15:06:28 #startmeeting KDE SIG Meeting 15:06:28 Meeting started Tue Nov 4 15:06:28 2014 UTC. The chair is Kevin_Kofler. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:06:28 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:06:32 #meetingname kde-sig 15:06:32 The meeting name has been set to 'kde-sig' 15:06:33 * pino|work here 15:06:38 #topic Roll call 15:06:44 * Kevin_Kofler present, leading the meeting today. 15:07:05 hi 15:07:16 * jgrulich is present 15:08:12 * jreznik is around 15:09:45 * danofsatx is here 15:10:01 <-- eating oatmeal. please carry on ;) 15:10:36 * ltinkl is here 15:11:13 #chair pino|work dvratil jgrulich jreznik danofsatx ltinkl 15:11:13 Current chairs: Kevin_Kofler danofsatx dvratil jgrulich jreznik ltinkl pino|work 15:14:20 #info Kevin_Kofler, pino|work, dvratil, jgrulich, jreznik, danofsatx, ltinkl present. 15:14:23 #topic Agenda 15:14:50 One topic I'd like to bring up is the system-clucene stuff in Qt Assistant. 15:14:52 present 15:15:08 #chair than 15:15:08 Current chairs: Kevin_Kofler danofsatx dvratil jgrulich jreznik ltinkl pino|work than 15:15:08 I'd like to solve the conflict between KDE 4 and Plasma 5/KDE Applications translations 15:15:24 KF5 5.4 update 15:15:26 * heliocastro here 15:15:29 #chair heliocastro 15:15:29 Current chairs: Kevin_Kofler danofsatx dvratil heliocastro jgrulich jreznik ltinkl pino|work than 15:15:35 #info than and heliocastro also present. 15:16:00 dvratil: Let's start with your topics. 15:16:28 let's start with the KF5 update, it's just an announcement 15:16:44 OK 15:16:46 #topic KF5 5.4 update 15:17:05 KDE Frameworks 5.4.0 are currently being built in Koji, so I expect the update to be done later today, or tomorrow morning 15:17:10 Hi all. 15:17:40 I create -libs subpackages in two or three frameworks to fix multi-arch conistallability 15:18:01 and that's about all 15:18:36 #info KDE Frameworks 5.4.0 are currently being built in Koji, -libs subpackages created in two or three frameworks to fix multi-arch conistallability 15:18:50 Thanks for your work on those packages! 15:19:06 #topic Conflict between KDE 4 and Plasma 5/KDE Applications translations 15:19:06 dvratil: There's no conflicts on kglobalaccel and kglobalaccel5 ? 15:19:20 heliocastro: no, the KF5 binary is called kglobalaccel5 15:19:42 Frameworks are 99.9% coinstallable with KDE 4, the only exception being KActivities 15:19:43 dvratil: I mean, on the startup of both desktops coinstalled 15:19:58 heliocastro: you can't have both desktops co-installed 15:20:08 Ok, this answers my question 15:20:13 thanks 15:20:22 ah, kglobalaccel is provided by kde-runtime,. which you actually can (Should) have installed. 15:20:42 I haven't seen any conflict there yet, but I haven't actually looked for one. I'll check it, thanks heliocastro 15:21:07 So we need to make sure that Plasma 5 starts only the services that make sense to start in a Plasma 5 environment, and likewise for Plasma 4. 15:21:18 (Even if you have both kdelibs 4 and KF5 installed.) 15:21:22 Yes 15:21:25 exactly 15:22:03 you need KDE 4 services when you start a KDE 4 application in Plasma 5 environment 15:22:20 Most likely yes. 15:22:35 KDE 3 apps definitely still fire up the KDE 3 kded. 15:22:43 What i saw is that f you start some kde4 application, who will takes care on key shortcuts ? 15:22:50 So I'd expect kded4 to also get fired up for KDE 4 apps. 15:25:21 move on? 15:25:28 Yeah 15:25:39 Let me check one thing… 15:25:41 Nothing we can do before proper verifying 15:26:27 I don't seem to have done anything special to kglobalaccel for kdelibs3/kdebase3, did KDE 3 not have such a binary? 15:26:37 (I also don't see anything in the kdelibs3 and kdebase3 packages.) 15:26:50 probably not. And if it would, it would be using DCOP, not DBus, so no conflicts 15:27:39 So I guess we can move on to what's already in /topic. :-) 15:27:53 Still about conflicts, but for translations. 15:28:05 yep 15:28:11 Indeed, new workspace and apps stuff conflicts with kde-l10n. 15:28:28 for those who don't know: In KDE 4 all translations are released in one tarball and we package them as one package (kde-l10n-$LANG) 15:28:38 in Plasma 5 each applications and library ships it's own translations 15:28:59 Maybe change the install directory ? 