/srv/irclogs.ubuntu.com/2015/09/16/#kubuntu-devel.txt

=== petersaints_ is now known as petersaints
=== tazz_ is now known as tazz
=== TJ- is now known as Guest8103
=== TJ_Remix is now known as TJ-
ovidiu-florinahoneybun: pong05:48
ovidiu-florinahoneybun: anything else than what we talked in Hangouts?05:48
sitterhttps://bugs.launchpad.net/bugs/149612108:53
ubottuLaunchpad bug 1496121 in plasma-nm (Ubuntu) "nm-plasma crashes plasmashell" [Undecided,New]08:53
sitterRiddell: did plasmanm not get a rebuild?08:53
Riddelldunno, I've not looked at how the rebuild went yet08:54
clivejoRiddell yofel: missing files in Baloo - https://paste.kde.org/pfzzcbxmz - where do I put them?09:16
clivejoare the all dev files to go in not-installed?09:17
Riddellthose dbus interfaces can go in the -dev package and the .mo files are translations and should probably go in baloo-kf5.install09:19
clivejoso I see :) - http://anonscm.debian.org/cgit/pkg-kde/frameworks/baloo.git/commit/?id=d64f2679761701285c2d0d7f1732edbd92295fd009:20
clivejomaxy is a helpful chappie09:21
sitter[11:25] <kubuntu-ci> Project merger_baloo build #42: FAILURE in 6.7 sec: http://kci.pangea.pub/job/merger_baloo/42/09:26
clivejositter: it has missing files, Ive just uploaded a fix09:27
sitteryou fix just broke the merger09:27
clivejocan you explain? :/09:28
sitterRiddell: ^09:28
clivejohow do I fix what I broke?09:31
Riddellone sec09:32
yofelclivejo: see http://kci.pangea.pub/job/merger_baloo/42/console09:33
yofelyou'll have to manually merge kubuntu_wily_archive into kubuntu_unstable09:33
yofelah right09:34
yofelsitter: could you please reset my CI login? I can't get in09:35
* sitter has been thinking about wiring it to launchpad logins but ETOOLAZY09:35
yofel^^09:35
Riddellclivejo: yep do the merge yourself, it was already fixed in a slightly different way in unstable branch09:37
clivejoI dont understand how I broke it09:37
Riddellclivejo: git pull; git checkout kubuntu_unstable; git merge kubuntu_wily_archive09:37
Riddellclivejo: when you commit to kubuntu_wily_archive the CI system will automatically try to merge that change into kubuntu_unstable09:38
Riddellif there's a clash in that merge you'll get a poke09:38
Riddelland you need to do it manually09:38
kubuntu-ciYippee, build fixed!09:38
kubuntu-ciProject vivid_unstable_kaccounts-integration build #85: FIXED in 13 min: http://kci.pangea.pub/job/vivid_unstable_kaccounts-integration/85/09:38
clivejobut why did it fail?09:38
clivejohow was it slighty fixed in a different way?09:39
Riddellclivejo: do the merge and you'll see09:39
kubuntu-ciYippee, build fixed!09:39
kubuntu-ciProject wily_unstable_dolphin-plugins build #10: FIXED in 17 min: http://kci.pangea.pub/job/wily_unstable_dolphin-plugins/10/09:39
kubuntu-ciProject vivid_unstable_kdeconnect-plasma build #98: NOW UNSTABLE in 13 min: http://kci.pangea.pub/job/vivid_unstable_kdeconnect-plasma/98/09:39
clivejoCONFLICT (content): Merge conflict in debian/baloo-kf5-dev.install09:40
yofelright, now edit the file09:41
clivejook09:41
clivejosomeone added them file by file09:41
clivejoand I followed maxy by adding the folder09:41
Riddellgo with your way then09:42
Riddelledit that file manually09:42
Riddellgit add it then git commit it then git p ush it09:42
clivejocan I use git gui?09:42
Riddelldunno, probably, I use qgit sometimes but not usually for commits09:43
clivejook I think thats done it09:44
clivejoand baloo is green on qa - happy days09:45
clivejoare you guys working on apps?09:58
Riddellnobody is so far10:00
clivejoRiddell: http://qa.kubuntu.co.uk/ppa-status/frameworks/build_status_5.14.0_wily.html10:01
clivejowould you cast your expert eye over them please10:01
