kde-sig
LOGS
15:01:37 <jreznik> #startmeeting kde-sig -- https://fedoraproject.org/wiki/SIGs/KDE/Meetings/2011-01-04
15:01:37 <zodbot> Meeting started Tue Jan  4 15:01:37 2011 UTC.  The chair is jreznik. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:37 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:01:54 <jreznik> #meetingname kde-sig
15:01:54 <zodbot> The meeting name has been set to 'kde-sig'
15:02:03 <jreznik> #topic roll call
15:02:14 <jreznik> who's present today?
15:02:36 <rdieter> here
15:02:52 <Kevin_Kofler> Present.
15:04:31 <jreznik> #chair rdieter Kevin_Kofler
15:04:31 <zodbot> Current chairs: Kevin_Kofler jreznik rdieter
15:04:57 * jsmith lurks just for the fun of it
15:05:20 * rdieter hands jsmith popcorn
15:05:27 <jreznik> :)
15:05:52 <jsmith> rdieter: Thanks :-)
15:06:17 <jreznik> #info jreznik rdieter Kevin_Kofler present, jsmith lurks with popcorn
15:06:43 <jreznik> #topic agenda
15:07:10 <jreznik> any New Years topic? except the New Year?
15:08:01 <Kevin_Kofler> Various status updates?
15:08:18 <Kevin_Kofler> E.g. any news about libreoffice-kde?
15:09:06 <rdieter> phonon for one, upstream split the core library and the backends.  will require packaging/reviewing backends
15:09:19 <Kevin_Kofler> Oh right.
15:09:55 <jreznik> Kevin_Kofler: libreoffice-kde is more topic for ltinkl, he's not here...
15:09:56 <Kevin_Kofler> Speaking of the default backend, Dragon Player still relies on xine-lib to show DVD menus and it looks like Phonon is lacking one function which would be needed there (a way to trigger the DVD menu).
15:09:57 <rdieter> I ran into that after Sho asked me to package a newer phonon snapshot
15:10:19 <jreznik> ok, so lets start with Phonon
15:10:20 <rdieter> Kevin_Kofler: darn, I thought apachelogger had fixed that
15:10:30 <jreznik> #topic Phonon
15:10:50 <jreznik> rdieter: could you give us some overview what happens in Phonon land?
15:10:57 <Kevin_Kofler> He's done some DVD menu work in Phonon-GStreamer, but the API to trigger the menu is missing.
15:11:11 <rdieter> to sum up:  phonon upstream has moved to git.kde.org and the core library and backends are now split
15:11:16 <Kevin_Kofler> I wanted to port Dragon Player to the new API and noticed there wasn't any. :-/
15:11:29 <Kevin_Kofler> I asked apachelogger and he said it just isn't there yet.
15:11:30 <rdieter> http://apachelog.wordpress.com/2010/12/23/the-phonon-family-is-moving/
15:13:06 <rdieter> Kevin_Kofler: suckage.  so... if we want to ship f15 xine-less, we'll lose out on that (in dragonplayer anyway)
15:13:24 <Kevin_Kofler> Yes, unless something happens there…
15:13:52 <rdieter> I guess I'm willing to live with that (for now)
15:14:15 <Kevin_Kofler> apachelogger now knows it's needed.
15:16:14 <rdieter> ok, I guess that's about it from phonon-land then.
15:16:28 <rdieter> I'll finish working on the phonon repackaging, and submit some reviews
15:17:43 <jreznik> rdieter: ok, ping me, and I'll do reviews
15:17:55 <rdieter> cool, will do
15:18:02 <jreznik> #action rdieter to finish working on the phonon repackaging, and submit reviews
15:18:18 <jreznik> #action jreznik to review rdieter's reviews
15:20:56 <jreznik> so that's all for Phonon
15:21:27 <Kevin_Kofler> So FYI we have KOffice 2.3.0 in Rawhide now, plus a backported patch to support the new libwp*.
15:21:28 <jreznik> [16:08] <Kevin_Kofler> Various status updates?
15:21:34 <jreznik> which ones?
15:21:44 <Kevin_Kofler> Well, that was one. :-)
15:21:44 * jreznik was completely out during Holidays :(
15:21:52 <jreznik> Kevin_Kofler: you were faster :)
15:22:02 <Kevin_Kofler> rdieter did the upgrade and I did the libwp* backport.
15:22:11 <jreznik> #topic Various status updates
15:22:37 <jreznik> #info we have KOffice 2.3.0 in Rawhide now, plus a backported patch to support the new libwp* by Kevin_Kofler and rdieter
15:23:37 <Kevin_Kofler> The new KPK for F14 hit stable, as well as a Qt fix for a tree view crash affecting KPK.
15:23:51 <rdieter> speaking of koffice, another related item... kdab has (finally) released a standalone kdchart, when I find time, I'll finish up packaging that up too (and then build both koffice and kmymoney against it)
15:24:20 <Kevin_Kofler> Yes, that'd be great!
15:25:01 <Kevin_Kofler> What do we do about the Lokalize mess (the crash is fixed but now there's junk appended instead).
