=== skreech_ is now known as Daskreech [11:43] ScottK: I've uploaded a new kde-l10n-da version as it FTBFS [13:15] Hey folks [13:15] Yo BluesKaj [13:19] yoyo smartboyhw :) [13:20] yoyo [13:23] debfx: I'll have a look. [13:37] debfx: Accepted. === yofel_ is now known as yofel [15:46] Up on 4.8.5 from proposed and it all seems good. [17:18] JontheEchidna: We're several releases behind on apt-listchanges, so having a look at it would be good even if it's just to update to Sid/Wheezy. [17:18] * JontheEchidna nods [18:49] apachelogger: what's the rationale behind a version like ~quantal0~ppa3~1 ? [20:16] Is everybody jamming? [20:17] taking a youtube break :P [20:19] Haha [20:19] * yofel tries to finish shadeslayer's package work [20:19] fell asleep on the couch yesterday -.- [20:19] long week [20:19] * txwikinger hosting the Waterloo Global Jam [20:20] * txwikinger woke up 4am this morning.. long day [20:40] * yofel wonders how kdepimlibs built in the PPA with "4.9.0" in the lib install files [20:41] something went seriously wrong with the 4.9.1 uploads [20:41] kdelibs5-dev (>= 4:4.9.0a) would explain how it built [20:43] do we have a general location for the automation scripts or are we still using the repos owned by debfx ? [20:43] apachelogger: ^ [20:54] apachelogger: can you please move your TODO out of the topic? At least put it on the packaging pad if nowhere else [21:03] Is do-release-upgrade -d not working yet? [21:04] txwikinger: check what your upgrade policy is set to, it seems like update-manager auto-set that to LTS for everyone in precise [21:05] then -d won't do anything until 14.04 [21:12] apachelogger: actually, wouldn't it make sense to use .so.4.*[.abiX] wildcards in install files? the abi version is 4 anyway, not 4.9 [21:18] Hmm.. maybe it is my proxy setting that is not valid here :-) [21:19] checked your upgrade policy? [21:21] can I do that commandline somewhere? [21:21] /etc/update-manager/release-upgrades [21:21] make sure Prompt= is not set to LTS, but Normal [21:21] *normal [21:23] thanks yofel.. that did it [21:24] that is certainly a stupid default [21:29] yofel: rationale: not having crafted magic to translate series name to version [21:29] yofel: automation: kubuntu-dev-tools [21:29] yofel: todo: no [21:29] why not? [21:29] yofel: abi: yes/no [21:30] cause the point is that I don't forget about it :P [21:30] uh... [21:30] nvm [21:31] as for kubuntu-dev-tools: do you mean 'put it there' ? because I don't see the stuff there [21:31] put proper stuff there [21:31] good [21:31] then I'll fix the kde-sc-dev replacement handling and move things over [21:31] the present main script is not proper [21:32] didn't take a too long look at it yet, all I noticed is that it's buggy [21:32] very buggy [21:32] anywho [21:32] if you feel like it, throw it in, but I'll be redoing half of it :P [21:33] sure [21:33] which reminds me that we should have metting [21:33] meeting even [21:33] I just don't want us to end up modifying the same file in different places and having to merge stuff again later [21:34] well [21:34] for now I'll just try to patch the worst things and somehow get a proper 4.9.1 build [21:34] that already happened, didn't it [21:34] i.e. kgetsource does half the stuff initial-upload does [21:34] dunno, ask debfx about that [21:34] we should probably stick to 1 language when writing scripts [21:35] kgetsource is ruby [21:35] yofel: I do recommend you write a quick script for 4.9.1 [21:35] to get proper we only need bzr co && get source && build source [21:35] yofel: no we shouldn't [21:36] yofel: what we should have is a whole bunch of scripts in whatever language makes sense and glue them together using le bash magic [21:36] fine with me [21:38] the keyword really is modularity, by keeping multiple scripts for multiple tasks they can easily be glued together differently, used manually etc. etc. [21:39] also it allows multi-threading on process style modularity... i.e. if the host script is bash we can easily fork 5 times to run e.g. kgetsource 5 times at the same time [21:40] which in turn makes concurrency a no-brainer [21:40] good point [21:40] on a related note: dinosaurs on a spaceship! :O [21:42] that would be the proof that intelligent