15:04:11 <jreznik> #startmeeting kde-sig -- http://fedoraproject.org/wiki/SIGs/KDE/Meetings/2012-05-08
15:04:11 <zodbot> Meeting started Tue May  8 15:04:11 2012 UTC.  The chair is jreznik. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:04:11 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:04:17 <jreznik> #meetingname kde-sig
15:04:17 <zodbot> The meeting name has been set to 'kde-sig'
15:04:23 <jreznik> #topic Roll Call
15:04:34 <Kevin_Kofler> Present.
15:04:41 <jreznik> anybody here? than_, rdieter_work?
15:05:30 <than> Present
15:05:54 <jreznik> #chair jreznik Kevin_Kofler than
15:05:54 <zodbot> Current chairs: Kevin_Kofler jreznik than
15:06:04 <jreznik> #info jreznik than and Kevin_Kofler present
15:07:08 <jreznik> looks like only we three are ready today :(
15:09:22 <Kevin_Kofler> So, anything to discuss, agenda-wise?
15:09:37 <Kevin_Kofler> I'd say F17 release blockers, if we have anything to discuss about those…
15:10:50 <jreznik> 4.8.3 for F16?
15:11:07 <than> jreznik: i will say yes
15:11:43 <Kevin_Kofler> It's officially released by upstream, have we built it yet?
15:11:46 <Kevin_Kofler> If not, we should!
15:12:02 <rdieter> hi, sorry I'm late
15:12:10 <jreznik> Kevin_Kofler: it's in -testing right now
15:12:41 <jreznik> rdieter: hi!
15:12:47 <jreznik> #chair rdieter
15:12:47 <zodbot> Current chairs: Kevin_Kofler jreznik rdieter than
15:12:57 <jreznik> #info rdieter presen too
15:13:07 <jreznik> #topic Agenda :)
15:13:29 <than> 4.8.3 is good release, it can be pushed into stable
15:13:29 <Kevin_Kofler> OK, it's been in testing for almost 6 days, it has +2 karma and no complaints, IMHO ship it (stable)!
15:14:11 <jreznik> #topic 4.8.3 for F16
15:14:28 <jreznik> #info it's been in testing for almost 6 days, it has +2 karma and no complaints
15:15:09 <rdieter> any objections then to queue'ing stable ?
15:15:26 <jreznik> it works well for me on f16, so I'm +1
15:15:26 <rdieter> (none from me, fwiw)
15:15:53 <Kevin_Kofler> stable +1
15:16:06 <than> +1 for stable
15:16:46 <rdieter> looks like a winner, queue'd
15:17:06 <jreznik> #agreed to push a winning 4.8.3 release to stable for Fedora 16
15:17:22 <jreznik> rdieter: could you push the red button? :)
15:17:28 <rdieter> done
15:17:47 <Kevin_Kofler> Speaking of big red buttons, I queued https://admin.fedoraproject.org/updates/FEDORA-2012-6552/PackageKit-0.6.17-2.fc15 to stable too, the 2 weeks finally expired.
15:18:21 <rdieter> cool
15:20:22 <jreznik> ok
15:20:38 <jreznik> #info PackageKit-0.6.17-2.fc15 queued to stable
15:21:14 <jreznik> that's probably all for updates, isn't it?
15:21:35 <Kevin_Kofler> There's also Kile 2.1.2.
15:22:13 <Kevin_Kofler> As released by upstream, it reintroduced the issue where it wouldn't terminate properly, so I did new builds with a patch from upstream which reintroduces the line which ensures termination.
15:22:24 <Kevin_Kofler> kile-2.1.2-2.fc1[567]
15:22:31 <Kevin_Kofler> These are now queued for testing.
15:22:44 <Kevin_Kofler> (replacing -1 which was broken)
15:22:55 <jreznik> #info kile-2.1.2-2.fc1[567] queued for testing
15:23:03 <Kevin_Kofler> (The stable updates still have Kile 2.1.1 which didn't have the regression.)
15:24:47 <jreznik> ok, let's move
15:24:59 <jreznik> #fedora 17 kde blocker bugs
15:26:20 <rdieter> just one, test build : https://admin.fedoraproject.org/updates/FEDORA-2012-7448/calligra-2.4.1-3.fc17
15:26:33 <rdieter> .bug 819371
15:26:35 <zodbot> rdieter: Bug 819371 All Calligra Products segfault when opening a new doc / new from template - https://bugzilla.redhat.com/show_bug.cgi?id=819371
15:27:19 <rdieter> i've only lightly tested, i'd only seen the reported crash once before, but I've been unable to reproduce it at all with this new build so far
15:27:56 <rdieter> see also upstream bug http://bugs.kde.org/show_bug.cgi?id=298412
15:28:24 <Kevin_Kofler> OK, I hope we can get this in.
15:28:37 <jreznik> let's hope it's fixed
15:29:03 <rdieter> it's an accepted blocker already, we just need to test that it works as advertised
15:29:46 <Kevin_Kofler> Unfortunately, I don't think we're going to see a solution for the ugly purple graphics, it was deemed only NTH, so they won't be waiting for a fix, we don't know how to fix it and neither upstream nor the Fedora X11 maintainers seem to care at all.
15:29:47 <jreznik> #info Bug 819371 All Calligra Products segfault when opening a new doc / new from template accepted as blocker bug (all apps in menu has to be working and not crashing)
15:30:20 <Kevin_Kofler> Qt doesn't support 16 bpp, ajax doesn't support 24 bpp and the cirrus driver doesn't support 32 bpp, so we're stuck. :-/
15:30:21 <jreznik> Kevin_Kofler: do we want to use a patch proposed by than? so use native for cirrus?
15:31:57 <Kevin_Kofler> Are we sure the patch doesn't have nasty side effects?
