15:06:38 <rdieter> #startmeeting kde-sig
15:06:38 <zodbot> Meeting started Tue Feb 17 15:06:38 2015 UTC.  The chair is rdieter. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:06:38 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:06:40 <rdieter> #meetingname kde-sig
15:06:40 <zodbot> The meeting name has been set to 'kde-sig'
15:06:44 <rdieter> #topic roll call
15:06:46 <tosky> o/
15:06:51 <jgrulich> \o
15:06:53 <rdieter> hi all, friendly kde-sig meeting, who's present today?
15:07:01 <jgrulich> hello
15:07:02 <Kevin_Kofler> Present.
15:07:38 <pino|work> o/
15:07:40 * satellit liistening
15:07:49 <than> present
15:08:22 <rdieter> #info rdieter tosky jgrulich Kevin_Kofler pino|work satellit than
15:08:38 * jreznik is here
15:08:40 <rdieter> #info danofsatx, ltinkl send regards
15:08:42 <rdieter> #info jreznik present
15:08:53 <rdieter> #chair jreznik tosky jgrulich Kevin_Kofler pino|work satellit than
15:08:53 <zodbot> Current chairs: Kevin_Kofler jgrulich jreznik pino|work rdieter satellit than tosky
15:08:57 <rdieter> #topic agenda
15:09:11 <rdieter> alright agenda, what to discuss today, topic suggestions?
15:09:50 <jreznik> f22 theming + plasma 5 change progress (deadline is in one week)
15:09:59 <dvratil> hi :)
15:10:08 <rdieter> #info dvratil present
15:10:55 <Kevin_Kofler> Also how to deal with Applications 15.04 and Plasma 5.3.
15:11:44 <Kevin_Kofler> See https://fedoraproject.org/wiki/Releases/22/Schedule for how the schedules (don't) align.
15:11:47 <rdieter> ok, that's a good start...
15:11:59 <rdieter> #topic f22 theming + plasma 5 change progress
15:12:03 <rdieter> jreznik: ?
15:12:24 <Kevin_Kofler> Wouldn't that better be split into 2 topics?
15:12:41 <jreznik> I'd say f22 theming is part of overall plasma 5 change
15:12:57 <jreznik> but theming was discussed a bit on #fedora-kde earlier today
15:13:21 <jreznik> we should make sure we will be ready, check how the old wp works, be prepared
15:13:25 * marcdeop is here...
15:13:40 <rdieter> #info marcdeop present
15:13:42 <jreznik> and reminder - the change has to be in MODIFIED state next Tuesday, it means testable
15:14:21 <Kevin_Kofler> It's already testable, isn't it?
15:14:34 <Kevin_Kofler> Unless GCC 5 broke it, but then that's not OUR fault.
15:14:37 <Kevin_Kofler> It worked before GCC 5.
15:14:50 <Kevin_Kofler> And GCC 5 also breaks KDE 4.
15:14:51 <rdieter> important pieces that need more work:  adapting comps, kickstart/live-image
15:15:42 <rdieter> any other priorities we should be working on?
15:16:04 <rdieter> besides the already-mentioned artwork/theming
15:16:12 * marcdeop wonders what he can help in....
15:17:20 <rdieter> .bug 1135103
15:17:23 <zodbot> rdieter: Bug 1135103 Plasma 5 Tracker - https://bugzilla.redhat.com/1135103
15:17:45 <rdieter> also have ^^ , plasma5 blocker, anything worth highlighting there?
15:19:11 <rdieter> I'll likely be looking into comps (and kickstart) myself this week
15:19:19 <Kevin_Kofler> For comps, we should also check the F20 and F21 comps for the package splits done lately.
15:19:29 <Kevin_Kofler> E.g. kmenuedit split out from kde-workspace.
15:19:30 <rdieter> jreznik: any eta on design team wallpaper selection?
