=== maclin1 is now known as maclin [05:41] anyone have any ideas as to why intel driver might cause X BadMatch/BadPixmap crashes, but only in Nautilus? [06:25] good morning! [06:46] Good morning [06:51] morgen pitti! [06:52] hey larsu, how are you? [06:52] good morning desktopers [06:52] hey larsu pitti [06:52] bonjour seb128 ! [06:52] pitti: good thanks! How are you? [06:58] larsu: gut, danke! [07:03] good morning [07:04] hey didrocks ;-) [07:04] hi didrocks! [07:04] hey larsu! [07:04] re seb128 [08:33] dobey, thanks for transitioning software-center to CI landing and merges the ubuntu changes back to the upstream vcs [08:35] seb128: sure. was glad to finally get that done yesterday :) [08:37] :-) [08:45] * seb128 starts wondering if upgrading nautilus was good idea [08:46] seb128: for the icon size or something else? [08:47] I still find the move action to be confusing btw (especially if you move files to a network share) [08:47] quite some of the changes feel like regressions [08:47] upstream is reverting some of those in 3.20 so they seem to agree [08:47] like they dropped the context menu to delete and are restoring it [08:47] same for the one to create symlinks [08:48] the "in toolbar" copy dialog is less obvious [08:48] the icons are too big [08:48] the menubar patch is way more work that I was expecting [08:48] the copy icon is clearly not obvious [08:48] like larsu has been on it for almost a month, I think we could have worked on more useful things [08:48] they change in trunk to open the popup copy dialog automatically [08:48] even after knowing where it is, I do the action and always have this 0.5s "did I really start copying/moving?" [08:49] to workaround a bit the fact that it's not obvious [08:49] I'm going to backport that and see how it feels [08:49] yeah, that one (and icon size) are the most obvious issues to me [08:49] but yeah, the icons size and the copy UI are the ones that concerns me most atm [08:49] I even didn't notice about the delete removal [08:50] but I either use del or command line [08:50] same here [08:50] so not a good example :) [08:50] same for symlinks [08:50] I never use those [08:50] or at least not in nautilus [08:50] yeah, I wonder if people use this [08:50] they are more "tech" work to me [08:50] yep [08:50] and I use nautilus to manage my datas a bit [08:50] like move photos around [08:51] or clean up my downloads folder [08:51] yeah, I wonder about the delete in context menu [08:51] I never see "regular users" using the context menu [08:51] i just want my nautilus to not crash arbitrarily [08:51] dobey, does it do that in xenial for you? [08:52] seb128: haven't tried xenial. it does in trusty, i think because of intel drivers. i sometimes get BadPixmap crashes in it :( [08:54] :-( [08:54] I'm not aware of that one [09:00] morning desktoppers [09:00] seb128, how did the CC catch up go? All ok? [09:01] buf = "BadDrawable (invalid Pixmap or Window parameter)", '\000' [09:01] :( [09:02] dobey, :-/ [09:02] good morning willcooke [09:02] willcooke, hey, it went very well, thanks [09:02] willcooke, didrocks and Laney joined [09:02] didrocks gave some updates on the online dash status and the code in [09:02] and Laney discussed a bit gnome-software [09:02] super, thanks seb128, didrocks Laney [09:03] I will go find the logs actually... [09:03] otherwise they seemed happy to know that we have active contributors on GNOME and other things and not much frictions [09:03] k [09:03] yw [09:03] i don't know where the crash reports are going though :( [09:03] yo [09:03] hey Laney [09:03] happy first friday! [09:04] hmm, nautilus package has 1800 open bugs :( [09:06] Laney: happy end of first week! [09:06] https://bugs.launchpad.net/ubuntu/+source/nautilus/+bug/1071155 is similar anyway [09:06] Error: launchpad bug 1071155 not found [09:07] hey Laney, happy friday! [09:07] hrmm, 9 open bugs in nautilus