clivejoRiddell: yofel was doing them last night, but couldnt get login to pause CI10:02
yofelI did nothing10:03
clivejoin fear of sitter shouting and the time of night he left it :)10:03
Riddellclivejo: khtml could do with symbols being updated10:03
* sitter pets yofel10:03
sitteron that note10:04
Riddellkitemmodels too10:04
sitterI am going to enable symbol tracking in CI again10:04
Riddellclivejo:  kwidgetsaddons three10:04
clivejook10:05
clivejoyofel is is going to implement my new idea, DUS10:08
yofelI never said that10:09
yofeland the parser is perl10:09
clivejoLOL10:09
sitterdo not touch the pyrl10:11
sitterit's dangerous10:11
clivejodont worry sitter, I is an expert ;)10:12
sitterI presume we are talking symbols management?10:13
clivejoyeah, but Im only joking10:14
sitterbecause the much bigger problem with that is that it has no autotests so any change might randomly break things10:14
sitterwhich seems slightly undesirable with symbols :)10:14
clivejoRiddell: hows that?10:31
Riddellclivejo: how's what?10:31
clivejothose three are done, the qa list10:32
Riddellclivejo: foo khtml broke on amd6410:33
RiddellI guess it needs some fancy use of that symbol update script for both arches10:34
Riddellsitter: any thoughts on what's up with this? http://mobile.kci.pangea.pub:8080/job/vivid_unstable_solid_pub/10:34
yofelthat's not fancy, that's what you should always do10:34
sitterRiddell: yes, read the mail I sent to bluesystems list about exactly this10:35
yofelfixing architecture mismatches later on is a lot more work10:35
=== zbenjamin_ is now known as zbenjamin
clivejoyofel: will you teach me how?10:35
yofelclivejo: you feed all buildlogs into batchpatch10:35
clivejoso download both the buildlogs and run it against each of them in turn?10:36
yofelno, pass all of them in one command10:36
yofeljust do something like buildlog* at the end10:36
clivejocan you show me an example of the command you run?10:36
sitteryofel: that's not gonna work10:37
sitterthe logs need to be in order10:37
yofelin what order?10:37
sitterat least if you explicitly define the architecture list10:37
yofelah, I never do that10:37
sitterin that case I'll shut up :P10:38
yofel^^10:38
clivejositter: how would you do it?10:38
sitterhm10:38
sitterubottu: info libtelepathy-qt5-dev10:38
sitterhttp://people.ubuntu.com/~apachelogger/tmp/symbol/symbolpatch10:39
sitterubottu: find libtelepathy-qt5-dev10:39
clivejo*coughs*10:39
sitterthat bot doesn't like me :(10:39
sitterpft10:40
sitterRiddell: our telepathyqt is too old for kdeconnect master :(10:40
sittersomeone should talk to apolo about this ;)10:40
* yofel -> lunch10:41
clivejoah lunch, that reminds me I must fill the cattle drinker10:42
sitterRiddell: is it possible that you didn't symlink plasma-mediacenter from kde-extras to plasma?10:42
Riddellsitter: in debian git ?10:43
sitteryes10:44
BluesKajHiyas all10:46
Riddellsitter: I did not, I'll do that now10:49
Riddellsitter: symlink added10:52
sitterRiddell: have you given any thought how to do builds in jenkins without CI?11:02
Riddellsitter: I don't understand, you mean with tars?11:03
sitteryep11:03
Riddellsitter: replace the _src job with one that uses uscan?11:04
sitterwell yes but how :P11:04
Riddellalthough I guess that wouldn't work with hidden tars on kde server but not on download yet11:04
sitteralso what if uscan isn't enough?11:04
sittere.g. kde releases that aren't public yet11:04
Riddelllet it fall back to using ssh11:05
sitterthat needs code abstraction somewhere tough11:05
Riddellwhy would it be any harder than working out how to check it out of git?11:06
RiddellI guess checking it out of git is built into jenkins somewhat?11:06