15:20:11 <jreznik> it may take some time, so maybe what I'd do is to use f21 wp to test it/prepare package
15:20:35 <rdieter> shrug, I think we'll just stick to plasma5 upstream default until something is ready (it's less work)
15:21:06 <rdieter> but I suppose preparing something won't hurt either
15:21:10 <jreznik> rdieter: it may work for alpha as it's defined as "wp has to be different to f21"
15:21:27 <Kevin_Kofler> Shipping the Plasma 5 default, yes, but that doesn't preclude working with the F21 wallpaper to have the processes ready.
15:21:51 <Kevin_Kofler> But no, we don't want to ship with the F21 wallpaper as the default, that's sure.
15:22:23 <Kevin_Kofler> (but making the Fn, n<22, wallpapers "just work" when installed from the repo would still be nice)
15:22:40 <rdieter> Kevin_Kofler: that's not a blocker though, we could fix that anytime, including after GA
15:22:41 <Kevin_Kofler> (Way too much artwork was lost because the format changed and nobody cared about the old artwork.)
15:22:42 <jreznik> maybe if I'd have some spare time, I can take a look on other QML theming part
15:23:09 <Kevin_Kofler> rdieter: Of course not, but having the processes ready to make the F22 wallpaper work when it comes out IS important.
15:23:20 <Kevin_Kofler> So jreznik's idea to try with the F21 wallpaper is not bad.
15:23:26 <Kevin_Kofler> (We just don't want it as the default.)
15:23:50 <Kevin_Kofler> What's the definition exactly? Different from THE previous release? Or from ALL previous releases? :-)
15:23:54 <rdieter> sure, anyone willing to take the artwork/wallpaper task ?
15:24:20 <Kevin_Kofler> I ask because if it's only from THE previous release, we could ship the F20 wallpaper, or some nice older one like Solar or Verne. :-)
15:24:35 <jreznik> Kevin_Kofler: previous, so it's clear you're running f22 not f21
15:24:58 <Kevin_Kofler> So going back to n-k, k≥2 would be valid? :-)
15:25:22 <rdieter> we could pull a gnome3 and insist on shipping the upstream default (once) :-/
15:25:49 <Kevin_Kofler> We will be doing that by default anyway until Design comes up with something.
15:25:50 <rdieter> (fallback plan if no one's able to work on it :) )
15:26:31 <Kevin_Kofler> We can't package something that doesn't exist (but as jreznik said, we can try with the older wallpapers).
15:26:54 <jreznik> Kevin_Kofler: and then it will be super easy to switch the trigger
15:27:04 <Kevin_Kofler> Right, that's the point.
15:28:00 <rdieter> just quick fyi, major milestone yesterday resolved a couple important gcc5 issues.  fixed qt4 FTBFS and one qt5-related crasher (caused by gold linker)
15:28:28 <jreznik> good to hear and thanks for hunting it
15:28:44 <rdieter> thanks mostly to smani
15:28:53 <rdieter> he found that qt5 one
15:29:22 <rdieter> anyway, we're back down to the startkde can't sync dbus , and same/old qt5 multiple display issue(s)
15:30:00 <Kevin_Kofler> Has anybody tried reverting dbus to an older build, to see whether it's dbus's fault or not?
15:30:15 <rdieter> original reporter on that bug said it happened with the older dbus too
15:30:23 <rdieter> (that was the original report)
15:30:43 <rdieter> orignal report was with 1.8.14
15:31:26 <rdieter> .bug 1191171
15:31:28 <rdieter> that one
15:31:29 <zodbot> rdieter: Bug 1191171 "Could not sync environment to dbus." (startkde) - https://bugzilla.redhat.com/1191171
15:33:05 <rdieter> https://bugzilla.redhat.com/show_bug.cgi?id=1191171#c7  in particular
15:34:08 <rdieter> anyway, anything else plasma5/feature related to discuss today?  otherwise, will move on soon
15:35:12 <than> does it work before (build with gcc <5) ?
15:35:32 <rdieter> than: reportedly yes
15:35:51 <rdieter> probems started as soon as plasma-workspace was built with gcc5
15:36:09 <Kevin_Kofler> The thing is, startkde is shell code, not C/C++ code.
15:36:31 <Kevin_Kofler> Either qdbus-qt5 is broken, or some tool startkde is starting to decide what to pass to qdbus-qt5.