life exists outside earth + someone thought of the ark before Noah [21:44] btw. what was the 'no' part about the ABI? There's not really a point in being stricter about it than dpkg [21:45] except if you want to catch upstream brain failures using that... [21:45] exactly that being the point [21:46] in fact strictly speaking them adding symbols and only bumping the last digit would already count as such [21:46] Any Qt D-Bus ninjas around that can tell me why this works: http://paste.kde.org/546074/ and this fails? http://paste.kde.org/546062/ [21:46] for that kind of thing you'd usually bump the middel digit [21:46] JontheEchidna: dinosaurs!!@#@!#% [21:47] strictly speaking, keeping 4 and breaking the ABI would already be a failure, but we have our symbol files for that [21:47] so not sure what 4.9 adds to that [21:47] they did not break the ABI [21:47] apachelogger: downloading now !@#E$!#@ [21:48] basically it goes like ... YOUREMOVECRAP.YOUADDCRAP.YOUFEELFUNNY [21:48] uh yeah... in theory [21:48] yah [21:48] so I believe only wildcarding the last digit is the most resaonble way of doing it [21:49] I'm not sure why some folks understand that (libkdcraw is at .so.21) and some not [21:49] some simply don't care :P [21:50] "uh this is a library only we use, yeah we install it publicly and also install it's headers, but really, no one would ever use it..." [21:50] now where did I hear that last... [21:50] oh yeah, nepomuk [21:51] * apachelogger ponders showering JontheEchidna in spoilers :P [21:51] or is it with [21:51] my my [21:51] I had too much coke [21:51] hooray [21:51] which reminds me that I need to package that nepomukwidgets thingy to unbreak neon [21:52] oh noes [21:52] JontheEchidna: wb [21:52] thx [21:52] JontheEchidna: what's not working about the dbus magic? [21:52] do you get the signal? [21:52] apachelogger: finished slot for my async never gets called [21:53] but calling waitForFinished to force it to be synchronous gets the error reply I want [21:53] hm [21:53] why the code looks alright [21:54] JontheEchidna: d->dbus is an interface I suppose? [21:54] yup [21:55] bleh, my connection is crap === jtechidna is now known as JontheEchidna [21:56] very nice :P [21:56] apachelogger: yes, it's a dbus interface [21:56] you sure the connect() is correct? [21:56] I fail to parse signals that long ^^ [21:56] apachelogger: yeah, Qt Creator tab completed it for me, plus it's the same as in the Qt docs === skreech__ is now known as Daskreech [21:57] plus there's no runtime error saying "no such signal" [21:59] * apachelogger looks at documentation [22:00] JontheEchidna: what type is the transaction? [22:00] JontheEchidna: it doesn't get deleted before the finished signal gets emitted, does it? [22:00] that would explain why it does not work ;) [22:00] JontheEchidna: it doesn't get deleted before the finished signal gets emitted, does it? [22:00] that would explain why it does not work ;) [22:01] I don't think it would, the slot's the only place where it could be deleted [22:01] right? [22:01] I am talking about the Transaction object === jtechidna is now known as JontheEchidna [22:01] not the watcher [22:01] oh, actually... [22:01] now that you mention it [22:02] * apachelogger saves the day [22:02] also dinosaurs [22:02] still downloading :( [22:02] in fact that may be what's causing my IRC issues [22:02] heh [22:02] the doctor munches all the bandwith [22:03] yus [22:08] JontheEchidna: I don't know if I should be woried but Muon hasn't worked for like amonth [22:09] * yofel has no issues with muon in quantal [22:09] what fails? [22:27] [lp:~kubuntu-packagers/kubuntu-packaging/kdepimlibs] Philip Muškovac * 189 * debian/ (30 files) * update install files * add new symbols to libakonadi-kde4.symbols [22:28] So I had brought an old router from home to uni. [22:29] But I forgot that it was the one that couldn't hold an IRC connection when you were running a torrent [22:29] apachelogger: thanks. I moved the dbus watcher to TransactionPrivate, which stays on the stack as long as there is a Transaction object with a QSharedPointer to it [22:30] *stays on the heap [22:30] now... dinosaurs! [22:30] hf [22:30] meh, apachelogger putting invalid mail addresses in changelog ^^ [22:31] that happens [22:31] * apachelogger used like 5 shells to get things moving faster than the script would allow :P [22:31] QSharedPointer is awesome [22:31] heh [22:32] automatic garbage collection for your private class, so you don't have to throw pointers to instances of your public class around :D [22:33] everytime I hear garbage collection I think of java... and shudder [22:34] * apachelogger failed to compute that line about public classes [22:36] So, transactions are generated by the worker daemon and live on a D-Bus path until the daemon deletes them [22:36] the client transaction class uses a dbus adapter to keep its copy of the transaction's data up-to-date on the client side [22:37] when a transaction on d-bus goes away, LibQApt will let it's client-side transaction instances go out of scope [22:37] and if an application using libqapt still has client-side transaction instances pointing to the TransactionPrivate data, the TransactionPrivate data will stay until the application lets it's Transaction object go out of scope [22:39] so you have multiple transaction objects using the same private data object? [22:42] yep, their only data member is a QSharedPointer [22:42] the backend can keep a copy to do with what it likes, and the app using the backend can keep its own copy for displaying the data to the user [22:43] when the daemon deregisters the transaction object on D-Bus, the backend will remove its instance from its list [22:43] and when the application is done showing data from its transaction instance and it goes out of scope, the TransactionPrivate is deleted [22:44] common approach [22:44] suffice to say I prefer not having a performance penalty :P [22:44] though I see the usefulness ;) [22:46] yofel: http://paste.kde.org/546128/ [22:46] ;) [22:46] isn't it beautiful [22:46] I just didn't want to give users of the library to delete stuff under my nose :P [22:47] and I'm sure fetching data over the D-Bus is the bottleneck. There shouldn't be more than a dozen transactions active at a time in the worker anyways [22:47] :D [22:47] apachelogger: nice :D [23:25] [lp:~kubuntu-packagers/kubuntu-packaging/kdelibs] Philip Muškovac * 431 * debian/ (37 files) * Change sonames from 4.9.0/5.9.0 to 4.9.*/5.9.* * Update kdelibs5-dev.install fix uploader email [23:34] I really liked that episode of Doctor Who [23:49] it had dinosaurs [23:50] * apachelogger recreeates initial-upload in bash [23:52] uhm [23:52] http://paste.kde.org/546134/ [23:52] why that was irritatingly simple -.- [23:52] * apachelogger waves fist at stupid xargs for having features all implemented [23:52] dunno [23:52] heh [23:52] inital-upload does seem to have a few extra features though [23:52] it even works(tm) [23:53] what a drag [23:53] no fun for apachelogger :( [23:53] that is if they work [23:53] yofel: like? [23:53] checking if there's a point in uploading a package (0-diff) [23:53] also, build-dep bumping needs implementing, although I've that half-done here [23:54] *I'm half done with that here [23:54] why that's just additional magic [23:54] also isn't diff checking like expensive? [23:54] less expensive than forcing more updates to users - also less that can break [23:55] it's freakin's expensive indeed though [23:55] -s [23:55] well I just wonder [23:55] how often does that happen? [23:55] past ppa1 you only do selective uploads anyway [23:55] xah [23:55] ah [23:55] now I get it [23:55] with the number of packages we have here, I believe over a dozen per point release [23:56] diffing 4.9.1 vs 4.9.0 [23:56] right [23:57] that is indeed slightly expensive, so I'd not put that in initial upload TBH [23:57] more like archive-upload [23:57] that won't help with the PPA stuff though - if we care [23:57] I am not sure we should care [23:58] I mean, if you opt for a PPA you probably have a not-too-crappy intartubes access [23:58] as long as it's at least in archive-upload I would be fine with it [23:58] stuff like that is why I think we should have a meeting :P [23:58] true [23:58] also it would be itneresting how many false negatives a 0-diff check yields [23:59] because at least with the large meta sources there were at least cmakelists changes that change a version or a whitespace