for sig 5 in _XReply() [09:07] and 5 of them have bug heat > 160 [09:07] other 4 are at 10 [09:08] I think the issue is that not many people know how to deal with those bugs [09:08] they don't feel like gtk/gnome issues [09:08] they are a bit of a pain to debug, yeah [09:09] really needs someone who knows X well enough, to poke at them [09:10] not sure who's working on X for us any more though [09:11] not many people, that's part of the issue [09:11] tjaalton is but I don't think he has enough free slots to look at those [09:11] what seb128 said [09:13] :( [09:14] dobey: happens only on intel? install the xenial driver, works fine in wily [09:15] willcooke: the phone has a phone icon now when you connect it oh pretty :D [09:16] tjaalton: i'm running the wily kernel already. also it works fine on my intel laptop. it only crashes on my workstation [09:16] dobey: no, intel X driver [09:16] or try forcing uxa in xorg.conf [09:17] what generations are those? [09:17] hi [09:17] when's the wily xorg going to end up in trusty for HWE? [09:17] once libdrm is past NEW [09:17] workstation is haswell [09:17] is when the rest is getting pushed [09:18] not sure on the laptop, i think also haswell [09:18] maybe ivybridbge [09:18] ok [09:18] my desktop is ivybridge, no issues there [09:18] on wily [09:19] i don't recall if it was crashing for me when i had ivybridge in my workstation [09:19] davmor2, nice!! [09:19] what's the resolution on workstation? [09:19] 4k [09:20] that could be it [09:20] i had an nvidia card in for a week, and nautilus didn't crash on it [09:20] but had enough other problems with that, that i just returned it [09:21] so try forcing uxa, or pull the X driver from xenial [09:22] Laney, is there anything keeping gnome-calendar in main? it doesn't seem to be on the current iso yet? [09:22] I guess it's on your todolist? [09:24] I seeded it [09:24] so the Task presumably [09:25] k [09:26] * Laney runs ./update [09:27] you can't do that until after the promotion [09:27] so easy to lose it if there is a gap [09:27] that's a bit backward [09:27] because usually we do the promotion once something tries to pull the binary in [09:27] seed -> component mismatch -> promote -> ./update [09:28] but oh well, it's minor [09:28] k [10:02] mitya57, hey, thanks for updating nautilus in decembre, you didn't use the packaging vcs though? (there was commit/not released work that you seem to not have included and your version is not in there) or did you just forgot to push? [10:06] larsu, I'm working on a nautilus update, how is the menubar change coming? can I give it a test build even if it's not ready it to be proposed? just to see how it's working [10:10] seb128: I'm done with the action rewiring and am writing xml right now [10:10] k [10:10] you'll need a gtk patch too [10:10] let me open a bug for that one first [10:10] danke [10:10] is that something upstream is likely to agree on? [10:10] or an ubuntu hack? [10:10] upstream [10:11] well, depends on desrt ;) [10:11] k [10:13] hi [10:13] hi desrt!!! [10:13] morning desrt [10:14] Laney! [10:14] morning desrt :) [10:14] willcooke: g'morn [10:14] hihi larsu [10:14] and hello didrocks and seb128 :) [10:14] hey desrt [10:15] good morning desrt [10:15] desrt: since you're here - care to look at https://bugzilla.gnome.org/show_bug.cgi?id=760312 ? [10:15] Gnome bug 760312 in .General "gtkapplication: export windows directly as action groups" [Normal,New] [10:16] bah [10:16] seb128: that's the patch I was talking about ^ [10:16] bzr bd-do doesn't like lp:~ubuntu-desktop/nautilus/ubuntu it seems [10:16] UnicodeEncodeError [10:16] grrrr at python [10:16] larsu, thanks [10:16] the usual :) [10:17] larsu: 'hi' was a subtle way of saying "i hope i get a bug link" :) [10:17] desrt: oh. I read it as "hi" [10:17] hey desrt, how are you? (and where?) [10:19] larsu: and yet i got my bug link =) [10:19] desrt: totally unrelated :P [10:19] pitti: good morning :D [10:19] pitti: i am before coffee, so i have difficulty answering "how are you" :) [10:20] and i am in toronto, where it is still very dark indeed [10:20] isn't it 05:20? [10:21] that is correct. [10:21] * Laney screams [10:21] larsu: are you sure this patch correctly handles the remoteactiongroup/platform_data stuff? [10:22] desrt: why would that have anything to do with it? [10:22] because GtkApplicationWindow doesn't wire that through, iirc [10:22] i'll do the review in the bug... [10:22] ugh, are you sure? [10:23] fairly sure, yes. [10:25] * desrt hopes that 5am pre-coffee reviewing doesn't come back to bite her [10:26] desrt: UNITY! [10:26] (thanks for the review) [10:26] "needs word" is certainly less dangerous than "accepted-commit_now" [10:29] heh [10:29] desrt: the gist: I want to export additional actions from inside a subclass of GtkApplicationWindow [10:30] amusing story of the day: (due to various online posts drawing equivalences between them) google translate is now confusing "russian federation" with "mordor", "russians" with "occupiers" and "sergey lavrov" with "sad little horse" [10:30] larsu: why can't the subclass simply add the actions like everyone else does? [10:32] desrt: that sounds horribly inefficient... [10:32] what kind of actions are we talking here, and how many? [10:32] but I'd rather do that than deal with the platform data mess [10:32] desrt: normal amount, nothing fancy [10:33] use action entries? [10:33] reimplementing the actiongroup infra is annoying and awkward and you're probably going to get it wrong [10:33] (for reasons like platform_data...) [10:39] desrt: hm? The actions already exist [10:39] in some widget deeper down [10:39] I don't like that reimplementing the interface doesn't work [10:40] seems like something that totally should work [10:41] we could possibly detect if you also reimplement RemoteActionGroup for yourself... [10:41] or we could implement it directly on the window itself, as well [10:41] that almost makes sense, in fact [10:41] yeah.... clearly [10:42] * larsu shall not shave this yak now though to not make seb128 more impatient ;) [10:42] desrt: thanks for catching this anyway [10:42] if you want to do the remoteactiongroup impl on the window itself and do a patch for that, i'd be happy to look at it... provided you give a better answer on the "why can't you just use actions" question [10:43] because i agree that it would be theoretically nicer [10:43] same argument could be made for GApplication itself [10:43] why didn't you do it this way in the first place i wonder? [10:44] it used to be possible to provide your own action group [10:44] maybe still is, although very much deprecated, and for a long time [10:44] it was the early days before we figured out what was really a good pattern here [10:44] ah, right [10:44] and before GActionMap existed [11:02] morning all [11:05] hey andyrock [11:19] Trevinho|OFF, [11:19] andyrock, [11:19] I have a question [11:20] yup [11:20] without any of my code [11:20] the old shadows update all the time [11:20] it's like a busy loop [11:21] do you know that it was intentional or there's something that marks regions as dirty etc? [11:21] i mean [11:21] that shouldn't happen anyway [11:21] i don't think it's intentional [11:21] but do you know why it happens? [11:21] maybe it's a workaround for something [11:22] to be honest nope [11:22] mmm then I have to fix this first, otherwise unity will calculate the shape (and perform convolution) at almost every frame with the new shadows... [12:05] andyrock, Trevinho|OFF ignore :) it was a program I left open that was causing the frequent updates! === Chipaca` is now known as Chipaca [12:19] pitti: probably a very stupid question, but how can I trigger an autopkgtests using the test request format? [12:21] didrocks: ah, ATM you can't, I'm