[07:21] good morning [07:41] morning. good news, I can go ahead with Qt 5.4.0 [07:41] you could help by getting https://launchpad.net/ubuntu/+source/plasma-framework/5.7.0-0ubuntu1 to release pocket so I can rebuild it successfully. [07:42] but I guess you're on it anyhow, getting all the bits migrated [07:43] oh, kio claims a regression :( http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#kio [07:43] but since it's in itself, maybe it's a wrong alarm you could force [07:45] but since I'm blocked by that anyway, I can do the final rebuilds on a bit more relaxed schedule, not pushing everything at once [07:47] I'd love to release Qt today, but maybe it's rather Monday [07:47] Mirv: sgclark was working through that excuses file earlier [07:48] and I'm sure will be working on it today as well [07:50] valorie: great! luckily there a few days until FF still. [07:51] Mirv: cool, waiting for QT 5.4 :) [07:51] valorie: are you on Vivid maybe ? [08:01] soee: not yet [08:01] I was going to jump, ran out of time, and now I'm sort of glad I waited [08:02] I only have this one laptop now [08:02] oh, ok :) [08:58] git.debian is IO blocked again >.< [09:05] sitter: Hi, do you have a pointer for Kubuntu Jenkins setup that you could point me at please ? [09:05] sitter: Doh! Ignore me me [09:06] I see the link in the CI channel now, [09:09] Sick_Rimmit: what kind of pointer are you looking for? [09:09] sitter: I wanted to find out more about how it's setup and configured for us [09:10] The CI Docs page give me the PPA's [09:10] Which I will setup a testVM for the daily build stuff [09:11] But I wanted to know more about the integration between Jenkins and our repo's how the tests are configured and stuff.. [09:11] there's no documentation for that [09:12] kci uses jenkins + LXC + launchpad to orcehstrate builds. https://github.com/apachelogger/pangea-tooling contains the tools that are used outside the LXC while http://anonscm.debian.org/cgit/pkg-kde/ci-tooling.git contains the tools run inside the LXC [09:12] sitter: Ah ha, OK then. that'll be why I can't find it :0 [09:13] update-projects.rb in former creates all jobs (all jobs' templates are in the jenkins-jobs directory) [09:13] sitter: OK those page are where I will start thanks [09:14] other than that one script pretty much all the things are really just somewhat generic jenkins jobs that call some or multiple scripts from the repo on git.debian [09:14] also fwiw, whenever you read dci somewhere, that's different stuff for the debian ci shadeslayer runs [09:15] sitter: OK dci = Debian Shadeslayer, Scarlett does the KDE stuff, and you've got the Kubuntu CI < right ? [09:16] yes [09:16] OK, Great. Well I've never looked at any Ruby before, there is a first time for everything.. [09:17] I'm setting up some stuff to play with to try get my head around it, so this will help me. Thanks :-) [09:20] ^ that'd be easier if I had already finished my pursuit of turning a full jenkins deployment into two commands ;) [09:23] sitter: Ooo that sounds interesting [09:28] yeah, needs some fixes still. right now the kci server has a manually maintained tooling directory which is in its layout /slightly/ divergent from the layout we have in git (mostly because the kci tooling predates the repos by like 6 months ;)) [09:32] We're building a CI rig here at work, I pointed the dev team at our stuff, and they've gone off and done the magic, and I'm now feeling inadequate, so now I have to get some more detail [09:32] * Sick_Rimmit shrugs [09:33] hey ho [09:34] ^^ [09:38] Sick_Rimmit: depending on what you intend to do, sgclark's stuff might actually be more useful. instead of the update-projects.rb script we have, kde's new CI uses a pseudo-language to describe build jobs. the reason we don't use that is because we need way more flexibilty in how and which jobs are created and how they related to one another that's potentially not needed for anyone but an OS [09:42] sitter: Well our deploylment is more straight forward, we have 4 software projects, PHP and C, We're using Cunit, PHPUnit and Fitnesse for REST APi testing. We build on top of Centos, and we plan to setup some containers with configured images in them, that's