:03] cjwatson: I kind of wish I was on your team hacking this, this is how much I want this to happen [21:03] I know, sorry if I sound silly [21:07] Also, Launchpad itself will likely be an earlyish adopter of git, so we'll get to investigate merge gatekeeping strategies; but of course our CI will be on-site [21:09] well, earlyish adopter of git on Launchpad, obviously not of git in general ;) [21:09] cjwatson: yeah, but I'm sure the result will be very interesting [21:14] Mirv: Successful system upgrade! [21:37] cjwatson: yeah sign me up to beta test that when it's ready