=== Ursinha is now known as Ursinha-afk [03:45] Good morning [05:33] good morning [05:36] hum, biab [05:59] bonjour didrocks [05:59] didrocks: so what's that "BaseException: foo bar" thing that apparenlty breaks ubuntu-developer-tools-center? [06:01] pitti: it's part of a test and expected (I don't remove stderr, but it's just a print) [06:01] morning pitti btw ;) [06:01] ah right [06:01] so it' sjust [06:01] adt-run [15:39:04]: test all: - - - - - - - - - - results - - - - - - - - - - [06:01] all FAIL stderr: 2014-08-27 15:38:35,645 [nose.plugins.manager] DEBUG: DefaultPluginManager load plugin cov = nose_cov:Cov [06:01] (i. e. the whole two metric tons of debug log) [06:02] argh [06:02] didrocks: alors, tu veux ajouter "allow-stderr" [06:02] or see if I can remove debug infos frmo nose-cov itself [06:03] or yeah, just adding your tag [06:03] on teardown of my tests, I'm ensuring I don't have any warning/error on stderr [06:03] (for each tests) [06:03] or it will raise in error [06:04] that's nicer for catching unexpected warnings, indeed [06:04] so, I guess that's good enough, I control my own stderr from my tests [06:04] that's also why the "fail on stderr" is still useful by default [06:04] yeah, but as it's covered in my tests for me… [06:04] I have also a flag that I turn in some tests [06:04] which do the contrary [06:04] like "I know I'll have a warning/error here, raise if there is none" [06:05] ok, so, let me add this tag [06:05] I fixed the coverage wrong report issue as well [06:05] so now, small tests covers 77% [06:05] and I'm still at 99% with the large ones [06:07] didrocks: wow [06:08] didrocks: in things like umockdev I find it excessively hard to get above 80%, as the remaining code paths are error handling for obscure conditions which are not practially reproducible in tests.. [06:08] so, congrats! [06:08] pitti: that's why large tests are integration tests (small are units/modules) [06:09] yeah, but rare corner cases are even harder to hit with integration tests than with unit tests [06:09] like, malloc() failures, or socket() failing, etc. [06:09] and medium are large tests without network and fake data (they inherit from large tests) + as, I'm in control of the mocks, I play with sending download not matching md5sum for instance :p [06:09] yeah === jpds_ is now known as jpds [06:15] pitti: ok, pushed now! Thanks for the help, I'll know where to look at now :) [06:15] pitti: so, as soon as you have a minute for enabling the medium tests… === jpds is now known as Guest68298 [06:15] didrocks: that should have failed the same way in a local run, though? didn't it? === Guest68298 is now known as jpds_ [06:16] didrocks: ah, that's the one which needs webternet access? [06:16] pitti: I'm pretty sure it didn't, (RET code was 0) [06:16] I can check again if needed [06:16] pitti: right, so medium only need to download a docker image from the docker hub [06:17] large, if enabled, would need a fake X server (I guess xvfb can work) + webternet access [06:19] didrocks: ok, you can sort out adding xvfb-run (you can test that locally), I'll look into wiring the http_prxy [06:19] I'll use the gem2deb autopkgtest for that, it fails for the same reason [06:20] pitti: yeah, I'll try with the medium tests first anyway, not sure I want to wire the large tests in that stage, or will do only in my daily tests run in another vm [06:20] pitti: ok, you're right, exit code was 4 btw without the allow-stderr [06:20] locally [06:21] I'll blame jetlag in China when I did this [06:21] (I guess the OK at the end derouted me) [06:43] good morning desktopers [06:44] bonjour seb128 [06:44] pitti, salut, wie gehts? [06:44] seb128: quite fine, thanks! [06:45] seb128: we made an interesting observation today [06:45] seb128: messaging-app has 31 strings upstream, and 65 strings in ubuntu [06:45] is "today" started enough to be able to do observations? ;-) [06:45] at first I thought it was an error in message sharing/imports, but it's actually correct [06:45] pitti, outdated upstream template? [06:45] seb128: right [06:45] that makes sense [06:45] seb128: IOW, it seems ubuntu's are always current (after landing, anyway) [06:46] so it's possible/better to translate the ubuntu one than the upstream one, as that's notoriously outdated [06:46] hum [06:46] I directly pushed a POT update to messaging-app, but it affects others, too [06:46] well [06:46] the issue is clicks [06:46] they don't use langpacks [06:46] yes, of course [06:46] we need to ship back the translations to trunk [06:46] but still, it helps somewhat for the debs [06:46] right [06:46] but those are supposed to be the minority [06:47] well, "supposed", but they aren't yet [06:48] seb128: anyway, it seems the imports worked well, and message sharing works [06:48] great [06:48] seb128: and FTR, yes, it was early enough :) 2014-08-28 05:43: arrived [06:49] so upstream got updated translations? [06:49] bah, you early bird ;-) [06:49] seb128: yes, I pushed the pot, and now upstream has the same ones [06:49] seb128: I noticed some broken strings, I sent an MP [06:49] https://code.launchpad.net/~pitti/messaging-app/i18n-fixes/+merge/232507 [06:49] (still needs to be reviewed) [06:49] https://translations.launchpad.net/messaging-app [06:49] 46 untranslated in french, what is our team doing! [06:49] right [06:49] strings are changing like mad [06:50] and the POT hasn't been updated in ages [06:50] and some strings like the attachment ones can't be translated as they are buggy [06:50] (see above MP) [06:50] right [06:50] I wish we had a proper solution/option to commit pot updates to trunk [06:51] ideally launchpad should have recipies or integration to refresh/commit those [06:51] well, you'd need to actually be able to update them without installing a gazillion build deps and building [06:51] then it could be done on landing [06:54] sweet! https://jenkins.qa.ubuntu.com/job/utopic-adt-ubuntu-developer-tools-center/2 [06:54] didrocks: c'est vert ! [06:55] pitti: j'aime le vert :) [06:55] didrocks: moi aussi ! [06:55] pitti, well, usually the pot update is basically running xgettext [06:55] c'est la plus bonne coleur [06:55] couleur [06:56] (je suis sûr que c'est ^ encore faux ☺ ) [06:57] belle [06:57] if that's what you mean [06:57] ou "la meilleur" [06:57] we don't use "plus bon" [06:57] well, we do, but not in the "+" meaning ;-) [06:59] seb128: ah, I thought "plus adv" -> "more adv", "la plus adv" -> "the most adv", but yeah, "good" is always special [06:59] good/better/best, gut/besser/bester [06:59] yeah [06:59] it's like "more good" in english [06:59] seb128: so it's mieux ("plus bon") and meilleur ("la plus bon")? [06:59] sounds wrong [07:00] yes [07:00] seb128: but it would be "cette pomme est plus grand que l'autre pomme", and "c'est la plus pomme"? [07:00] "c'est la plus grande pomme"? [07:00] grande [07:00] but yes ;-) [07:01] oui; merci ! [07:01] de rien ! [07:24] Is bug 1362086 a known issue in current daily build? [07:24] Launchpad bug 1362086 in ubuntu "install the system, start the os, some windows flashing" [Undecided,New] https://launchpad.net/bugs/1362086 [07:24] bug 1362086 in Ubuntu "install the system, start the os, some windows flashing" [Undecided,New] https://launchpad.net/bugs/1362086 [07:29] Morning all [07:29] Hi TheMuso [07:29] hey ypwong [07:29] hey willcooke [07:29] not a known issue [07:29] seb128, morning [07:30] seb128, another bug 1356825 saying compiz is not running [07:30] Launchpad bug 1356825 in ubuntukylin "Compiz doesn't run after log in system" [Undecided,New] https://launchpad.net/bugs/1356825 [07:30] bug 1356825 in Ubuntu Kylin "Compiz doesn't run after log in system" [Undecided,New] https://launchpad.net/bugs/1356825 [07:30] Hey willcooke. [07:31] ypwong, look at .cache/upstart/gnome-session-Unity.log? [07:32] morning willcooke [07:32] ypwong, is that a liveCD iso? or an installed system? [07:32] hey TheMuso [07:32] seb128, not sure, it's tested by NUDT QA [07:33] Hey folks. [07:34] evening TheMuso :) [07:37] didrocks: don't forget the fcitx MIR [07:37] :) [07:38] * happyaron off of the day, 0:37AM... [07:38] happyaron, goodnight [07:38] happyaron: don't worry, told you I couldn't yesterday with the backlog, but it's on my list [07:39] ok [07:39] ypwong: thanks, :) [07:49] happyaron, where are you atm? [07:49] happyaron, do you want to meet today? [07:49] willcooke: I'm at Debconf, can't make it I think... [07:49] ahhhh - yes of course. [07:49] My brain is like swiss cheese at the moment [07:49] ok, happyaron I'll email you some stuff [07:49] happyaron, have fun! [07:50] thanks! [07:51] didrocks: ok, I adjusted autopkgtest's VM build script for proxy handling and tested gem2deb manually in the DC; now deploying this [07:52] pitti: excellent! [07:53] didrocks, is the PPA for dev desktop in your PPA? [07:53] willcooke: for trusty yeah (ppa:didrocks/ubuntu-developer-tools-center), for utopic, just use distro [07:53] didrocks, thx [07:53] install ubuntu-developer-tools-center should be enough [07:53] yw [07:54] (the command is udtc) [07:55] didrocks: the command name obviously should have started with "udev-"! [07:55] * pitti apologizes; you can tell I'm waiting for test runs to finish, which don't take long enough to do something useful in between :) [07:56] pitti: I can see in particular that you still endure pain with the halsectomy and all this udisk/udev/ubrokenbrain things :p [07:56] udev is fine :) [08:04] didrocks, jibel: *screw* *hammer* *lube* *wrench* et voilà -- your new interweb tube for autopkgtests! [08:04] https://jenkins.qa.ubuntu.com/job/utopic-adt-gem2deb/? [08:05] interweb tube! *giggle* [08:06] pitti: \o/ [08:07] pitti: no metadata to add to enable the proxy? [08:08] didrocks: there is, the proxy needs to be set in the cloud-init metadata [08:08] didrocks: but that's done automatically now [08:08] didrocks: so the tests themselves don't need to do anything [08:08] excellent :) [08:08] I need to build a new vm locally, to try? [08:08] with cloud-init? [08:08] didrocks: no [08:08] I just add the tests then and cross fingers without testing locally? ;) [08:08] didrocks: that's just for the DC as it is firewalled; your local VMs (presumably) have unrestricted webternet access [08:08] ok [08:09] didrocks: no, please do test locally first [08:09] if they don't pass locally, they won't pass in the DC either [08:09] sure [08:09] I'm fixing the medium tests as we speak, I understand why it's failing, but not why it changed [08:10] didrocks: btw, now that the test succeeded once, any future failure will block the package [08:10] (we block on regressions, but not on "always failed") [08:10] yeah === hikiko-lpt is now known as hikikoo === hikikoo is now known as hikiko [08:19] Issued certificate has expired. [08:19] snif :p [08:31] moin! [08:33] hey Sweetsha1k [08:33] wrong name :) [08:34] didrocks, JackYu will have those translations done today he thinks [08:35] hey Sweetsha1k [08:35] willcooke: yeah, I received an email as well, thanks a lot! === Sweetsha1k is now known as Sweetshark [08:38] * Sweetshark feels better. [09:19] didrocks: I'll retry the failed udtc run on i386 [09:19] pitti: I was wondering if I can do that myself? [09:19] self.assertTrue(len(map_result) == 1) [09:19] argh! [09:19] didrocks: yes, if you have VPN [09:19] * didrocks changes silently for an equals :p [09:20] didrocks: I find this useful: [09:20] self.assertEqual(len(map_result), 1, str(map_result)) [09:20] didrocks: then you see the actual value on failure [09:20] pitti: oh nice, didn't know that one, will update in multiple places then, thanks! [09:20] unless you can assert the string contents of course, then assertEquals(map_result, ['foo']) is even better [09:20] I can't assert the content [09:21] didrocks: yes, assert*() all take an optional last argument which is shown on failure; usually a string [09:21] like assertEqual(a, b, 'got %s, my debug variable x = %s' % (b, x)) [09:21] ok, let me update that, while I'm testing the fixed medium tests with new certificate in adt [09:22] didrocks: do you see http://d-jenkins.ubuntu-ci:8080/job/utopic-adt-ubuntu-developer-tools-center/ ? [09:22] didrocks: i. e. do you have VPN set