afraid; firewall limits that to snakefruit (or do you have access there?) [12:21] pitti: I do have access here, ~ubuntu-archive [12:21] didrocks: ah [12:21] but I don't see anything handy in run-autopkgtest --help [12:21] didrocks: now, I didn't yet expose those in run-autopkgtest [12:21] ah, I'm not that crazy :) [12:22] amqp-publish -u amqp://autopkgtest-worker:SECRET@10.25.180.81 -r debci-trusty-amd64 -b 'ubuntu-make-master {"test-git": "https://github.com/ubuntu/ubuntu-make.git", "ppas": ["ubuntu-desktop/ubuntu-make-builddeps", "ubuntu-desktop/ubuntu-make"]}' [12:22] didrocks: that's what I used for testing back then [12:22] didrocks: shoudl be straightforward to add it to run-autopkgtest, similar to --ppa [12:23] SECRET will be replaced by the right thing in amqp-publish? [12:23] didrocks: if you want to have a go at it, please do; otherwise I'll try to do that on Monday [12:23] didrocks: no, I just edited it out for pasting into public IRC :) [12:23] pitti: well, I can have a look, but let me ensure it can runs properly first :) [12:23] ok ;) [12:23] run* [12:23] pitti: also, that I can play with the env variables and such, let's see [12:24] between that and https://wiki.ubuntu.com/ProposedMigration/AutopkgtestInfrastructure#Test_request_format, should be workable :) [12:24] didrocks: for experiments it would be good if you could skip armhf for now, as that's horribly swamped [12:24] didrocks: go nuts on x86, they have plenty of capacity [12:24] . o O { me is *sooo* happy to be able to say that now, since Tuesday ☺ } [12:25] haha! ok :) I just need one arch anyway for testing this [12:29] * didrocks looks fot amqp-publish [12:29] for* [12:31] pitti: amqp-tools isn't installed on snakefruit, though? === alan_g is now known as alan_g|lunch === hikiko is now known as hikiko|ln [13:18] didrocks: no, it's not; I ran that from my adt control instance [13:18] didrocks: I can run stuff for you if you wwant [13:18] pitti: well, it was easier for me to implement the options, I did this :p [13:19] I saw the job running in the web ui, but then, it seems to fail [13:19] didrocks: ah, just got a worker failure mail [13:19] the issue is that I can't find the logs back [13:19] W: Failed to fetch http://ppa.launchpad.net/ubuntu-desktop/ubuntu-make-builddeps/ubuntu/dists/xenial/main/source/Sources 404 Not [13:19] +Found [13:19] didrocks: right, sorry; this counts as a "tmpfail", i. e. it thinks that this is a temporary glitch [13:19] yeah, I think if test-git or test-bzr are provided, we shouldn't force a ppa or trigger [13:19] didrocks: so it retries instead of posting the log [13:20] didrocks: you don't need a trigger, but you need a PPA [13:20] no, it's a real failure, I don't have anything in the ppa for the current devel serie [13:20] let's please not clutter the ubuntu result container for these [13:20] so the archive doesn't exit [13:20] exist* [13:20] ah, makes sense [13:20] hum, should I do a dummy upload in the ppa then? [13:21] didrocks: usual workaround is to copy (or upload) a package into it and delete it again [13:21] yeah [13:21] didrocks: upload, or copy-package with binaries [13:21] can do this [13:21] anyway, I tried on trusty then [13:21] with the ppas [13:21] and same, seems to have failed quite quickly [13:21] (the 2 archives exists here) [13:21] but I don't see anything in the ui [13:21] adt-run [13:19:34]: @@@@@@@@@@@@@@@@@@@@ apt-source ubuntu-make-master [13:21] E: Unable to find a source package for ubuntu-make-master [13:22] making sense :) [13:22] this looks like this didn't get the git argument [13:22] but the doc says it won't try to use the "pkg" as a "pkg" [13:22] ah [13:22] * didrocks check for typos [13:22] * pitti looks at the full log [13:24] Received request for package ubuntu-make-master on trusty/amd64; params: {u'ppas': [u'ubuntu-d [13:24] esktop/ubuntu-make-builddeps', u'ubuntu-desktop/ubuntu-make'], u'test_git': u'https://github.com/ubuntu/ubuntu-make.git'} [13:24] didrocks: s/test_git/test-git/ [13:24] did I typo that on the wiki? [13:24] I guess you did [13:25] no, it's ok there [13:25] waow [13:25] I can't read [13:25] indeed, I reopened the page and it's there [13:25] well, to be fair these pretty well look the same :) [13:25] * pitti hugs didrocks [13:26] * didrocks hugs pitti back [13:26] ok, so running on trusty [13:26] let's see [13:27] didrocks: I killed all adt-runs for u-make for now, to break the infinite retry loop [13:27] seems for this kind of operation this needs to become more clever [13:30] urgh, sorry for the retry loop on this then :) [13:30] did you kill last one? [13:30] * didrocks looks [13:30] no, so looking at http://autopkgtest.ubuntu.com/running.shtml#pkg-ubuntu-make-master :) [13:31] didrocks: no need to be sorry; it's my ongoing quest for minimizing human activity in a system where everything can (and does) fail all the time [13:31] didrocks: killed that too [13:31] pitti: oh? [13:31] that one wasn't with the correct parameters? [13:31] (my last retry after the fix) [13:32] didrocks: oh sorry, I thought you said "this one is also old" [13:32] didrocks: as it already ran for 11 mins [13:32] pitti: well, it's installing ubuntu-desktop! :p [13:32] * didrocks reruns [13:32] ok, running again [13:33] pitti: so, then, I should see it stored in http://autopkgtest.ubuntu.com/packages/, right? [13:33] (after the first run) [13:33] http://paste.ubuntu.com/14437805/ .. totally trivial command line, eh! [13:33] didrocks: no, there's no web UI frontend for PPAs right now [13:34] pitti: ahah, and that's without the --env parameter still! :p [13:34] didrocks: if you run it against the ubuntu archive, it'll appear there, but not for PPAs; so far we haven't needed one, but we could set up something [13:34] ok [13:34] didrocks: no problem -- just start another one with env and let them run in parallel [13:34] pitti: oh right, the *cloud* [13:34] * didrocks tries [13:35] (done) [13:35] this one should only run a pep8 test [13:35] didrocks: it's fine if you run 10 in parallel occasionally; if you need much more, and it starts blocking things, we need to ask for moar powah [13:36] indeed, it will still takes time though until we have real ubuntu-desktop images [13:36] pitti: I don't remember, the ppas artefacts archive in swift will be per ppa per release, right? [13:37] (so we can remove the vivid one at some point, right?) [13:37] didrocks: correct [13:37] sounds good :) [13:37] didrocks: oh, you don't have "ubuntu-desktop" in d/t/control Depends:, but call apt-get yourself? [13:37] didrocks: do you call it with $(which eatmydata 2>/dev/null) apt-get ? [13:38] didrocks: if not, it'll be awfully slow (test Depends: do that automatically) [13:38] pitti: ah, I don't, I guess I should. There was a reason I minimize the deps in d/t/control (that was before the holidays…) [13:38] pitti: ok, adding eatmydata [13:38] ah yeah, the lines with test deps was getting really long [13:38] and I have this "setup" tests [13:38] didrocks: testbeds have that magic dpkg force-unsafe-io flag in the config, but it doesn't seem to be nearly as fast as eatmydata [13:39] and I need to have all them listing the same deps, to reuse the same test bed [13:39] that's why I put the dep setup logic in setup (in addition to the other setup parts) [13:39] WARNING: apt does not have a stable CLI interface yet. Use with caution in scripts. [13:39] 0 upgraded, 1254 newly installed, 0 to remove and 0 not upgraded. [13:39] Need to get 534 MB of archives. [13:39] yeah, I'm using simple apt, which is fine in trusty and xenial [13:39] After this operation, 2015 MB of additional disk space will be used. [13:39] pas mal :) [13:39] ahah :) [13:40] if you want to kill those 2 runs because they will be slow due to no eatmydata, please feel free [13:40] didrocks: I don't