15:09:53 <rdieter> #startmeeting kde-sig
15:09:53 <zodbot> Meeting started Tue Nov 12 15:09:53 2013 UTC.  The chair is rdieter. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:09:53 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:09:57 <rdieter> #meetingname kde-sig
15:09:57 <zodbot> The meeting name has been set to 'kde-sig'
15:10:04 <rdieter> hola everyone
15:10:08 <rdieter> #topic roll call
15:10:11 <jgrulich> hi
15:10:24 <Kevin_Kofler> Present.
15:11:49 <rdieter> than, dvratil, mbriza, kde*foo : ping
15:11:59 <dvratil> hi
15:14:22 <rdieter> #info jgrulich Kevin_Kofler dvratil rdieter present
15:14:27 <rdieter> #chair jgrulich Kevin_Kofler dvratil
15:14:27 <zodbot> Current chairs: Kevin_Kofler dvratil jgrulich rdieter
15:14:33 <rdieter> #topic agenda
15:14:48 <rdieter> what to discuss today?  I can do kde-4.11.3 status
15:15:09 <Kevin_Kofler> 4.12 Beta?
15:16:36 <rdieter> alright, sounds like a good start.
15:16:40 <rdieter> #topic kde-4.11.3 status
15:17:01 <rdieter> pinged folks earlier in #fedora-kde and went ahead and queue'd what we had in -testing for stable updates
15:17:36 <Kevin_Kofler> Good.
15:17:49 <rdieter> on related topic, phonon-4.7 stuff had a few wrinkles ironed out, submitted fresh builds for -testing earlier today
15:19:25 <rdieter> that's all I have on that
15:19:46 <Kevin_Kofler> And KDevelop 4.5.2 is now queued for stable.
15:21:07 <rdieter> alrighty moving on ..
15:21:13 <rdieter> #topic kde 4.12 beta
15:21:16 <rdieter> Kevin_Kofler: ?
15:22:00 <Kevin_Kofler> So, that thing has been released for a while, shouldn't we try to get it into Rawhide?
15:22:48 <Kevin_Kofler> It's going to be hard to get 4.12 ready if we don't take advantage of the betas to do the usual packaging changes, review requests if any, etc.
15:22:50 <rdieter> it's probably more a matter of finding someone (or someones) with time to do it
15:24:36 <Kevin_Kofler> Yeah. I used to do some of those updates in the past, but I'm not really in touch with the current processes now that we need to work with scripts etc.
15:24:51 <rdieter> personally, i may have time tomorrow (slight), or this weekend
15:24:51 <Kevin_Kofler> (No, I'm not going to bump the hundreds of packages by hand.)
15:25:34 <Kevin_Kofler> I can help if you need someone to look at individual packages (fixing builds, rebasing patches etc.).
15:25:40 <rdieter> i have a helper build-it.sh script, where you can just do:  ./build-it.sh <pkg>
15:26:03 <rdieter> that does the basic updating, sans patch management/rebasing
15:26:15 <Kevin_Kofler> That helps when everything goes fine, yeah… :-)
15:26:30 <rdieter> ok, let's see how tomorrow goes (for me), else weekend it is
15:26:36 <Kevin_Kofler> It's still a pain when there is anything that needs fixing, but I doubt we can make that any easier.
15:26:57 <rdieter> <nod>
15:27:14 <rdieter> only minor gotcha is to find/fix any assumptions where kde version = kde-workspace version
15:27:23 <rdieter> since that will on longer be true
15:27:36 <Kevin_Kofler> Yeah, upstream is really doing everything they can to make packaging a PITA.
15:27:53 <Kevin_Kofler> First the split tarballs and now that.
15:29:21 <rdieter> ok, well, anything on 4.12 ?
15:29:40 <rdieter> anything else, that is
15:29:53 * jreznik is listening
15:29:57 <Kevin_Kofler> I think pretty much everything that links to anything from kde-workspace (and maybe even some stuff that doesn't ;-) ) has a versioned BR kde-workspace >= %{version} and possibly also a versioned runtime Requires.
15:30:26 <Kevin_Kofler> These all need to be fixed (I guess BR kde-workspace >= 4.11 and Requires >= the version detected at runtime).
15:31:01 <Kevin_Kofler> (detected at build time, I mean)
15:31:10 <rdieter> yeah, for the latter I was thinking of putting in a _kde4_workspace_version macro somewhere
15:31:48 <Kevin_Kofler> In the Frameworks world, we'll need to cope with EVERYTHING having a separate version, like in GNOME. THAT will be a PITA.
15:32:08 <Kevin_Kofler> Some developers have admitted the workspace thing is a way to force us to get used to that nonsense.
15:32:32 <Kevin_Kofler> So we'll need _*_version macros for every package?
15:32:34 <rdieter> heh :(
15:32:54 <Kevin_Kofler> Or we just do what most packages do and ignore the runtime >= constraint, just desupporting selective upgrades.