sitteractually this is a bit shitty because we handle git-source in jenkins but we'd be doing the others in ruby11:06
sitterso yeah, the separation hurts a bit from an architecture POV11:06
Riddellor python.  it could be written in python.11:06
sitterit can't11:06
Riddelloh11:06
Riddellpor que no?11:07
sitterbecause it's a shit language11:07
sitterthat's why11:07
Riddellah I see11:07
sitteralso on the more practical side we have no provisioning for python deps11:07
sittereitherway that doesn't matter11:08
sitterso on the jenkins job side we already have tech in place that can sort of tell apart a packaging repo from an upstream repo IIRC and will simply work with only packaging but no upstream11:09
sitterso it would only use the packaging repo and then hand over to ruby or whatever11:09
sitterinside there we'd have a couple of steps to generate the source: 1. get the source and tar if necessary -> 2. merge packaging into source- > 3. mangle packaging to CI -> 4. buildpackage -S11:10
sitter3 must not really happen for non CI builds11:11
Riddellthis sounds good11:11
sitterand 1 would need to have different classes with a factory GitSource(copy from clone dir, gen orig.tar), TarSource (uscan > arbitrary code stored somewhere for ssh)11:13
sitterformer would eventually run through releasme to get l10n11:13
RiddellI know that's been your hope for a long time but I would guess it's still a bit long term11:14
Riddellreleaseme still needs extra stuff added, like all the plasma scripts I have11:15
Riddelland not everything uses releaseme11:15
sitterwell11:16
sitterfirst step is using releasme's l10n classes to get l10n for our clones :P11:16
sitterwhich should be somewhat trivial11:16
sitteralbeit super low priority11:17
Riddelland super slow11:20
Riddellit's still a good chunk of the plasma release process getting l10n11:20
sitterRiddell: didn't we make it fast?11:21
Riddellsitter: sure it's lots better than before11:21
Riddelltakes an hour rather than 4 hours11:21
sitteran hour seems a bit extreme though11:21
sitterah11:21
sitterincluding sources11:21
Riddellbut if you have everything in CI doing that it'll take quite a while11:21
sitterwell11:21
sitteryou've got some 30 sources in plasma?11:22
Riddellyes11:22
sitterthat's 2 minutes per source. seems pretty reasonable TBH11:23
sitterso say 1 minute for l10n which is pretty reaosnable11:23
Riddellok, you know CI better than me11:23
sitterthe thing with CI is that it would be somewhat parallel11:23
sitterso it wouldn't be idle building sources for 60 minutes but more evenly spread it out11:24
Riddellright11:25
sitterkci going down for maintenance in ~15 minutes11:25
shadeslayersitter: Isnt vcsbuilder exactly what you described above?11:49
sitterno11:50
sitterit abstracts on the concept of source building11:51
sitterwhich is actually too high11:51
sitteras I said, there's 4 steps only 2 deviate between source types11:51
shadeslayerI thought that's what vcsbuilder was going to become, hmm ...12:01
sitterso did I, but it's not quite what we need :P12:02
shadeslayer:)12:02
sitteror maybe it is, but then we still need a base of sorts to contain the shared 2 steps12:02
shadeslayer*nod*12:02
sitterlogic abstraction generally is nicer to read than design abstraction though12:02
shadeslayersitter: BTW where do I move that mgmt folder out to?12:03
sitterso we have one source class it just has behavior properties that change depending on the environment it is used in or something12:03
shadeslayerci/mgmt ?12:03
sittershadeslayer: mgmt/12:03
sitterdon't matter much I guess12:03
sitter    @data = YAML.load_file("#{File.dirname(__FILE__)}/data/maintainer.yaml")12:08
sittershadeslayer: that's wrong I think btw12:08