15:29:03 How is this going to work in the Applications 14.12 release, which is mixed KDE 4 and 5? 15:29:18 Is kde-l10n going away? 15:29:35 Kevin_Kofler: not sure, I assume kde-l10n will contain only translations of KDE 4 applications 15:29:35 We can redirect where kde pick their translations, no ? 15:30:32 I think we should probably wait for the Applications 14.12 prereleases that are going to be tagged this week. 15:30:46 ok 15:30:59 See what upstream does to kde-l10n, before we spend time on solving a problem upstream may be solving for us. 15:31:16 ok, let's discuss this next week 15:32:22 If upstream doesn't have a good solution, IMHO, we should do what I did for kde-i18n vs. kde-l10n, which is to write some scripts to list all the files/directories that conflict, then for each, blacklist the translation that doesn't match the code we ship. 15:32:42 It's even easier now with kde-l10n vs. bundled in the package, because if the translation is bundled in the package, it's always the right one. 15:33:19 The only drawback is that this means the Plasma 5 Copr needs to carry a different kde-l10n than stock Fedora (probably just a flag set, which would be always 1 in Rawhide too). 15:34:06 But it's not worth spending our time on doing those scripts if upstream is dropping kde-l10n or reducing it to kdelibs only in 14.12. 15:35:16 Again, what halt us to just install kf5 translation in other directory ? 15:36:59 /usr/share/locale is a FHS standard directory. 15:37:18 And it's stupid to install tons of kde-l10n translations for apps which then ship their own translations. 15:37:18 but what about usr/share/local/kf5 15:37:27 So -1 to the subdirectory. 15:37:40 There's only one right translation for any given gettext domain. 15:38:08 Conflicting stuff just needs to be blacklisted (and in fact, IMHO, we should have done this weeks ago). 15:38:44 It's very easy to blacklist unwanted stuff from kde-l10n: http://pkgs.fedoraproject.org/cgit/kde-l10n.git/tree/kde-l10n.spec#n802 15:39:32 I was also thinking about introducing per-app sub-packages to kde-l10n, that would allow external packages (think Copr) to obsolete them, together with the app they are updating 15:40:42 dvratil: I can help you on that. 15:40:45 (e.g., we're blacklisting kpilot translations because kpilot was deprecated from kdepim and moved to a separate package (that has since become unmaintained upstream again, sadly), but some languages were still shipping conflicting old kpilot translations) 15:40:59 dvratil: Ewww!!! 15:41:12 kde-l10n is already per language. 15:41:28 Per-app subpackages mean we have one package per (app,language) pair, i.e. quadratically many. 15:41:32 here, sorry I'm late 15:41:41 #chair rdieter 15:41:41 Current chairs: Kevin_Kofler danofsatx dvratil heliocastro jgrulich jreznik ltinkl pino|work rdieter than 15:41:43 Kevin_Kofler: it would be...a lot, yes :) 15:41:47 I think those will be more than TeXLive! 15:42:05 So IMHO, that approach does not scale. 15:42:42 dvratil: Like this: http://svnweb.mageia.org/packages/cauldron/kde-l10n/current/SPECS/kde-l10n.spec?revision=787665&view=markup 15:42:44 In addition, there would still need to be a kde-l10n-$language metapackage that Requires the subpackages, for upgrade paths, and so you'd still have to maintain a whitelist for the Requires. 15:43:03 (The Obsoletes would not help that.) 15:43:51 * rdieter thinks some form of whitelist/blacklist is the way to go 15:43:52 I think a blacklist (or even whitelist, that can be done too with just a bit more scripting) that works on the CMakeLists.txt files (like the existing blacklisting that we already do) is the much simpler solution. 15:44:27 we're not planning on shipping both kde4 and kf5 versions of (most) apps, so shipping only one appropriately makes sense 15:44:47 It's even possible to do the blacklisting this way: http://pkgs.fedoraproject.org/cgit/kde-i18n.git/tree/kde-i18n.spec#n701 15:44:54 keeping track of that isn't fun, but neither is any other approach fun either 15:44:57 but it's better to do it in CMakeLists.txt. 15:45:23 (KDE 3 used autocrap which I was too lazy to deal with, so I did that. ;-) ) 15:47:09 do we have any bugs yet tracking the translation/conflicts ? 