[01:24] yuriy: that's what you get when you write software for Ubuntu I guess :/ [01:24] does it even have me listed as the author? or a author? [01:24] or something? [01:28] Just fix it an upload it then. [01:32] ha [01:35] nixternal: it does -- i'm talking about the kaboutdata, you probably changed it [01:39] I can't remember..I remember setting myself as the author and leaving the (C) for Canonical though [01:40] I seriously don't care about the copyright...as long as it stays free software of course :) [01:41] * nixternal goes and eats, then after than is going to send a railing email to the moveon.org idiots who claim to be "nonpartisan" yet bash republicans on their website and ask for donations to a democrat [04:43] bug 427841 [04:43] Launchpad bug 427841 in ubuntu "[FFe] new package kamoso" [Wishlist,Fix released] https://launchpad.net/bugs/427841 [04:43] wow, i was quite late on that one === obama is now known as vorian [04:43] Change has been achieved [04:51] heh [04:51] within each small move is the kernel of a radical change [06:03] apachelogger: weren't you working with scarabeus on the gettext branch a few months ago? :) [06:03] we had it working "in theory", just that some setups wouldn't use translated strings :) [08:41] ryanakca: ping === Blizzzek is now known as Blizzz [09:02] ScottK: kamoso dev already sent mail to lure and gilles [09:02] proposing either a shared lib or improving the kipi interface or rewriting the code [09:13] Sput: in theory :P === yofel_ is now known as yofel [09:14] Sput: IIRC the only reliable soution is to use lowlevel gettext, i.e. it's c++ implementation directly [09:14] and even then I would not know what to do with the ui files [09:15] Sput: I would just limit localization to the KDE enabled client :P [09:15] AFAIK that is how marble does it [09:15] actually [09:15] .... [09:16] * apachelogger just had an idea [09:19] ok, that is so insane it could actually work [09:20] Sput: following scenario: you stack klocale on top of the KDE integration using i18n() only inside KDE components, unless I am completely wrong KLocale should also ensure that tr() gets translated [09:20] then we use gettext to obtain two pot files [09:20] one for qt only and one for KDE integration [09:21] the latter will manually add the qt one to the klocale translation catalog [09:21] so [09:21] depending on the cmake arguments we either build a mo or a qm [09:22] so, if you build with KDE you get 2 mos per language, one kde and one qt [09:22] if you build qt only you will get one qm file per language [09:22] as insane as this sounds, it might actually work :D [10:30] HURRR [10:30] apachelogger: ping [10:30] Quintasan: pong [10:31] apachelogger: what should I add to rules when package is build like this "qmake dooble.pro && make" ? [10:32] apachelogger: I tried adding those to default rules but it keep failing because of 'clean' part :/ [10:32] cdbs' qmake.mk [10:32] let see [10:32] Quintasan: you probably need to sort clean yoursef [10:32] qmake is a PITA [10:33] for the developer and the packager and the user [10:33] qmake is one big fail IMHO [10:33] apachelogger: debuild -S starts with clean, isn't clean supposed to run AFTER compilation? [10:33] poke upstream and recommend cmake [10:33] Quintasan: no [10:33] before and after [10:33] the source package needs to be clean aswell :P [10:33] but I start with clean source, don't I? [10:33] well [10:34] if you run debuild -us -uc you have poluted source [10:34] which is why clean is always run before the source package is built [10:36] damn, each package I pick is fckd up or upstream is too lazy to fix it [10:36] well, all the more to learn :) [10:37] train clock plasmoid borked, acetone was propably to lazy to sort out copyrighted images [10:37] well, you can always dfsg copyrighted stuff [10:38] it usually is more annoying to upstream when users come whine about how there are no images than packagers whining about copyrights :P [10:38] :D [10:38] * apachelogger kicks hg [10:38] I found Silicon, looks better than Acetone [10:38] darn hg convert refuses to convert konvis full history [10:39] * apachelogger imports konvi into git hoping that hg has more luck converting git :D [10:39] Quintasan: well, acetone would be a no-go for packaging for me [10:39] way too ugly website [10:39] http://www.hyperray.net/hyperget/silicon-auto-image-mounter <-- over 9000 times better [10:40] why auto-mounter though? [10:40] what is so auto? [10:40] dunno [10:40] cool :D [10:40] It works just like daemon tools [10:41] Quintasan: btw, did you cooridnate a KDE release packaging session