sitterin build_source12:08
sitteroh wait it is in there12:08
sitternot a fan of data in lib/ :P12:08
shadeslayersitter: added to list of things to fix12:31
sitterRiddell, yofel: shouldn't marble-qt be dead?12:47
sitter00:00:49.506 KCI-W :: W: marble-qt: empty-binary-package12:47
Riddellsitter: who says it's empty? it gets built when I build it12:47
sitterCI says12:48
sitterit also has no install file12:48
sitterso I don't see how it can contain anything12:48
yofelThe original plan was to kill it, but I never finished that - and someone said that it might still be useful12:48
RiddellI did have to add one I remember12:49
sitterdebian/not-installed:# marble-qt12:49
sitterdebian/not-installed:./usr/bin/marble-qt12:49
yofelbut I'm not quite sure if QTONLY needs a double build12:49
sitterRiddell: you did not push then12:49
Riddellsitter: debian/marble-qt.install is in kubuntu_unstable12:51
Riddellsitter: which branch are you on?12:51
sitterwily_archive Oo12:51
sitter[remote "origin"]12:52
sitter        url = debian:applications/marble12:52
sitterRiddell: http://anonscm.debian.org/cgit/pkg-kde/applications/marble.git/commit/?h=kubuntu_unstable&id=c5615d60b1debe15c3255cd23085e4b174c3269012:53
sitterI would say something about the fact that there is stuff in the commit that has nothing to do with the changelog but I'll just pretend I dont' care anymore12:53
Riddellpushed marble-qt.install to wily_archive, it'll get updated when someone does applications 15.08.112:58
sitterok13:00
yofelwhen I read lp 1496121 I wonder why we put so much effort in symbol tracking if other people end up braking stuff anyway.. *sigh*14:22
ubottuLaunchpad bug 1496121 in plasma-nm (Ubuntu) "nm-plasma crashes plasmashell" [Undecided,New] https://launchpad.net/bugs/149612114:22
yofel*breaking14:22
yofelactually, perfect example14:24
yofelclivejo: THAT ^ is the reason why we have symbol files14:24
sitterthat's why we need ACC ADT, no? :P14:47
sittersymbols files we need to do precise link dependencies14:48
yofelwell, true, for c++ we really need ACC14:51
tsdgeoswe need AC/DC!15:07
yofellol15:09
* clivejo pictures sitter head banging to AC/DC15:10
* clivejo giggles15:11
ovidiu-florinyofel: ping16:24
ovidiu-florinanyone here?16:30
yofelovidiu-florin: HM?17:10
yofeloops17:10
yofeldrat, back in a few mins17:12
soeenew frameworks almost ready ? :)17:50
soeeRiddell: shouldn't Frameworks 5.12 for vivid be removed already from stagin ppa (https://launchpad.net/~kubuntu-ppa/+archive/ubuntu/staging-frameworks/+packages) as they were released ?18:07
clivejosoee: nope they arent ready, symbol issues18:24
* ahoneybun has rebuilt his linode server18:48
clivejowhat happened to it?18:50
ahoneybunwhen I made it I did it wrong or something18:57
clivejooh18:57
clivejohave apps 15.08.1 been uploaded yet?19:36
clivejoyofel Riddell ^19:42
ahoneybunweird Plasma does not like you using TWRP19:55
ahoneybunI crashed it lol19:55
clivejohow do I fix khtml?19:57
clivejocan I just copy debian's soluation?19:58
yofelsounds reasonable20:09
clivejoyofel: how would I do a revert of my commit20:10
yofelI wouldn't revert anything but just copy&paste that line from master into your file20:10
clivejoI would like to do it properly20:10
clivejofor practice20:11
clivejorevert the commit, grab the two build logs and feed them into pkgkde-symbolshelper20:11
clivejoyou think Im crazy?20:12
yofelno, just it'll take about half an hour if you want to leave the builds to launchpad until we're done...20:12
yofeland it won't even work in the archive because armhf and powerpc will be missing from the symbol annotation20:13
yofelhm...20:13
clivejowhat should I do?20:14
yofelI just noticed another thing20:14
