14:07:12 #startmeeting KDE SIG Meeting -- https://fedoraproject.org/wiki/SIGs/KDE/Meetings/2009-09-08 14:07:12 Meeting started Tue Sep 8 14:07:12 2009 UTC. The chair is rdieter. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:07:12 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:07:19 #topic Init 14:07:22 who's present today? 14:07:46 Present. 14:08:22 than, jreznik, ltinkl: ping 14:08:28 * jreznik is here 14:08:37 present 14:08:47 here 14:08:50 No idea where ltinkl is, I pinged him on #fedora-kde. 14:09:27 alrighty 14:09:30 #topic Agenda 14:09:37 any other agenda items for today ? 14:09:56 have phonon, kde-4.3.1... so far 14:10:13 maybe red hat developers conference... just in the end 14:10:26 ok 14:11:05 + constantine kde theme 14:11:23 #topic switching back to standalone Phonon and Phonon-xine, as Qt's Phonon and its backends will not be updated anymore 14:11:45 references, http://doc.trolltech.com/4.6-snapshot/phonon-module.html , http://doc.trolltech.com/4.7-snapshot/phonon-module.html , http://labs.trolltech.com/blogs/2009/09/03/multimedia/ 14:12:02 Kevin_Kofler: was this yours? 14:12:21 Yes, Sho_ pointed me to this stuff on #fedora-kde. So let me sum up the situation: 14:12:35 So the breaking news is that Qt is planning to phase out Phonon in Qt in favor of something written by the mobile team (Symbian developers and/or former Qtopia developers). 14:12:58 They're apparently not planning to update their Phonon anymore, and they already don't support features newer than Phonon 4.1 (!) in their backends. 14:13:00 " phase out Phonon" isn't exactly true, but go on... :) 14:13:00 Current in Phonon 4.3! 14:13:12 "Qt Backends are currently developed for Phonon version 4.1. The Phonon project has moved on and introduced new features that the Qt Backends do not implement. We have chosen not to document the part of Phonon that we do not support. Any class or function not appearing in our documentation can be considered unsupported." 14:13:22 Even the 4.7 snapshot says that, with unchanged version number. 14:13:40 And some of the comments on that QtMobility blog post are saying that Qt is deprecating Phonon. 14:13:52 So I think relying on Qt's copy of Phonon is not going to be viable in the long run. 14:13:53 but that multimedia framework works above the phonon 14:13:55 yeah, that's just one person saying that. 14:14:20 And I have big worries about the future of Phonon-GStreamer too, it seems according to Qt's own documentation that it's already incomplete. 14:14:29 still, I support the idea to move back to standalone phonon, that's where the upstream development happens 14:14:30 And they probably aren't going to add anything to it. 14:14:51 But in upstream Phonon, the main backend is the xine-lib one. 14:15:09 GStreamer support was contributed by Trolltech and they don't seem to care about it anymore. 14:15:09 it's not a good idea to have two phonons, two not finished backends etc... 14:15:40 moin 14:15:49 For example, the new version has support for decoding single frames used for video preview, the xine-lib backend already has it, the GStreamer one only "almost" according to a Phonon developer's comment on that blog. 14:15:51 and continuing to use phonon-xine backend. sandsmark (new phonon lead developer) said point-blank that xine is superior, and gets more attention/love 14:16:36 but from our perspective, gstreamer is better as it's much more better supported in Fedora 14:16:39 And funnily, the QtMobility folks claim Phonon is no good because it doesn't have those exact features they aren't interested in merging / writing the backend for. :-/ 14:16:48 I'm told by the Phonon developers that the GStreamer backend is of poor quality, and advertises to support features it doesn't really support 14:17:01 (single frame decoding was quoted, for example) 14:17:35 It's quite likely that it doesn't really support everything it's supposed to, given that it was written for 4.1 and doesn't support anything newer according to Qt's documentation. 