it [09:46] But as ever I wanted to use the exercise in both the context of my daily work and Kubuntu, as I'd like to graduate from Yeloow belts some thime [09:48] * Sick_Rimmit has dodgy typing skills this morning [09:53] mh, you definitely should look into what sgclark is doing for kde then, sounds like a better fit [09:53] she's using docker for container management I think [09:54] an alternative would be a virtualization management called vagrant, which I perefer over docker simply because it is plugin based so one can pick different virtualization tech depending on what is needed and possible and useful xD [09:54] travis-ci.com is built using vagrant for example [09:54] Oooo, I hadn't heard of that, I will check that out [09:55] vagrant and chef primarily and then a whole heap of other tech [09:55] Ok, Hadn't heard of Chef either, is that similar to Puppet ? [09:56] pretty much the same [09:56] Ah he, right. Excellent info thanks [09:56] also, the travis stuff is almost all on github somewhere I think, so something to draw inspiration from if needed [09:56] also pretty well documented [09:57] sitter: Brilliant.. thanks === rdieter_work is now known as rdieter [12:40] morning [12:41] good sunny morning [12:43] hi sgclark! [12:46] Riddell: seen my error report yesterday ? [12:51] sitter: Sick_Rimmit: well I had to abandon docker for now do to the fact I have to work with Windows and OSX as well. [12:51] sgclark: Hellooo there :-) [12:51] hi :) [12:52] Ah so we don't have that issue here, everything we deploy on is Linux [12:52] sgclark: what are you using instead? [12:52] but sitter put onto Vagrant which looks very interesting, I am going run up a play around setup to check it out [12:54] What I hope is to have a build under Jenkins control that works like BuildOS and Configure with set of packages, Check out code and build our system, run all the tests, and report [12:54] yes, I looked at that too. looks very nice [12:55] yes the 3 platforms has made my life very interesting haha [12:55] * sitter would totally use vagrant :P [12:55] and this kde_projects.xml is driving me batty [12:57] sgclark: how so? [12:57] I've got to put some more spare time effort into the KDE stuff, I am supposed to trying to learn Muon code, so that I can help with maintenance.. [12:57] other than that I am writing all the job creation code in groovy. which is fairly easy to learn. checkout job-dsl-plugin [12:57] Oh boy Groovy Ruby... more new stuff [12:57] sitter: trying to parse it and the nesting is not all created equal [12:57] * Sick_Rimmit mind boggles [12:58] I think I got it tho.. [12:58] haha [12:58] sgclark: what out for ktp [12:58] ? [12:58] it's super nested [12:58] bah [12:58] not sure what you need to do but that totally made my code fail [13:01] the inconsistent nesting gets real fun when you need to find something specific as every node can bey a project or a component/module thing [13:02] so you get to iter all elements in the document and guess whether or not it is what you are looking for xD [13:02] most inefficiently terrible thing ever -> https://projects.kde.org/projects/playground/sdk/releaseme/repository/revisions/rewrite/entry/lib/project.rb#L189 [13:02] yeah I had to create a json file with every job with a boolean for if it gets a job created [13:03] ^^ [13:03] because active does not mean it gets a jenkins jobs heh [13:05] nice, pretty much what I came up with but in groovy lol === kbroulik is now known as kbroulik-lunch === mgraesslin_ is now known as mgraesslin [13:17] sgclark: can you fix the kio migration autopkgtest failure today (or maybe it just needs a rerun? obviously the unit tests succeeded during build) [13:18] I think after that my qt 5.4 landing wouldn't be necessarily blocked if plasma-framework also migrated to release pocket [13:19] * sgclark looks [13:19] Mirv: do we have an eta for the 5.4 landing? [13:20] tsdgeos: as soon as Kubuntu's 5.7.0 has migrated to relese pocket so I can rebuild them :) otherwise we should be now good to go since we got confirmation emulator is working and I've been building the final packages (with proper version numbers) today [13:20] Mirv: cool [13:21] there was just in a ninja qtbase