15:32:06 <rdieter> sorry again, I've an appointment to go to, but I'd be ok with than's patch (pending details, I only vaguely recall how it was implemented)
15:32:28 <Kevin_Kofler> IIRC, it was running lspci | grep cirrus in an /etc/kde/env script, IMHO that's a very bad thing to do.
15:32:39 <Kevin_Kofler> The grep might match stuff which isn't a display at all.
15:32:39 <jreznik> Kevin_Kofler: it is...
15:33:03 <jreznik> with better handling of cirrus I'm ok with it
15:33:07 <Kevin_Kofler> I'm not convinced the script as written won't cause side effects on other configurations.
15:33:10 <jreznik> native still seems working quite well
15:33:32 <jreznik> Kevin_Kofler: I was a little bit scared too when I saw it for the first time
15:33:43 <jreznik> but on the other hand - native is still quite safe option
15:34:00 <than> Kevin_Kofler:  why is it bad?
15:34:20 <Kevin_Kofler> than: You grep every internal device on the computer for the string "Cirrus".
15:34:23 <than> it's only set if qt runs in kvm env
15:34:45 <than> Kevin_Kofler: i don't see problem here
15:34:53 <Kevin_Kofler> This could match completely unrelated devices.
15:34:59 <than> how can you check it?
15:35:06 <Kevin_Kofler> E.g. isn't there a brand of disk controllers called CirrusLogic?
15:35:52 <Kevin_Kofler> The PCI ID might possibly be a safer thing to match.
15:36:07 <Kevin_Kofler> Or even the whole line, with ^ and $ markers.
15:36:34 <jreznik> Kevin_Kofler: +1 for ^ and $ markers
15:36:49 <than> Kevin_Kofler: PCI ID is good, i will take a look again
15:36:52 <jreznik> it's going to be same always as there's only one existing device in kvm
15:37:15 <Kevin_Kofler> Yes, so we want to check the whole lspci line.
15:37:24 <Kevin_Kofler> That'll avoid false positives.
15:37:41 <than> Kevin_Kofler: agree it's safer
15:38:10 * jreznik does not expect anybody is running kde on real cirrus hw
15:38:21 <Kevin_Kofler> It'll still be an ugly hack, but we are unlikely to get anything more complex through for F17 Final through NTH. Time is short.
15:40:09 <Kevin_Kofler> The right solution would be to revert to 24 bpp and fix the llvmpipe problems with that, but unfortunately Fedora does not have a strong "no regressions" policy and so we cannot force them to do that. :-(
15:41:13 <than> how the bugzilla number for this issue (cirrus)?
15:41:26 <Kevin_Kofler> The blocker process is very broken because it allows causing arbitrary regressions as long as they're not blockers, and you cannot revert them because "OMG Blocker!!!!111!!!!!!!!!!!!!". :-/
15:42:02 <Kevin_Kofler> The rule should be that what used to work must keep working no matter what!
15:42:33 <Kevin_Kofler> than: https://bugzilla.redhat.com/show_bug.cgi?id=810161
15:44:07 <jreznik> #link https://bugzilla.redhat.com/show_bug.cgi?id=810161
15:44:32 <jreznik> #info we are going to solve the cirrus 16 bpp problem by using the native engine
15:45:06 <jreznik> #info the cirrus matching pattern has to be extended to match cirrus video only
15:45:13 <jreznik> ok, anything else?
15:45:26 * jreznik is also not happy how qa took care of this issue
15:45:40 <jreznik> and I hope cirrus/vnc combo is going to die soon...
15:46:14 <than> Kevin_Kofler: grep -qi "VGA compatible controller: Cirrus Logic GD 5446"
15:46:22 <than> i think it's safe enough
15:46:52 <Kevin_Kofler> I guess that's OK.
15:47:11 <Kevin_Kofler> But why not the entire line with ^ and $?
15:48:17 <than> Kevin_Kofler: i don't want to use ^ because 00:02.0 could be changed
15:48:48 <Kevin_Kofler> OK, well I guess what we have is safe enough.
15:49:05 <Kevin_Kofler> So +1 for applying it as is and filing it through NTH.
15:49:19 <than> jreznik: could you please add this workaround and rebuild?
15:49:38 <Kevin_Kofler> (I'm still very unhappy about the fix because I think it's a very wrong fix, but I'd rather have a workaround than nothing at all. :-/ )
15:50:09 <jreznik> than: I can do it but not sure I'll make it today...
15:50:28 <Kevin_Kofler> (It really sucks that ajax is unwilling to fix the mess he caused. It should be HIS job to either fix Qt for 16 bpp if he thinks it's so great or find another solution for the original blocker.)
15:50:40 * jreznik wants to enjoy free day a little bit after the linuxwochen weekend :)
15:50:47 <than> jreznik: no hurry
15:51:00 <Kevin_Kofler> than: Well, time is short. :-(
15:51:19 <Kevin_Kofler> It's only NTH, so they won't wait for us.
15:52:14 <jreznik> if anyone can do it now, feel free... otherwise I'll do it tmrw in the morning... /me has to leave after mtg
15:53:05 <than> jreznik: it's fine with me, thanks
15:53:43 <than> Kevin_Kofler: take it easy, we can fix it in f17 update :)
15:54:02 <jreznik> than: ok, tmrw morning
15:54:08 <jreznik> than: this should go to final...
15:54:19 <Kevin_Kofler> than: That'd leave the live image very broken in QEMU. :-(
15:54:22 <jreznik> or better has to go... as it's nasty
15:55:21 <jreznik> anything else we want to discuss?
15:55:26 <jreznik> 5 minutes left
16:01:42 <jreznik> ok, thanks for coming
16:01:45 <jreznik> #endmeeting