14:17:46 (see quote I posted) 14:19:35 They also complain about lack of capturing support in Phonon, well that too is planned, but nobody is caring about writing it. :-/ 14:20:06 And sadly, they think reinventing the wheel with those QtMobility classes is the way to go. :-( 14:20:33 the latest is that there might still be some interest to do feature development on phonon from nokia's side. 14:20:51 as for not caring ... well matthias is overloaded with his thesis work 14:21:32 I don't take switching back to standalone phonon and phonon-xine lightly, it'll suck and be painful again to get the packaging right (again). we could consider these 2 items separately, if folks like. 14:21:48 They also threw away mkretz's existing partial work (on a work branch) on a low-level (PCM) sound API and reinvented the wheel in Qt 4.6 with the QtMultimedia module. 14:22:00 * Sho_ personally thinks that both are good ideas 14:22:09 The Phonon bugfixing being done right now is being done in KDE 14:22:14 And the Xine backend is better. 14:22:17 I think we just need to revert the changes we did. 14:22:32 Phonon-xine will automatically be preferred if we ship the standalone Phonon. 14:22:42 Sho_: maybe it's better but not in Fedora... 14:22:43 Kevin_Kofler: mostly, yeah, not sure if an Epoch is needed 14:22:46 (same SRPM, shorter name) 14:22:48 Kevin_Kofler, i don't think it's good idea revert it 14:23:06 We'd just have to remove the priority patch which takes effect when both are installed. 14:23:25 jreznik: the problem is that it even if gstreamer is super in fedora, when the phonon backend sucks it makes for a bad user experience 14:23:27 let's not get ahead of ourselves here... there's some decisions to be made first 14:23:49 than: We can't ship a frozen Phonon forever, stuff will require a current one. 14:23:56 And it seems Qt is no longer updating it. 14:24:00 Sho_: but still not complete and bad gstreamer backend works better than xine, for me, ltinkl... 14:24:16 hm, the xine backend is working fine for me 14:24:20 Sho_, where is the problem with gstreamer.backedend 14:24:42 than: dunno actually, but the phonon developers are very down on it 14:24:50 i don't see any problem with gstreamer.backend 14:25:19 jreznik, than : phonon upstream says so. I believe them. further, they're willing to continue work on making xine backend better (and consider gstreamer a largely lost cause) 14:25:20 The Phonon developers said Qt hasn't changed anything in Phonon since 4.5.0, or at the very least not merged any of their changes. 14:25:48 And the people who maintained the GStreamer backend are the very ones (Nokia) who don't care anymore. 14:25:49 maybe after polkit work we can aim on gstreamer backend... but I'm not sure it's worth as there are plans to rewrite gstreamer... 14:25:56 Sho_, it don't see any problem to switsch back to xine-backend if Trolltech may drop phonon support in the future 14:26:08 jreznik: That just makes GStreamer even more of a dead end. 14:26:54 it's now too early to switch it to xine-bacnkend for me 14:26:56 anyway, proposal 1: Which back to stanalone phonon. any opposition to this ? 14:27:06 rdieter, no 14:27:41 No, but we'll need some solution for the circular Qt<->Phonon dependency (due to QtWebKit). 14:27:47 the problem is under phonon - gstreamer and xine are not showcase of best piece of software :( 14:27:51 That causes no ends of trouble. 14:28:04 sure, modulo packaging details. 14:28:15 Another problem is qtconfig building stuff only if Qt's Phonon-GStreamer is being built. 14:28:15 we want one phonon, not too... maybe angry mail to nokia and kde upstream? 