care [13:40] I'm adding that to master meanwhile [13:40] ok :) [13:40] let's just let it finish [13:40] didrocks: queues are empty right now anyway (except arm) [13:40] yeah [13:40] pitti: btw, if you are interested in the structure: https://github.com/ubuntu/ubuntu-make/tree/master/debian/tests [13:41] you basically have: pep8/small/medium/large [13:41] which are the real tests [13:41] (you can pass TESTS="small large" for instance) [13:41] no parameter is what we used to have "pep8 small" [13:41] collect-coverage is for collecting and archiving global coverage report [13:42] custom is used when you set one or more specific tests [13:42] like TESTS="tests/small/test_foo.py:Bar.baz tests/…" [13:42] and TESTS=all runs everything but custom ofc === Wellark_ is now known as Wellark [13:43] (well, it runs custom which is skipped) [13:43] didrocks: ah, https://github.com/ubuntu/ubuntu-make/blob/master/debian/tests/control is clever -- so you have a separate "setup" job and re-use the testbed for everything [13:44] didrocks: btw, there's no need to repeat all the stanzas [13:44] didrocks: as long as the depends:, restrictions: etc. are identical, you can just do [13:44] Tests: foo bar baz [13:44] oh, even better then! [13:44] * didrocks modifies [13:45] didrocks: wouldn't it be easier to run https://github.com/ubuntu/ubuntu-make/blob/master/debian/tests/setup-testbed#L17 with "needs-root"? [13:46] didrocks: calling sudo in tests is a bit meh (if you do it, please call "sudo -n", to avoid eternal hangs because of password questions) [13:46] didrocks: you can do su -c "commands here" $ADT_NORMAL_USER [13:46] ah, there is a variable for this :) [13:46] I was going to ask how to not hardcode "ubuntu" for user name [13:47] (I still have some commands I don't want as root) [13:47] meh, apt-get install has some trouble in the test [13:47] argh? [13:47] * didrocks looks === hikiko|ln is now known as hikiko [13:48] didrocks: you can't :) (I just looked at the tail -f of the log) [13:48] http://paste.ubuntu.com/14437883/ [13:48] ah ok, so it's not the one running I see (I only see one btw? I maybe typoed the env variable…) [13:49] didrocks: that's the old one without --env, yes (no "env" in the command line) [13:49] hum, I didn't get this kind of no tty issue on my xenial vm at least [13:49] DEBIAN_FRONTEND=noninteractive ? [13:49] so, debconf doesn't really like it [13:49] yeah, I didn't setup anything for this though here [13:50] I wonder if a default changed [13:50] although wait, this is already set in tests [13:50] but this doesn't make it through sudo [13:50] didrocks: that test run finished now (due to apt-get error), should be in swift now [13:51] didrocks: which PPA was that against? [13:51] pitti: ubuntu-desktop/ubuntu-make [13:51] (as I takes last ppa from the list IIRC to match against) [13:51] https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-xenial-ubuntu-desktop-ubuntu-make?format=plain [13:52] sorry, not xenial, trusty [13:53] didrocks: ah, this one looks happier, but that didn't install -desktop: https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-trusty-ubuntu-desktop-ubuntu-make/trusty/amd64/u/ubuntu-make-master/20160108_134929@/log.gz [13:54] ah, they are not stored [13:54] * didrocks was stupidely looking at the first one [13:55] so, I can ofc now change the Depends: line [13:55] as I have everything in one line [13:56] didrocks: you can line-break test control just the same way as debian/control, FYI [13:56] but that would mean that I don't add needs_root for setup (but I can add -n to sudo) [13:56] ah nice :) [13:56] didrocks: you still need it to configure autologin and stuff though? [13:56] yeah, and also the setup does some extra stuff === alan_g|lunch is now known as alan_g [13:59] oh, actually, I didn't really try to setup the testbed that way [13:59] hah, didrocks is serial-killing workers again :) [13:59]