15:47:25 or is that something to worry about later, once plasma5 is introduced into rawhide? 15:47:59 I was suggesting to wait a week to see what upstream is doing to kde-l10n with the Applications 14.12 release. 15:48:03 its a problem for my Copr, as people can't have both KDE 4 and Plasma 5 apps localized 15:48:07 They're preparing the first prelease now. 15:48:55 but I agree, let's discuss this next week when we know how the tarballs will look like 15:49:08 I guess the only one who knows is tosky, who's not here :) 15:51:07 #info Deferred to next week, waiting for what upstream is doing to kde-l10n in 14.12. 15:51:17 #topic System CLucene support in Qt Assistant 15:51:40 So I worked on the patches, and according to rdieter, they seem to work now. 15:51:47 So, how do we proceed with those? 15:52:29 The current status is: Qt 4 uses the bundled CLucene in Fedora 19, 20 and 21. In Rawhide, I made it use the system CLucene, not sure whether anybody tested it yet. 15:52:47 I forget, is qt5-qttools-5.3.x fixed or not? iirc, no. 15:52:49 Qt 5 tries to use the system clucene09 in Fedora 19, 20 and 21, but it doesn't work. It should be fixed in Rawhide. 15:53:08 if not, let's wait until 5.4.0 lands 15:53:10 I haven't done any fixed 5.3.x builds yet, indeed. 15:53:18 But I did the patch against 5.3.2. 15:53:36 So from the Qt side, it's just a matter of replacing the patch with the one in master. 15:53:54 Kevin_Kofler: if you want to do it now with 5.3.2, I'm fine with that. otherwise, I'll handle it later with 5.4.0 15:54:27 batch updates of qt5, qt, and... one other thing depended on clucene09, zarafa ? 15:54:44 What needs to be done is merging clucene09 from Rawhide and rebuilding zarafa against it. 15:54:57 And qt5-qttools needs to be bootstrapped, because of the soname bump. 15:55:32 Another good question is, has anybody tested zarafa with the refcounted clucene09 yet? 15:55:51 It SHOULD work, but still, somebody should test it… 15:56:06 The problem is, I have no clue about Zarafa and definitely don't want to set an instance up. 15:56:12 (It's a webmail/groupware thingy.) 15:58:51 contact their maintainers when prepping the update , and ask for feedback/testing then 15:59:54 The Zarafa maintainer is Robert Scheck, who also maintains clucene09. 16:00:33 So he already knows about the plans, but I'm not sure what if any testing he has done on the Zarafa side of things. 16:06:28 rdieter: And for Qt 4, do we want to enable the system-clucene support there too? 16:06:43 (Of course, we can do it only when or after updating clucene09.) 16:07:00 I have it in Rawhide and it should be working there. 16:07:20 yes, it's worth trying at least 16:07:29 (assuming testing goes ok) 16:07:31 The advantage would be to share the same clucene09 between the Qt 4 and 5 Assistant (~1 MiB saved). 16:10:32 Kevin_Kofler: Tihs isn't the only advantage 16:10:45 single point of failure, if any bug appears 16:11:16 Kevin_Kofler: To be honest, I don't know how to test that change. 16:11:28 Well, there's still the QtCLucene wrapper layer, which is 99% copy&paste between Qt 4 and 5, but which can't be shared/unbundled (because it's built against a different Qt). 16:11:30 Kevin_Kofler: interesting, I have zarafa installed on my f20 box, and I don't see any clucene09-core dependency 16:11:48 rdieter: It's a dependency only of the indexer for search. 16:11:57 rdieter: only zarafa-search depends on clucene09 16:12:21 ah, ko 16:12:22 rsc: Well, does indexing/searching still work after you update clucene09 and zarafa? 16:12:23 ok 16:12:27 If it works, we're fine. 16:12:55 (I don't see why it wouldn't work, but you know, a software engineer's famous last words. ;-) ) 16:13:10 I'll include a new zarafa build in the qt5 copr for testing 16:15:06 rdieter: OK 16:15:23 Maybe you can also include a qt (4) build with system-clucene? 16:15:33 Kevin_Kofler: yes, will do 16:15:38 (Just use the SRPM from Rawhide.) 16:20:14 #topic Open discussion 16:20:22 So, is there anything else to discuss? 16:20:26 We're already over time. 16:21:29 Looks like not. :-) So, thanks all for coming, goodbye! 16:21:31 #endmeeting