yet? [10:41] is that even comon these days? [10:42] * apachelogger sure looks forward to october when he can jump in with all the godlike powers he got again :P [10:42] I don't think so. You mean session to explain what ninjas do? [10:42] Quintasan: nah, to ensure they dont fck up [10:42] i.e. revu all changes and basically jump for the important, yet difficult packages [10:43] JontheEchidna did some IIRC, he probably can explain when he gets up [10:43] fck, qmake is srsly a PITA [10:43] * apachelogger notes that importing konvi into git takes ages [10:43] Quintasan: told ya :P [10:44] Quintasan: you most likely need to implement everything but qmake/make yourself [10:44] i.e. clean and install [10:44] HA [10:44] okay, two ways to get it to generate makefile [10:45] delete win and osx *pro files or hack rules :P [10:45] debuild will restore deleted files, so the former is an option [10:45] but [10:45] if you want to learn something... [10:46] implement the whole build using low-level debhelper [10:46] you can pass a specific pro file to qmake IIRC [10:46] debhelper...urgh [10:46] so if you have ultimate control over the build you probably dont need to delete that stuff [10:47] * apachelogger is wondering if the git konvi import takes the branches into account [10:47] there is a noobs guide to debhelper or I need to go through manuals? [10:47] stopped obtaining revisions at 665000 and searches up the revisions now ... at 835000 already :S [10:48] Quintasan: well, there is a very very unuseful guide in the wiki [10:48] :| [10:48] maybe it improved though [10:48] worth looking I suppose :D [10:49] simply put: you need a clean target and a binary-arch target that depends on a build an install target, where install depends on build and build on configure [10:50] you might want to stamp build though :P [10:50] lol [10:50] * Quintasan never used debhelper [10:50] sec [10:51] http://aplg.kollide.net/kubuntu/debian/rules [10:51] it's best to start reading at the bottom [10:51] http://www.gnu.org/software/automake/manual/make/Phony-Targets.html [10:52] what you need to understand is that debian/rules is for the better part a very simple make file [10:52] thus makefile paradigms appy, like, every line break also equals going back to the execution dir ... [10:53] reading it like a script might make sense though [10:53] easier to understand [10:53] so if you start at the bottom you have the phony targets [10:53] then binary which is really just a wrapper for binary-arch and binary-indep [10:54] this splitting is mostly useful when you need different stuff to be done for an arch:all and arch:any package [10:54] both "depend" on build and install [10:54] if you look at install, it depends on build [10:54] so build must be done before install [10:55] build "depending" on build-stamp and that is depending on config.status [10:55] in config.status you would run cmake .. or configure or qmake [10:56] once that is done build-stamp continues ensuring we are in the right dir [10:56] you can obtain information about debhelper scripts via their man page [10:56] e.g. man://dh_testdir [10:56] then it just runs make and creates a file build-stamp [10:57] that file is prevent useless runs of build [10:57] ...look at the clean target youll see that there the stamp gets removed [10:57] which basically means that make will only be run after a clean [10:57] so build is done, returning to install [10:58] that again runs some dh magic to ensure everything is in place for make install [10:58] one word of advice though ... for qmake in 90% of all cases it makes sense to abuse man://dh_install for installation rather than placing everything manually via the install target [10:59] once install is finished we return to binary-arch, all "dependencies" of that are now done and the binary-arch target is self is processed [10:59] here you only find a pile of debhelper scripts to build the package and place additional files [11:00] I recommend you take a look at the manpages of each of those for clearity [11:00] if everything went fine the dh_builddeb script builds the actual deb [11:00] Quintasan: all clear? :D [11:01] hmm [11:02] yup, but knowing myself I will need to package two or three things before understanding it fully :P [11:02] nothing wrong with that :) [11:03] kformula was remove from koffice 2.0.82 again, but it was part of 2.0.81. How can I get it deinstalled when upgrading to 2.0.82, make koffice-kde4 replace kformula (>=2.0.81) ? [11:04] neversfelde: you don't [11:04] neversfelde: do we deploy via PPA? [11:05] apachelogger: yes [11:05] it is in the backports