up? [09:22] pitti: sure, I do, the whole CI was based on it :) [09:22] didrocks: if so, you can log in and click on "Build now" [09:22] pitti: yeah, doing that meanwhile [09:23] thanks! [09:23] didrocks: ah, and now of couse udtc holds back python-apt :) (but nevermind, we are in freeze anyway) [09:23] ahah, sorry ;) [09:23] (relaunched) [09:23] #4 running, Gestartet durch Benutzer Didier Roche [09:23] \o/ [09:24] * pitti ^5s didrocks [09:24] didrocks: don't use the "matrix reloaded" btw, it confuses things [09:24] * didrocks high five back pitti [09:24] pitti: yeah, we had issues with it as well in CI [09:25] ok, I had assertEqual for all the others [09:25] let me add the arg though [09:31] pitti: ok, I'll need to dig in medium tests why it can't start a docker container there. Ideally, I will be able to find how to drop myself inside the chroot === alan_g is now known as alan_g|afk === alan_g|afk is now known as alan_g [09:56] pitti: 2014/08/28 09:56:24 Post http:///var/run/docker.sock/v1.12/containers/create: dial unix /var/run/docker.sock: permission denied [09:56] pitti: do I need to relax any constraint? (I guess maybe the allow-root?) [09:57] the docker service (which needs more perms) is activated by the docker command (which doesn't need root) [10:01] didrocks: if you need to be root to access docker, then yes, "needs-root" [10:01] pitti: but it doesn't log me as root, right? [10:01] didrocks: you can run those in a separate Tests:, so the others run as user [10:01] (as I have tests where I don't want that) [10:01] hum [10:01] and it's better to have finer-grained tests anyway [10:01] yeah, it will just change my coverage report logic :p [10:02] didrocks: it runs the test script as root with "needs-root"; otherwise, a user (usually "ubuntu", but can be "phablet", or something else -- that depends on the testbed) [10:02] ok, let's try the medium tests with that separately first [10:03] pitti: the command itself doesn't need to be root, it's the activated service that needs that, let's see if the tag will allow that [10:03] didrocks: activated by what? [10:03] didrocks: if it's a system service which is running, then your test doesn't need root [10:03] didrocks: if you want to access the socket yourself, then you need to, of course [10:06] hum, /me checks the docker documentation [10:06] ah, actually, I think it's just that the "ubuntu" user needs to be in the docker group [10:06] but then, it needs to refresh the groups list [10:07] didrocks: hm, that makes it complicated indeed [10:07] yeah, need to add manually, and the launch bash? [10:07] you'd need to start as root, detect some user (uid >= 500), add that to the group, then call the tests through su "testsuite" $user [10:08] yeah, not sounds hackish… [10:16] willcooke, around? [10:18] hey tkamppeter === jpds_ is now known as jpds === MacSlow is now known as MacSlow|lunch [11:51] pitti: ok, I'm quite unsure, I'll be able to get the medium tests running soon inside autopkgtests though [11:51] I can download the container, starts it [11:52] but for weird reason, I can't ssh to it, like if the user (that has harcoded password for this temporary session) wouldn't let me log in [11:52] and same container (checked the ID) would let me log in outside the autopkg chroot, weird [11:52] * didrocks starts a second container manually inside the chroot to check users [11:53] ah, it seems that stdout is blocked, if I cat anything… [11:54] and so then, I get no prompt back [11:54] (I guess someone in the autopkg setup) [11:55] ahah, it all comes down to that: su: System error [11:56] hence ssh not answering [12:02] 52 results only on google [12:02] that's the cost of trying something really new I guess ;) [12:02] * didrocks tries to build an image with utopic docker's version, just in case there is an incompatibility === alan_g is now known as alan_g|lunch [12:18] didrocks: yeah, I don't remember any test that involves docker so far [12:19] pitti: I'm trying 2 things: building a new trusty image with the new docker or trying utopic docker image on utopic [12:19] but