14:28:45 Kevin_Kofler: we'll likely have to build it, then omit it from packaging 14:28:54 eww, yes, but oh well 14:29:16 proposal 2 (likely more controversial): For F-12, default to phonon-xine 14:29:32 rdieter, it's not recommend 14:29:48 It is, by the people actually working on Phonon. 14:30:00 it should be said that there is ongoing discussion with nokia right now about what they intend to do with phonon, and there's the possibility of a multimedia meeting (real world) on the table 14:30:11 Also by the Amarok developers, they say Phonon-GStreamer is still as buggy as ever for Amarok. 14:30:14 rex and me are the only guys with access to that thread right now 14:30:20 so maybe it's too early to make any decisions 14:30:35 than, jreznik should be on kde-ev list 14:30:37 Sho_: +1, let's wait what will happen 14:30:39 Sho_: The problem is that the F12 final freeze is coming really soon. 14:30:48 rdieter: is it open list? 14:30:56 Sho_, yes, it agree with you it's too early to make decition 14:31:08 jreznik: I thought you were in kde-ev ? (sorry, if not, you should be!) 14:31:18 for the immediate future though i agree it seems you're going to get more bugfixes from us (kde) than nokia 14:31:32 than: It's never too early to revert to something working. 14:32:48 Kevin_Kofler, at the moment i don't see any advantage to switch back to xine 14:33:15 we didn't clearly state it, and that was partly a process failing, but I considered the -gstreamer default experiment to be successful only if -gstreamer ended up being *clearly* superior 14:33:18 Phonon upstream and Amarok upstream both say Phonon-GStreamer is crap. 14:33:25 if we know Qt drop the phonon support in Qt, we always can swictch back to xone 14:33:41 Kevin_Kofler, it's not true 14:33:42 And I don't think people actually ran my Phonon-GStreamer test plan. 14:33:47 I have little to no test returns. 14:33:49 and by the criteria, the experiment has failed 14:33:55 Kevin_Kofler, i don't see any problem here 14:34:03 Kevin_Kofler, i did 14:34:20 i did your Phonon-GStreamer test plan 14:34:40 for posterity, can someone post a link for said test plan? 14:35:01 https://fedoraproject.org/wiki/SIGs/KDE/Phonon-GStreamer_Test_Plan 14:35:07 and it works fine with your Phonon-GStreamer test plan 14:35:24 than: better than -xine ? 14:35:28 yes 14:35:40 interesting, how so? 14:35:54 for me it's working better too... sometimes xine backend does not play any music, amarok is just skipping 14:36:12 hm works just fine for me 14:36:23 it never happend with gstreamer, that was why I switched (I was xine fan :D) 14:36:36 jreznik, +1 14:36:50 it's same reason for me 14:37:01 rdieter: What was the tracker bug for Phonon-GStreamer again? Any open issues there? 14:37:03 seems to me that reliablity tests of each backends seems to be even. works for some, not for others. no definitive answer why (backend issues, driver/pulseaudio interaction) 14:37:42 rdieter: yes, it's more than just backend - pa, drivers, weather 14:38:54 https://bugzilla.redhat.com/show_bug.cgi?id=513418 14:38:55 Bug 513418: medium, low, ---, than, NEW, phonon-backend-gstreamer tracker 14:39:32 Looks like no open issues are there. 14:39:51 rdieter, it's already fixed in current qt-phonon 14:39:57 I think I remember somebody complaining about Phonon-GStreamer issues, but unfortunately, there's no bug report, at least not on the tracker. 14:40:17 So I have no idea whether the complaints are even still valid. 14:40:52 most users still have old phonon setting in 14:40:53 I use Dragon and Amarok a lot, and I've had no problems with the xine backend either on Gentoo (more than a year of usage) nor on Fedora (a little over a week) 14:41:06 though Gentoo's xine maintainer is an upstream xine developer, so Gentoo has good xine packaging 14:41:13 I was ok with using -gstreamer default too, until I had a couple of conversations with sandsmark. 