15:25:38 <Kevin_Kofler> Do we use the workaround from Lokalize upstream? (Is it already in 4.5.5?) And what about Psi, is that hit by the "junk appended" issue as well?
15:25:54 <Kevin_Kofler> I'm considering to just revert the performance optimization in Qt which caused that regression.
15:25:58 <Kevin_Kofler> I'd rather have slow, working code.
15:26:20 <Kevin_Kofler> There have been 2 followup crash fixes to that "optimized" code and it's still broken!
15:26:36 <jreznik> Kevin_Kofler: +1
15:26:49 <rdieter> Kevin_Kofler: that's the lokalize upstream workaround?
15:27:00 <rdieter> err, s/that's/what is/
15:27:35 <Kevin_Kofler> IIRC, the upstream workaround is a hack inside Lokalize where they use some convoluted process to "select all" instead of the normal method.
15:27:41 <Kevin_Kofler> See the upstream bug.
15:27:55 <Kevin_Kofler> https://bugs.kde.org/show_bug.cgi?id=249373
15:28:25 <Kevin_Kofler> I haven't checked whether that was committed to 4.5.5 or only 4.6.0.
15:29:12 <Kevin_Kofler> jreznik: The thing is, that optimization was considered a bugfix by Nokia (otherwise they wouldn't have applied it to a point release).
15:29:31 <Kevin_Kofler> This trend really worries me, they keep doing performance optimizations which break things.
15:29:33 <rdieter> and for reference, http://bugreports.qt.nokia.com/browse/QTBUG-15659
15:29:41 <Kevin_Kofler> They care about speed above quality these days. :-(
15:29:55 <jreznik> Kevin_Kofler: optimalizations are needed but should not break things...
15:31:26 <Kevin_Kofler> Another big problem is that upstream Qt keeps asking for non-KDE testcases each time they break something in KDE.
15:32:00 <rdieter> not necessarily non-kde testcases, just minimal test-cases (non-trivial test-cases are obviously harder for them to use)
15:34:22 <jreznik> that makes sense
15:35:30 <rdieter> I guess we have 2 options:  just use localize workaround   or revert the qt optmization
15:37:08 <jreznik> does it affect other apps (Kevin mentioned Psi)...
15:37:10 <jreznik> ?
15:37:25 <jreznik> in that case, we should revert the qt optimalization
15:38:04 <Kevin_Kofler> The crash was also found in Psi, I don't know whether the "junk appended" issue affects it, too.
15:38:12 <Kevin_Kofler> See http://bugreports.qt.nokia.com/browse/QTBUG-15857
15:40:44 <rdieter> ok, revert +1
15:41:35 <jreznik> revert +1
15:42:28 <Kevin_Kofler> Of course we need to test whether that optimization is really the patch to blame, but I'm fairly sure it is.
15:44:39 <jreznik> Kevin_Kofler: could you try it?
15:46:48 <Kevin_Kofler> I guess I can try to build it, then see if I can test it.
15:47:20 <Kevin_Kofler> Of course, the uphill battle will be convincing upstream Nokia that they should revert their broken optimization.
15:48:00 <Kevin_Kofler> Hmmm, that said, in the upstream reports, the Lokalize author suggests a possible fix, which looks to be a 1-line fix.
15:48:14 <Kevin_Kofler> (fix which doesn't require reverting the optimization)
15:48:18 <Kevin_Kofler> I guess we need to try that too.
15:48:34 <Kevin_Kofler> But I think he hasn't tested it, he just says it "should" work.
15:48:57 <jreznik> but then you have to fix Psi and whatever affected
15:49:05 <Kevin_Kofler> jreznik: 1-line fix to Qt
15:49:11 <jreznik> Kevin_Kofler: ok
15:49:21 <jreznik> sounds even better
15:49:48 <Kevin_Kofler> "the problem with current fix is that select() method can fail, it doesnt check for return value of movePosition() calls, which is false on a non-valid cursor
15:49:48 <Kevin_Kofler> my guess is that using setPosition(0) instead of movePosition(Start) should eliminate the problem"
15:50:36 <Kevin_Kofler> What annoys me a bit is that they're patching over the bugs repeatedly instead of just reverting the regressing optimization, which would IMHO be a much safer action on a release branch.
15:50:54 <Kevin_Kofler> I guess they really need that stuff to be fast for their mobile targets.
15:51:14 <Kevin_Kofler> But if that bug is really 1 line to fix, we should probably do that.
15:51:45 <Kevin_Kofler> I think what I'll try is 1. the suggested 1-line fix and (if that doesn't work) 2. reverting the optimization.
15:52:06 <Kevin_Kofler> (And if that doesn't work either, I'll have to give up.)
15:53:38 <jreznik> makes sense
15:58:47 <jreznik> #action Kevin_Kofler to try the 1-line fix, then reverting optimalization
15:58:57 <jreznik> ok that's probably all for this meeting
15:59:03 <jreznik> thanks
15:59:21 <rdieter> agreed
15:59:42 * jreznik is going to prepare meeting summary
15:59:51 <jreznik> #endmeeting