15:36:35 <rdieter> it calls /usr/libexec/ksyncdbusenv
15:36:43 <rdieter> and that call returns an error
15:36:56 <than> it looks like new gcc miscompiled the kde code again
15:37:01 <Kevin_Kofler> Ah, it's not a command-line qdbus call? Then it could be a GCC issue alright.
15:37:37 <rdieter> ltinkl even offerred a workaround patch to change the code a bit, but it didn't help
15:38:03 <Kevin_Kofler> Ah right, I vaguely remember that.
15:38:22 <rdieter> http://pkgs.fedoraproject.org/cgit/plasma-workspace.git/tree/plasma-workspace-ksyncdbusenv.patch
15:38:33 <rdieter> but apparently the error is the same
15:38:59 <Kevin_Kofler> I see the patch now. I guess then the error is not the QStringList construction, but the QDBus code.
15:39:06 <rdieter> that said, one person reporting a similar bug said the recent rebuilds fixed things
15:39:12 * rdieter looks
15:39:50 <rdieter> .bug 1190817
15:39:52 <zodbot> rdieter: Bug 1190817 krunner - Segmentation fault - https://bugzilla.redhat.com/1190817
15:40:29 <than> maybe we could try to build with -O0 and see if it works, of course it's not a fix but more a workaround temporary before we know how to fix the issue
15:40:55 <rdieter> maybe, I've no better ideas
15:41:18 <than> i will try next days
15:41:33 <Kevin_Kofler> At least it'd tell us WHAT is being miscompiled.
15:41:42 <than> Kevin_Kofler: yes
15:41:47 <Kevin_Kofler> Of course, it's no guarantee that -O0 will help either.
15:42:22 <than> Kevin_Kofler: but it's worth trying to build with -O0
15:42:49 <than> we need to know why the code is miscompiled with gcc-5
15:43:24 <than> or maybe  it's the fault in kde code
15:44:45 <rdieter> not sure if it's worth reverting that patch first or not, I'll leave that to you
15:44:53 <rdieter> let's move on to scheduling then
15:44:55 <Kevin_Kofler> So it's indeed all done in C++ code, the shell script (startkde.cmake, which I'm looking at now) only starts @CMAKE_INSTALL_FULL_LIBEXECDIR@/ksyncdbusenv with no arguments.
15:45:08 <Kevin_Kofler> That said, there are other things in startkde that I think we should fix.
15:45:17 <rdieter> agreed
15:45:38 <than> yes
15:45:40 <Kevin_Kofler> 1. It prepends the Qt bindir to PATH. We removed that from the KDE 4 startkde, it needs to go away also from the Plasma 5 one.
15:45:42 <Kevin_Kofler> It breaks Qt 3.
15:45:57 <Kevin_Kofler> And also qtchooser setups, depending how the user set them up.
15:46:03 <rdieter> Kevin_Kofler: please file a bug (or just patch it)
15:46:28 <Kevin_Kofler> 2. The default font family is Oxygen. Do we really want that? IMHO, we should default to plain "Sans" / "Monospace", which is aliased to DejaVu systemwide.
15:46:30 <rdieter> long term, we should probably just ship our own startkde
15:46:38 <Kevin_Kofler> +1, as we've done in KDE 4.
15:46:43 <Kevin_Kofler> That's exactly what I want to do, in fact.
15:47:09 <Kevin_Kofler> (Fixing 2. also means we could get rid of the Requires: oxygen-fonts.)
15:47:19 <rdieter> I wouldn't mind losing oxygen fonts, they look bad to me (on my screens at least), but I don't feel strongly about that
15:47:36 <Kevin_Kofler> They also have nowhere near the glyph coverage of DejaVu.
15:47:51 <than> as i remember Oxygen font has rendreing issue,
15:48:08 <than> i'm really against it as default
15:48:09 <Kevin_Kofler> So, agreed to change the default back to the systemwide default?