14:41:47 and disregarding his opinion/recommendation here would be a bad idea, imo 14:42:05 Sho_: could you try our -gstreamer test plan? 14:42:26 fwiw, I think we should reproduce the test plan for both backends. 14:42:28 jreznik: I have no rawhide install :/ 14:42:34 FYI, for the "weird mimetypes" test, it's normal if MIDI fails. 14:42:38 rdieter: +1 14:42:39 xine-lib does not support MIDI. 14:42:49 It should support all the other weird stuff though. 14:42:50 Sho_, please remove the old seeting in Trolltech.conf before tyring the -gstreamer test plan 14:43:28 Well, first he should test if everything passes with -xine. 14:44:26 amarok plays fine the midi file with phonon-gstreamer-backend 14:44:47 GStreamer supports MIDI, that's one point for GStreamer. 14:44:58 Kevin_Kofler, i know 14:45:05 anyway, let's move on... can't use up the whole hour here... looks like we've more feedback to get 14:45:30 Kevin_Kofler, it's no reason for me to switch back to xine now 14:45:35 So gather feedback, discuss again next week? 14:45:49 or continue to hash it out in #fedora-kde later. :) 14:46:05 Maybe getting the Phonon and/or Amarok folks to come up with a list of unfixed Phonon-GStreamer issues could help? 14:46:08 but postpone final decision 14:46:22 #topic KDE-4.3.1 state 14:46:24 It's always better to have specifics than vague "it's broken" statements. 14:46:41 Kevin_Kofler: sure, whatever it takes to help convince the nay-sayers. :) 14:46:56 For KDE 4.3.1, updates are queued for updates-testing now. 14:47:03 I queued the core stuff, svahl the extragear stuff. 14:47:21 and there's an updates push underway started a few hours ago 14:47:46 OK, so it should be in testing soon. 14:47:51 any other comments ? 14:48:09 * Sho_ comments: "myyyy preeeeciousssss 14:48:14 " 14:48:38 14:48:40 +1 14:48:59 #topic constantine-kde-theme 14:49:01 jreznik: ? 14:49:23 yep, I'm here - please test it 14:49:51 I plan some changes but this should be near the final one 14:49:53 * rdieter has constantine-kde-theme-11.90.2-2 installed, nice. kdm still missing wide screen background here 14:50:02 I can help fix/patch that soonish 14:50:22 rdieter: strange, it's ok on my widescreen laptop... what's your resolution? 14:50:27 1280x800 14:50:41 * jreznik has 1280x800 too... 14:50:49 Maybe you have some -extras package installed? 14:51:03 shouldn't matter, but I'll take a closer look after meeting 14:51:33 ok, thanks 14:51:46 kde-settings should be updated to use constantine 14:52:07 Yeah. 14:52:10 I can do that too, sure. 14:52:17 unless someone beats me to it 14:52:23 Thu 2009-09-17 is deadline for final splash screens 14:52:54 I can do it, just mentioning it 14:53:02 #topic redhat developers conference 14:53:03 jreznik: ? 14:53:58 we have devel conf in brno this week and we're going to have talk with ltinkl, somethink we should mention? I hope Kevin_Kofler will come too :-) 14:53:59 * rdieter is looking for a URL to post 14:54:00 https://fedoraproject.org/wiki/DeveloperConference2009 14:54:04 ah. :) 14:54:36 I'm thinking of coming, but I have no train tickets yet and no accomodation for the night yet. As organized as ever. ;-) 14:54:53 Kevin_Kofler: accomodation should not be problem 14:55:15 Our manager asked me if I need room for you or I can offer my flat ;-) 14:55:30 but you should decide asap 14:56:41 One more thing: there are 2 kdebluetooth bugs filed, one for missing deps, one for bugfixes from SVN. 14:56:44 We should take care of those. 14:57:08 We'll just need to coordinate the push with the 4.3.1 push or we'll end up pushing stuff built against 4.3.1 before 4.3.1, which is always a source of trouble. 14:58:17 Anything else to discuss? We're running out of time (less than 2 minutes left). 14:58:37 that should be it, I think 14:58:47 thanks everyone 14:58:47 I could bring up the Konvi support link thing .. 14:58:56 but, 2 minutes is a bit short :-) 14:59:07 #fedora-kde for continuing discussion is fine. 14:59:10 #endmeeting