building an image is taking ~30min, so I switched back to the MIRs meanwhile [12:19] * pitti hands didrocks the "Pioneer!" shirtr [12:19] s/r$// [12:20] pitti: ahah :) on the back, is it written "suffering"? [12:20] didrocks: aren't these synonyms? [12:20] I guess you're right ;) [12:20] To boldly test where noone has tested before! [12:20] heh [12:20] speaking of pain! no space left inside the chroot [12:20] * didrocks <- sad face [12:21] didrocks: reminds me of the first time we tried to run autopilot tests in autopkgtests [12:21] yeah, I guess that was "fun" [12:21] Sweetshark: of course, it stopped on libreoffice! all your fault [12:21] after 3 days or so we finally figured it out; I spent a whole afternoon with jibel trying to track down a 30 s timeout which eventually turned out to be a bamf issue [12:22] argh "nice" [12:22] didrocks: uh, you're building a biiig image there .. why not start with something smaller? [12:22] pitti: well, that's what I used for my medium tests, I'm building the desktop image [12:23] to "simulate" the real env udtc in running in [12:24] I can strip down for this test at least [12:27] pitti: oh, what is doing wrong bamf? [12:28] pitti: yeah, there is really something inside the chroot: PAM: System error [12:28] /usr/bin/chfn -f user returns 1 [12:28] (inside docker) [12:31] Trevinho: I think back then the issue was that autopilot talked to bamf without declaring a dependency, so we didn't install it in the test env [12:31] Trevinho: so it just dbus-timed out for 25 s on each test, which blew up the test time from the usual 10 seconds to like 20 minutes [12:31] but it was not at all easy to see that :) [12:32] ouch :P [12:32] and then some fun setting up enough of Xvfb, dbus-server, some Xvfb options etc. to make everything actually work [12:32] yeah, bamf is a subtle dependency... [12:32] then autopilot-sandbox-run was born to do all that automagically :) [12:33] probably that bamf module should be moved away from new AP, and just kept for the desktop version [12:33] cool === MacSlow|lunch is now known as MacSlow === Ursinha-afk is now known as Ursinha === alan_g|lunch is now known as alan_g [13:07] didrocks: what stopped on libreoffice? [13:08] didrocks: nobody expects to stop on libreoffice! [13:08] Sweetshark: apart if your disk is full! ;) [13:08] lo [13:08] lol === Ursinha is now known as Ursinha-afk [13:10] didrocks: see the text on the back of the t-shirt: http://vmiklos.hu/blog/so-many-bugs.html [13:11] ahah :) [13:11] haha, nice one === Ursinha-afk is now known as Ursinha === jmunsch is now known as Guest11861 === gatox is now known as gatox_lunch [14:37] ok, i18n translations merged back in udtc done (and udtc 0.0.4 uploaded), now, time for a very late run :) [14:41] didrocks: it's not that late… [14:41] enjoy anyway ;) [14:41] larsu: no break for 9h, I count that late :) [14:41] thanks ;) [14:42] didrocks: man, take more breaks [14:42] but you know this "just that and I'll be done"… :p [14:43] and it's just to reveal more issues, of course :) [14:43] yeah I know the feeling === m_conley_away is now known as m_conley === gatox_lunch is now known as gatox [15:26] willcooke: ping [15:27] happyaron, hey - in a meeting, but can type === tiheum_ is now known as tiheum === alan_g is now known as alan_g|EOD [17:15] happyaron: I'll finish the MIR 2nd pass tomorrow morning [17:15] * didrocks EOD, see you tomorrow guys! [17:16] great, see you, :) [17:19] kenvandine, hey [17:19] kenvandine, would it be annoying if I ask to rename that DialpadSounds property? [17:19] like DialpadSoundsEnabled [17:19] or something that makes it clear from reading it that it's a on/off bool [17:19] and not a sound name [17:22] MakeDialpadSounds [17:22] bad day bad day [17:22] ? [17:23] relatively personal issue.. [17:23] seb128, fine with me [17:24] kenvandine, ok, please do it then ;-) [17:24] seb128, and i still want feedback from boiko [17:24] k [17:35] seb128, renamed [17:36] kenvandine, thanks, approved === shiznix_ is now known as shiznix === m_conley is now known as m_conley_away