beta ppa [11:05] hm [11:06] neversfelde: kformula will get nuked when the koffice libs become incompatible with it [11:06] apachelogger: one question, what's the point of configure depending on configure-stamp? [11:06] neversfelde: replacing it is quite ugly TBH [11:06] there seem to be some files, which where in kformula-kde4.install that are now in kplato-kde4.install, so I will have to make kplato-kde4.install conflict with kformula [11:07] Quintasan: prevent configure from being executed when it is not necessary ... all stamp targets are only to prevent that the containing stuff gets executed if the appropriate stamp file is available [11:07] hmmm [11:08] neversfelde: oh, then replace and conflict kformula with kplato [11:08] apachelogger: k [11:08] so things like ./configure or qmake dooble.pro should go to configure-stamp? [11:08] thank you [11:09] Quintasan: depends on the other available targets, but generally yes :) [11:09] * apachelogger hg converts konvi git to hg :D [11:10] that is so freaking insane [11:10] best thing ever: you can't bzr fast-export branches with tags containing an epoch, git does not appear to be liking that very much and breaks on fast-import [11:12] WTF [11:13] apachelogger: debian/rules:21 *** missing separator (TAB instead 8 spaces?). Stop [11:13] I used TAB there :| [11:18] * Quintasan is going crazy [11:20] YAY [11:20] ITS ALIVE [11:23] apachelogger: exacly as I though, clean did not depend on anything, adding configure and build to depends helped :P [11:27] Quintasan: why would you have to build before being able to clean? [11:28] that does not make sense [11:28] that is like: you have to fill your harddisk before you can format it [11:30] the quasselVskonvi codeswarm does not take konvi-kde3 into account while it was branched from trunk [11:30] still quite interesting [11:30] at the time konvi went to the KDE 3 branch quassel grew quite a bit :D [11:40] apachelogger: I looked at it like this [11:40] apachelogger: a) you start with pure source b) you configure (+Makefile) c) then you build it(+ *.o files) d) you clean Makefile and *.o files === mgraesslin_ is now known as mgraesslin [11:42] Quintasan: you cannot assume that the intial source is purged [11:42] Quintasan: always run clean [11:42] clean must work anyway [11:42] otherwise the package will get rejected [11:42] apachelogger: but it fail all the time [11:42] so fix it :P [11:43] says - "no rules to make 'clean'" [11:43] no matter what I put there [11:43] batpaste your rules file [11:43] http://pastebin.com/f2a3868e5 [11:45] make clean only works when there is a make file ;) [11:45] should be distclean IMHO anyway [11:45] [ ! -f Makefile ] || $(MAKE) distclean [11:45] unless makefile does not exist run make distclean [11:46] ah k [11:46] also, if you prefix any command with - an error will be ignroed [11:46] so -make distclean [11:46] would have about the same affect [11:46] but I lol'd since built deb file did not contain any binaries :O [11:46] just that lintian will bitch about it :D [11:46] Quintasan: make install is defuncted [11:46] as I anticipated :P [11:47] Quintasan: use dh_install to put everything into place [11:47] $(MAKE) DESTDIR=$(CURDIR)/debian/dooble install [11:47] replace this with dh_install? [11:47] no [11:47] Quintasan: did you already take a look at the dh_install manpage? [11:48] looking [11:48] mhm [11:48] the thing with dh_install is that it needs to be run for _every_ binary package [11:49] first let me deal with clean [11:49] ill explain meanwhile :P [11:49] so it shouldnt depend on build or anything [11:49] Lure: why do you have an unlock dialogue for autologin? the whole point of autologin is there's no password [11:49] so dh_install in the install target would work for your package [11:49] but [11:49] say you add a -data package it will not work that well anymore [11:50] thus it is common to have dh_install run in the binary-* target [11:50] since those get run for every binary package anyway [11:51] so just implement an appropriate debian/install file and remove the comment from dh_install [11:51] ... it is already in your binary-arch target, just commented out :) [11:51] Riddell: autologin to decrease login time but still being somewhat secure [11:52] okay, I want to know wtf is with this error about 8 spaces instead of tabs [11:52] Riddell: like you turn on the pc and go get a cup of coffee, when you come back you just need to enter the password and can pointy click around [11:52] apachelogger: I need to put [ ! -f Makefile ] || $(MAKE) distclean in clean rule, right? [11:52] Riddell: IMHO a very minimal