15:48:17 <than> Kevin_Kofler: +1
15:48:31 <rdieter> Kevin_Kofler: I don't think we have enough present to decide that now, I'd suggest proposing onlist
15:48:32 <Kevin_Kofler> (The real systemwide default (DejaVu), of course, not the Cantarell font GNOME Shell uses as the default. :-) )
15:48:52 <rdieter> but consider me a tentative +1 too
15:49:03 <than> no Cantarell font please :)
15:49:13 <Kevin_Kofler> I think we definitely all agree on that. :-)
15:49:50 <jgrulich> I'm used to Sans Serif, so +1 for not making oxygen-fonts as default or requirement
15:49:59 <Kevin_Kofler> (The funny thing is that I had to fix a hardcoded reference to Cantarell in Calamares of all things. ^^ Upstream agreed with me that hardcoding a GNOME font is not a good idea. :-) )
15:50:54 <rdieter> I suppose a compromise of sorts would be to at least continue to install oxygen-fonts by default, so users will always have them available
15:52:34 * than is fixing build issue in qt5-qtwebkit
15:52:53 <tosky> rdieter: that's not a bad idea
15:53:00 <Kevin_Kofler> rdieter: I'm opposed to installing anything that's not required by default.
15:53:04 <Kevin_Kofler> Download sizes matter.
15:53:22 <rdieter> than: I suspect it may suffer the same problem qt4's bundled webkit did
15:53:34 <rdieter> (it tried to detect gcc version, but only checked up to v4)
15:54:33 <rdieter> Kevin_Kofler: it's relatively small in this case
15:54:49 <rdieter> anyway, let's get to scheduling before we run out of time
15:55:01 <rdieter> #topic f22 schedule vs kde-apps/plasma schedule
15:55:14 <rdieter> https://fedoraproject.org/wiki/Releases/22/Schedule  references both
15:55:19 <Kevin_Kofler> #link https://fedoraproject.org/wiki/Releases/22/Schedule
15:55:20 <than> rdieter: it's another issue in qt5-qtwebkit. it loooks like a conflich with new glib in rawhide
15:55:40 <than> s/conflich/conflict
15:55:47 <rdieter> ok
15:55:48 <rdieter> :(
15:56:22 <rdieter> as far as scheduling goes, I think both plasma-5.3 and kde-apps-15.04 are too late to include in f22 GA
15:56:43 <rdieter> that is, unless either includes killer features we *must* have
16:03:08 <rdieter> now, if any slips happen we can certainly reconsider
16:03:20 <Kevin_Kofler> (The GCC already team tried to help with that, it seems. ;-) )
16:03:36 <Kevin_Kofler> Uh, shuffled sentence there…
16:03:41 <rdieter> woo, got confirmation in startkde bug that the latest qt5 build fixes it
16:03:41 <Kevin_Kofler> (The GCC team already tried to help with that, it seems. ;-) )
16:03:49 <rdieter> startkde/dbus bug, that is
16:04:01 <Kevin_Kofler> Good, looks like it was a GCC bug, too.
16:04:19 <rdieter> either gcc or gold(linker)
16:04:37 <rdieter> er, wait, the linker thing was just qt4
16:04:39 <Kevin_Kofler> Yeah, Qt using gold by default is silly, there's a reason it's not the default.
16:04:56 <Kevin_Kofler> Oh well, GCC 5 then.
16:05:17 <rdieter> ok, I'm confusing myself again
16:05:25 <rdieter> yes, it was qt5 that used gold
16:05:42 <rdieter> I forgot that qt5-qtbase actually uses a configure script (vs other modules that use qmake)
16:05:51 <Kevin_Kofler> IMHO, we should just permanently stick to --no-use-gold-linker and use it everywhere.
16:06:00 <rdieter> maybe :-/
16:06:02 <than> great that  startkde/dbus bug is fixed!
16:06:05 <Kevin_Kofler> I don't see a valid reason why Qt needs to use a different linker than Fedora default.
16:06:24 <Kevin_Kofler> It saves a few seconds of build time, I guess, but who cares?
16:06:56 <rdieter> alright, hour is up, any last comments before closing meeting?
16:09:58 <rdieter> I'll take that as 'no'
16:10:01 <rdieter> thanks everyone!
16:10:02 <rdieter> #endmeeting