15:33:07 <rdieter> if that is so, I guess we could whack them will clue sticks as retribution
15:33:23 <rdieter> s/will/with/
15:33:36 <rdieter> Kevin_Kofler: yeah
15:33:37 <Kevin_Kofler> We should try again to get them to use symbol versioning.
15:34:25 <Kevin_Kofler> Even if they don't actually use the versioning the intended way (as versioning with different version supported), just the fact to ASSIGN a version (which could even be automatically copied from the @since annotations) to every symbol would solve the problem for us.
15:34:38 <rdieter> <nod>
15:34:43 <Kevin_Kofler> (because RPM then adds dependencies on the library name + symbol version (ignoring the symbol name).)
15:34:49 <rdieter> even that would be an improvement over the status quo
15:35:11 <Kevin_Kofler> And using symbol versioning that way would not change anything for the targets that don't support it.
15:35:31 <Kevin_Kofler> We should try to sell that to them for the new KF5 world.
15:35:48 <rdieter> I've been tempted to implement something like that downstream, but big downside is that it means we are now abi-incompatible with everyone else
15:36:02 <Kevin_Kofler> I guess I'll try to write something up (nag me if I forget), but maybe somebody ELSE should send it, given how "popular" I am on mailing lists. ;-(
15:37:12 * rdieter has no idea what you're talking about, your rock star status is untouchable :)
15:38:06 <rdieter> I think a good idea would be recepted, regardless of who it comes from
15:38:40 <Kevin_Kofler> I think if you sign on that proposal, that'll help though. :-)
15:38:46 <Kevin_Kofler> But first we need to write it up.
15:39:16 <rdieter> ok, definitely count me in
15:39:34 <Kevin_Kofler> We'd sign our names then, and I hope we'll get support from other RPM-based distros, and maybe deb ones too if that also helps dpkg.
15:39:50 <rdieter> we could spin it like... if you're screwing us with inconsistent versioning, at least do this to minimize the damage... kind of thing
15:41:04 <Kevin_Kofler> Yes, but rather worded like "Because the Frameworks will by their nature have separate version numbers, this becomes important. In the past, we used the following approach: … but now that will not scale anymore."
15:41:07 <Kevin_Kofler> :-)
15:41:08 <rdieter> debian ... not sure, they already track symbol versions themselves in many places, but I doubt this will hurt things from their perspective
15:41:44 <rdieter> Kevin_Kofler: your wording sounds much better
15:42:01 <Kevin_Kofler> I can sound diplomatic when I'm not too angry. ;-)
15:42:12 <rdieter> heh
15:42:40 <rdieter> when you're tempted to use words like crappy and braindead, you know the line has probably been crossed. :)
15:42:51 <Kevin_Kofler> ^^
15:43:16 <Kevin_Kofler> So I should rather use words like sh…y and f…king? ;-)
15:43:50 <rdieter> har, anyway, anything else on the 4.12 topic?
15:44:04 <Kevin_Kofler> Speaking of betas, there's also a KDevelop 4.6 beta.
15:45:07 <rdieter> #topic open discussion
15:45:27 <rdieter> and... I just queue'd a fresh batch of qt5 5.2.0 beta builds for f20-testing
15:45:38 <Kevin_Kofler> So one more thing: Do we really want KMix to give noisy volume feedback by default?
15:45:47 <rdieter> these, in particular (re)set qreal on arm to be double (intead of float)
15:46:10 <Kevin_Kofler> I'm a bit torn, I see why setting the volume is the one place making noise is justified, but I'm also not sure our users will like the noise.
15:46:18 <rdieter> Kevin_Kofler: i'm biased toward yes
15:46:39 <rdieter> esp after a bit of feedback on test day
15:46:41 <Kevin_Kofler> Re qreal on ARM, yay, finally!
15:47:30 <Kevin_Kofler> That will allow us to stop worrying about ARM-only issues in the Qt 5 future.
15:48:12 <rdieter> fortunately, not much stuff is built against qt5 yet, so the abi break is relatively easy to deal with
15:48:23 <Kevin_Kofler> Having a typedef that is always double except on one platform where it is float was a constant source of trouble.
15:49:49 <rdieter> it was a bit interesting fixing some of the code that assumed qreal = double
15:50:05 <rdieter> for various definitions of "interesting"
15:50:28 <rdieter> too bad we can't fix qt4 the same way. :-/
15:50:52 <rdieter> (though most of the issues there have been fixed already)
15:52:34 <Kevin_Kofler> Well, I guess we could, if we ignore ABI compatibility with other distros and the LSB, and do a mass rebuild after that. :-)
15:53:00 <Kevin_Kofler> But I don't think it's a good idea.
15:54:26 <Kevin_Kofler> Qt 5 is the right time to do it.
15:54:31 * rdieter will close meeting in 2 min, if there's nothing else to discuss
15:56:38 <rdieter> thanks everyone!
15:56:41 <rdieter> #endmeeting