usecase though [11:53] Quintasan: yeah, replace your make clean with it [11:53] Quintasan: do you have spaces somewhere that cause that error :P [11:53] now I need to know where the hell the binaries are put [11:54] apachelogger: how do you set that? [11:54] there is a special setting for that in kdm [11:54] last tab of the KCM IIRC [11:54] along the no-password for login option [11:55] oh I see it [11:55] Quintasan: ../ by default [11:55] Quintasan: depends on how you build really ;) [11:55] shit, this sucks [11:55] lol [11:56] that is regular stuff no matter what is in your rules file [11:56] Binary is build in the source directory [11:57] oh [11:57] the app binary :D [11:57] * apachelogger thought the deb binary [11:57] yeah, and I need to put Icons dir somewhere [11:57] This is harder than I expected [11:58] the harder the better I always used to think :D [11:58] now I am too lazy to add more than 3 cdbs includes :P [12:00] apachelogger: http://pastebin.com/f64cdd090 <-- this is direcotry layou [12:00] I obviously need Icons, Images, Tab and Doc [12:01] also qss [12:01] but where I should put those? [12:02] Quintasan: you might not [12:02] Quintasan: first check the binary [12:02] it is common that qmake binaries have all that crap embedded in the binary [12:02] wut [12:02] so at the most you need the app icon in /usr/share/pixmaps and a desktop file [12:04] lol, with Dooble we have 3 QtWebKit browsers :D [12:04] :/ [12:05] They propably are not embedd [12:05] way too easy to creat a browser these days [12:05] we should write our own [12:08] apachelogger: the icons are not embbed into binary, there are lots of files in Icons directory [12:08] Quintasan: are none of them embedded? [12:09] apachelogger: nope, no icons until I manually select .cfg file from Icons directory [12:09] then it's getting dirty [12:09] Quintasan: inspect the code [12:09] either that thing would expect the cfg somewhere to be or you need to patch it so that it does [12:11] arg("Icons/nuvola/configuration.cfg")); [12:11] :/ [12:13] arg? [12:13] Quintasan: please paste that file [12:13] EgS: ping ping ping [12:13] I am a sexy hot biatch [12:13] lol [12:13] http://pastebin.ca/1564103 [12:14] -(~/src/git/quassel/build:$)-> grep -r -i quassel_qt.mo trace [12:14] 31892 access("/home/me/.kde/share/locale/de/LC_MESSAGES/quassel_qt.mo", R_OK) = 0 [12:15] Quintasan: that code is a bit different :D [12:15] if the settings do not yet contain iconset, then add iconset with currentdir/Icons/... [12:15] Quintasan: probably needs to be patched [12:16] well, the browser right after compiling starts with no icons [12:16] though I find it insane that upstream writes that stuff to the settings object even if the user did not set it [12:16] Quintasan: if you start it a second time? [12:17] compiling :P [12:17] anyway [12:17] just replace the code there [12:17] QString("%1/%2").arg(QDir::currentPath()). [12:17] arg("Icons/nuvola/configuration.cfg")); [12:17] to [12:17] QString("/usr/share/foobar/blah/bluh..."); [12:18] no icons even after 5 restart [12:18] yeah, needs patching :P [12:18] patchy patchy [12:18] and poky poky upstream [12:19] you might borrow the MSOD [12:19] or was it LSOD [12:19] something that sounds like a drug for sure :P [12:20] Nightrose: ping [12:20] I will put Icons dir in /usr/share/dooble right? [12:20] apachelogger: pong [12:20] Quintasan: yep [12:21] Nightrose: do you have Sput's mobile numba?... can you text him? [12:21] I am close to a nervous breakdown [12:21] wus! [12:21] oh [12:21] there he is [12:21] ;-) [12:21] Sput: I HAZ GETTEXT [12:21] ooh [12:22] what I was talking about earlier actually works [12:22] right after creation of the kapp: KGlobal::locale()->insertCatalog("quassel_qt"); [12:22] do we need to use i18n() at all? [12:22] nope [12:22] not at all [12:22] hm, and what happens with !kde? [12:22] not the tiniest bit [12:23] and! can we convert our existing translations, or are I gonna make some people extremely unhappy? :) [12:23] we can [12:23] hahahahahahha [12:23] they need proof reading and stuff [12:23] but for the better part it seems to work [12:24] apachelogger: there are even java modules for Dooble XD [12:24] basically what we do is introduce a Messages.sh like KDE extracting all qt stuff to a pot and all KDE stuff to a pot (just in case you ever want i18n for some KDE specific strings) [12:24] then at build time either create a qm (which I did not test yet...) [12:24] or a mo [12:24] depending on whether built with KDE or not [12:25] the existing translations should be convertable using gg://ts2po [12:25] which seemd to