yofel_ZNK3KJS7JSValue9getNumberEv@Base 5.1420:14
yofelthe upstream version is 5.14.0, not 5.14. Has no real impact in this case, but please try to use the *exact* version next time20:15
yofelclivejo: what *I* would do is: take the symbol from the debian file, put that into ours, fix the version to say 5.14.0 just to match what we're doing and remove any architectures we don't have20:16
clivejohow did I introduce that, by using pkgkde-symbolshelper batchpatch -v 15.14 khtml*.build?20:19
yofelprobably20:20
yofelactually, you only gave it the i386 build log20:20
yofel# SymbolsHelper-Confirmed: 5.14 i38620:20
yofelshows that20:21
clivejoI give it the amd64 first, then the i386 failed to build so I gave it the i386 buildlog20:22
clivejodidnt realise that sometimes you have to give it the two buildlogs20:22
clivejoI have the 386 one so if I download the amd64 and apply both would that fix it?20:23
yofelalways try to feed it all build logs. symoblshelper can only determine information for the logs it has, so as it saw a symbol missing on i386, but didn't have the amd64 log which said that no symbols are missing, it couldn't figure out that the symbol is architecure-specific20:24
clivejoit is failing on the amd64, so if I feed it with the amd64 buildlog will it remove the i386 symbols?20:26
yofelif you do that, you'll again have a symbol diff on i386 with an added symbol20:27
yofelyou need to feed it ALL logs20:27
clivejook20:28
clivejograbbing amd64 now20:29
clivejoso use -v 5.14.0 ?20:30
yofelyes please20:30
clivejoor should I leave it and "learn"20:30
yofelnah, if you're already at it finish it20:30
clivejopkgkde-symbolshelper: warning: * patch 'libkf5khtml5_5.14.0-0ubuntu1~ubuntu15.10~ppa1_i386 (--- debian/libkf5khtml5.symbols)' for i386 ... FAILED.20:32
clivejoso that is expected as I had already applied the i386 buildlog?20:32
clivejo (arch=i386)_ZNK3KJS7JSValue9getNumberEv@Base 5.14 (was arch=)20:33
yofelwell, it did the right thing at the end20:34
clivejoyou reckon thats fixed it?20:34
yofelfor the PPA yeah20:34
yofelwill most likey throw warnings in the archive, but we can fix that later20:35
clivejoso commit, push and rebuild?20:35
yofelyep20:35
clivejothere is another file has changed in git gui20:35
yofelhuh?20:36
clivejoa .orig file20:36
clivejois that a git file?20:36
yofeluhm, no20:36
yofelsomething might've gone wrong when applying the diff20:36
clivejolibkf5khtml5_patched.symbolsJ212jy,orig20:37
yofelany errors in the real symbol file?20:37
clivejonope just two diffs20:37
yofelthen delete the file20:37
clivejo+# SymbolsHelper-Confirmed: 5.14 amd64 i38620:37
clivejoand + (arch=i386)_ZNK3KJS7JSValue9getNumberEv@Base 5.1420:38
clivejoif I just ignore it and dont stage it20:38
clivejois there an easy way to download the buildlogs?20:41
yofelI personally go to the status page, right click on arch > copy link address, then wget the file20:43
clivejothats the way I been doing it20:43
yofelwe could write a script that does that, and we have one for the archive, but so far nobody bothered20:43
clivejoyofel: did you get your CI login sorted?20:46
yofelyep20:46
clivejoupload apps?20:46
yofelno motivation...20:46
clivejooh?20:47
clivejocan I?20:47
yofeluh.. lets see if I can figure out where the pause button is20:47
clivejohttp://kci.pangea.pub/view/mgmt/job/mgmt_pause_integration/ ?20:48
yofelthere20:49
clivejoyou did something :)20:49
yofelI clicked on a button :P20:50
clivejoI clicked the button too, but it doesnt know who I am :P20:50
yofelok, so if you want to upload apps go ahead20:50
clivejoyou sure you dont want to?20:51
yofelyes20:51
clivejodont want to fall out over it :P20:51