upload by tjaalton that means I'm also rebuilding qtbase, but other than that + frameworkintegration/plasma-framework/kdelibs4support rebuilds it's all done [13:26] Riddell: you uploaded kf5 5.7.0? if so you have the magic files to rebuild kio for Mirv please. [13:27] or sitter? ^ [13:29] actually, it looks like the tests you have at https://jenkins.qa.ubuntu.com/job/vivid-adt-kio/lastBuild/ARCH=amd64,label=adt/consoleText are not executed doing package build, ie are autopkgtest only, so you could have an actual test bug? [13:29] 3/5 of kiocore-listdirtest failing [13:30] we hardly run tests during build [13:30] shall I just override it? [13:31] if you're alright with overriding for now, I'm all for it so I can hopefully move forward with Qt [13:32] by god, it's blocking qt, override! [13:32] * sitter wants his shiny new qt :D [13:32] there's still a bunch that need built on arm64, that arch is always slow [13:32] QWARN : TestTrash::initTestCase() couldn't create slave: "Unknown protocol 'file'." <- that could entirely be a ksyscoca issue btw [13:32] Actual (KProtocolInfo::icon("trash")): "" [13:32] Expected (QString("user-trash-full")) : "user-trash-full" [13:33] ^ that'd then be explained by that as well as if the cache is busted or not built the protocol lookup would fail and no icon woudl come back [13:33] Should never happen: couldn't stat $HOME No such file or directory [13:33] * sitter should have started at the top xD [13:35] Riddell: override [13:37] shadeslayer: does dci have autopkgtest yet? [13:46] Added all of sgclark's failed upload packages to the supported seed. That should be sufficient to get them into the packageset. I should know in less than an hour. [13:47] thanks ScottK [13:47] ScottK: thanks!! [13:47] oh good bug 1378789 got in [13:47] bug 1378789 in kubuntu-settings (Ubuntu Utopic) "[SRU] Set the default IO scheduler to CFQ in Kubuntu Trusty" [Undecided,Fix released] https://launchpad.net/bugs/1378789 [13:47] kio, marble, cantor overridden [13:48] That's step one. We actually support all those packages anyway, so it seemed reasonable. (I mean as much as we support anything) [13:48] Riddell: pushed a potential fix for kio, haven't uploaded though to not block Mirv in case it still doesn't make the tests pass [13:49] * sitter sighs at the kate5 vs kate4 and konsole5 vs konsole4 cluster fuck [13:49] sitter: We can override the test failure if needed. [13:50] yeah, that's what Riddell is doing I think... but if I upload a new version it would block again, no? [13:50] Shouldn't slow thigs up as long as jr or I are aroud to override. I'll be here (mostly) for the next 10 hours. [13:50] fair enough [13:50] Wait for the curret one to migrate. Then upload the new one. [13:51] If it fails, let us know and we can update the override. [13:53] ScottK: do you know the difference between force-badtest, force-skiptest and force-autopkgtest ? [13:53] Vaguely. [13:55] I think it's use force-badtest if you know the test is screwed up so you want the test results ignored. [13:56] I think force-skiptest is if you want to skip a test for another package to get your package in. [13:56] Not sure about the autopkgtest one. [13:57] So if the test fails in package a due to the test being broken, you can force-badtest. [13:57] sitter: re: cluster fuck, I think I muttered those exact words too [13:57] That will also cause the test results for a to be ignored if an upload of package b triggers a retest of a. [13:58] sitter: at least kate doesn't have a part anymore in kf5 (ktexteditor is self-contained) [13:58] Something like that. [13:58] I think. [14:00] shadeslayer: would you mind explaining to me why this change is in the utopic branch http://anonscm.debian.org/cgit/pkg-kde/kde-applications/kate.git/commit/?h=kubuntu_unstable_utopic&id=8ada7d92e5b908f7e9bf702dbeedcdb3c0a3aa02 [14:00] rdieter: yeah [14:04] * shadeslayer looks [14:04] sitter: no, should only be in kubuntu_unstable [14:05] oh wait [14:05] Hiyas all [14:05] sitter: IIRC it was a issue with Netrunner where kate wasn't co installable with kate5 === kbroulik-lunch is now known as kbroulik [14:05] what? [14:06] sitter: did you merge in santa's changes to qca-qt5 in git.debian ? [14:07] "powerdevil_5.2.0-1_amd64.changes is NEW [14:07] looks like debian have started on plasma [14:11] Riddell: yes [14:11] sitter: great, but git confuses me cos I don't think it logs who did that so it looks like he just committed directly [14:12] well, he did? [14:12] I just merged what he commited into our clone [14:12] distributed version control ya know :P [14:14] ^^ :D [14:15] I actually like that, so I can track down people who made the actual change instead of the people who commited the change for someone else [14:17] would be nice to have both in my opinion, like bzr [14:18] Riddell: how would that work? all I did was push his changes onto our clone. there is nothing going on that identifies me as me [14:20] it's a new commit in bzr land I'm sure [14:21] sitter: you've been running away in the late afternoons, where do you go? is there gossip we need to hear? [14:21] Riddell: a merge commit, git doesn't do one unless it is necessary [14:21] and in git it isn't necessary because it doesnt use silly numbers :P [14:21] one of many things which confused me about git [14:22] * sitter be running to get drunk [14:22] Riddell: well, we could make it policy to always have merge commits [14:22] I do not see the point TBH, we already have enough merge clutter due to CI automerging [14:23] Riddell: other option is to add a signoff to the commits [14:24] would have looked like this then http://paste.ubuntu.com/10205208/ [14:25] notable disadvantage being that this rewrites the commit so the sha changes and santa would have had to make a merge with commit if he wanted to reuse his branch [14:26] Riddell: what would be the point of knowing who merged something? [14:30] sitter: it would stop me being confused because I read an e-mail that says "I committed over here please merge over there" and it's already over there [14:31] I'm not saying anything needs to change, just another bit of git I need to get used to [14:31] new blocker for kf5 https://launchpadlibrarian.net/197535998/buildlog_ubuntu-vivid-armhf.kdeclarative_5.7.0-0ubuntu1_FAILEDTOBUILD.txt.gz [14:31] and I need to run off to spanish class [14:41] mmm GLDouble issues === JoseeAntonioR is now known as jose [14:52] right, that blocks plasma-framework still [14:53] stellarium GLDouble was fixed with http://launchpadlibrarian.net/197291047/stellarium_0.13.1-1build1_0.13.1-1ubuntu1.diff.gz , maybe try something similar as the easy route? === mgraesslin_ is now known as mgraesslin [15:01] Mirv: discussing upstream [15:09] ok [15:14] Mirv: so qt 5.4's qopengl actually doesn't have that typedef anymore? [15:25] sitter: yes, it'd seem [15:37] (..on armhf) [16:34] i think this can be closed https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1421480 as it seems to be alreay fixed [16:34] Launchpad bug 1421480 in ubiquity (Ubuntu) "Dependencie problem libdebconf-kde1 Kubuntu Vivid Plasma Desktop" [Undecided,New] [16:41] Riddell: Mirv: FTBFS should be fixed [16:42] also considering kio migrated from proposed my test fix worked as well [16:49] sgclark: The cron job that updates the packagesets runs once a day at 0830 UTC. I think I committed the needed changes, but we need to wait for the cron job to run to know for sure. You can check http://people.canonical.com/~ubuntu-archive/packagesets/vivid/kubuntu tomorrow to see if those packages are listed. [16:50] Also, if you see any more missing, add them to the Kubuntu supported seed and let me know so I can run the script again. [16:50] ScottK: great thanks :) [17:22] ah yes, got a confirmation email, kio tests fixed properly [17:29] shadeslayer: just fyi I think I'm going to abort my phonon mission, reading pkg-kde-commits it seems they already fixed it. another thing: I think I still have a "merge request" mail awaiting in the moderator queue of kubuntu-devel, could you (or any other moderator) have a look? thanks in advance [17:30] sure [17:30] santa_: I don't see anything [17:30] only spam [17:31] ah, nevermind. it seems it got accepted (I can see it in the web archives) [17:31] cool [17:32] (however I don't see a confirmation mail that got accepted) [17:32] thanks anyway === JoseeAntonioR is now known as jose === rdieter is now known as rdieter_work