result in a prety decent .mo for me [12:25] sould's cool [12:25] though the "qm which I didn't test yet" frightens me [12:26] me too [12:26] as we already tried that one back then and it didn't work for several people including me :) [12:26] but maybe we did it differently back then [12:26] we tried mo loading in Qt I think [12:26] I thought we were building .qms? [12:26] I dunno :P [12:26] anyway, if that worked, it would be extremely cool [12:26] lets see [12:26] of course, today is UI freeze :) [12:26] for us [12:26] :) [12:27] worst that could happen is that a po and a ts need to be maintained, cross converted via ts2po or po2ts [12:27] or get the gettext --qt export fixed up [12:29] sounds interesting in any case [12:30] we prolly would get a ton more languages if that .po approach worked [12:56] apachelogger: there are some lines that look like (QString("%1/Tab/Default/search.html").arg(QDir::currentPath())); [12:57] oh man [12:57] apachelogger: should I patch them to point to /usr/share/dooble/blash? [12:57] yes === mgraesslin_ is now known as mgraesslin [13:02] Lure: ok, I know what the issue is, I'll take a look at it when I get a moment [13:03] Riddell: thanks [13:20] Riddell: btw http://aplg.kollide.net/kubuntu/swarm-09-09-12.avi [13:20] Sput: even more btw http://aplg.kollide.net/kubuntu/KonversationVsQuasselSwarm/2009-09-13_12-22-19.avi [13:22] Sput: does the core actually have translations? [13:22] yeah [13:22] that needs special treatment :S [13:22] though in the future we'll probably change that [13:22] sensible [13:22] well, except for cli options and warnings and stuff [13:23] but client-visible stuff should be generated in the client [13:23] well, I would pass cli option translations though low-level cpp gettext anyway :P [13:23] qlocale seems like a bit of an overhead for that [13:24] ok [13:24] * apachelogger notes that po->mo->klocale seems to work pretty well alright [13:24] on to qm [13:24] huh, what's this swarm stuff about? looks fun [13:26] commits over time of existence [13:26] though unfortunately I didnt get konvi in when it was in the KDE 3 branch [13:26] aah based on commit history [13:26] neat idea :) [13:26] on the other hand that illustrates how that boosted quassel :D [13:27] I am not sure how to handle that the core got translations of its own -.- [13:27] apachelogger: commits to what? [13:27] Sput: could add quassel_core.pot [13:28] Sput: always make that a qm [13:28] Riddell: version control [13:28] with a bit of fiddling it doesn't matter which VCS, nah, actuall a lot of fiddling [13:29] cor [13:30] quite useful if you want to know how many times Riddell did not manage to set his bzr nick :P [13:36] Riddell: Dooble is a PITA to deal with but I think I will finish it today [13:37] thanks for that [13:41] Quintasan: pong, hurray :) [13:43] Quintasan: are you packaging all the little blurbs it seems to quasi-depend on? (ex, the search engine in runs as default) [13:43] ryanakca: nah, just core [13:43] ryanakca: there is epic mess with configurations for icons etc [13:47] of course our patches are as buggy as always [13:47] hoorays for cant remove language [13:49] please test koffice 2.1 beta2 from the staging ppa :) [13:50] karmic only atm [14:51] this is retarded [14:52] http://pastebin.com/f50c38a87 <--- can anyone tell why it complains about missing separator? I'm using tabs there [14:53] Quintasan: can you upload the file somewhere? [14:53] also, what is the precise error [14:54] debian/rules:26: *** missing separator (TAB instead 8 spaces?). Stop [14:55] apachelogger: http://hs.quintasan.pl/rules [14:55] no [14:55] you have spaces [14:56] Quintasan: what editor are you using? [14:56] vim [14:56] or are you doing echos and cats :D [14:56] hm [14:56] Quintasan: misconfigured maybe? [14:56] dunno [14:56] vim would by default tab debian/rules [14:56] Quintasan: paste your .vimrc === Blizzzek is now known as Blizzz [14:57] http://hs.quintasan.pl/vimrc [14:58] * apachelogger notes that cmake is freaking him out today [14:58] Quintasan: set expandtab will replace tabs with spaces [15:00] thanks :O [15:54] apachelogger: (about kamoso dev talking to Kipi people) Great. I'll modify the bug to be less threatening. [16:00] apachelogger: I got a U/I freeze waiver for quassel expiring tonight, so if you have any intent to help Sput push the translations stuff through, now would be a good time. [16:01] well it is b0rked [16:01] Sput: ping ping [16:01] doesn't make any sense really [16:01] when I install current master and replace the german qm with my one built from a po the UI still stays translated [16:02]