yofelyou know the git script better than I do, and I don't want to involve myself into something long-running right now20:52
yofeljust remember to put the bottom of the upload script output with the errors on the pad20:53
clivejoso bzr merge to refresh my kubuntu-automation folder?20:53
yofelprobably not needed, but do it anyway to be on the safe side20:54
clivejosomeone has modified ppa-build-status20:55
clivejowas that you on weegie?20:55
yofelyep, the log would tell you that as well :P20:56
clivejoI didnt see the log20:56
yofelbzr log --limit 120:56
clivejoso I need to commit that to my personal branch?20:56
clivejoIm only seeing myself!20:57
yofelI guess? I'm not too familiar with the branched bzr workflow20:57
clivejoweird Im not seeing that you made that change20:58
clivejowould you make it as Riddell?20:58
clivejomaybe I should delete my local folder and pull it from bzr20:59
yofeldoes 'bzr log -n 0' show it?20:59
clivejonope21:01
yofeldid you do a bzr pull?21:01
clivejonothing to pull21:01
clivejothis is weird21:02
yofelno idea then. As I said, I don't really know how to use bzr like that21:02
clivejome enither21:02
clivejoyou are definately there - http://bazaar.launchpad.net/~kubuntu-packagers/+junk/kubuntu-automation/changes/616?start_revid=61621:03
clivejorm it and recreated21:05
clivejoyou are there now!21:05
clivejohumm it wants to remove libkexiv2 from the list21:09
yofelyeah, dunno why21:09
clivejoif I add it back into the applications-wily will it break it?21:11
yofelno, the tarball is on depot from what I see21:13
clivejosomething badly worng21:17
clivejo==== Skipping akonadi, failed to fetch tarball from depot.kde.org, moving to manual/21:17
yofel akonadi-15.08.1.tar.xz21:18
yofelit's there21:18
clivejoI know21:18
clivejobut the script cant see it21:18
yofelanyway, I'm off for the evening21:19
yofelgn21:19
clivejook, have a good evening21:19
clivejoO/21:19
clivejoRiddell: ping21:41
Riddellhi clivejo21:53
clivejohowdy21:57
clivejois the staging ppa ready for these apps (15.08.1)21:57
clivejoRiddell: ^21:58
Riddellclivejo: as ready as it'll ever be :)21:58
Riddellclivejo: have you got them ready to upload?21:58
Riddellclivejo: how's frameworks doing?21:58
clivejoyodel helped me fix khtml earlier21:59
clivejoI think its good to go ?21:59
clivejoyofel even21:59
Riddelllovely21:59
clivejoI see 15.08.1 fixes that problem in ark22:00
Riddellclivejo: if you want to do the next step of uploading it to ubuntu you can run the comment "./ubuntu-archive-upload -v 5.3.2 -t ~/src/plasma/ -r plasma" (but with bits changed as relevant)22:00
Riddellthen send a tar of all the .changes files to me or someone to sign them22:00
Riddellor I'll just do it tomorrow22:00
clivejoRiddell: there was talk that it cant go into wily?22:01
clivejodue to feature freeze?22:01
Riddellclivejo: you could also file a feature freeze exception bug request22:01
clivejodoesnt bother me22:01
clivejosurely it should come from someone on KC?22:02
clivejobut needs testing 22:02
clivejogod knows what mistakes Ive made with the symbols22:02
Riddellclivejo: nope it should come from any packager and be approved by someone on release team (like me)22:03
Riddelland yes you should install and test frameworks indeed22:03
clivejoIm keen to test apps22:04
Riddelland upload that to the staging-applications PPS22:05
RiddellPPA22:05
Riddellmy head is collapsing, I need to sleep22:05
clivejoRiddell: do you know why it wants to remove libkexiv2 from the packaging list22:05
clivejoit being the script22:05
Riddellno idea, it is in there http://download.kde.org/stable/applications/15.08.1/src/22:06
* Riddell snoozes22:06
clivejonight night22:06

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!