[Do Feb 7 2008] [17:41:09] hi onestone :) [Do Feb 7 2008] [17:41:32] this is turning out to be compiz fusion dev meeting [Do Feb 7 2008] [17:47:03] Betreten david_r hat den Kanal betreten (n=david@nat/novell/x-549b02f1883e73eb). [Do Feb 7 2008] [17:47:29] cyberorg, maniac103 , onestone : ask and ye shall receive [Do Feb 7 2008] [17:47:43] don't keep him too long though, he's a busy man :-) [Do Feb 7 2008] [17:48:10] cyberorg, maniac103 , onestone: hey guys [Do Feb 7 2008] [17:48:20] hi david_r :) [Do Feb 7 2008] [17:48:24] david_r: hi! :-) [Do Feb 7 2008] [17:48:26] hi david_r [Do Feb 7 2008] [17:48:49] who goes first? [Do Feb 7 2008] [17:49:07] onestone: did you get the 'make distcheck' issue sorted out ok? [Do Feb 7 2008] [17:49:20] david_r: Yes [Do Feb 7 2008] [17:49:24] good [Do Feb 7 2008] [17:49:27] david_r: just a quick question from my side - did you receive my session management patch review request? [Do Feb 7 2008] [17:49:44] adding -I$(top_builddir)/include fixed it [Do Feb 7 2008] [17:50:03] maniac103: no, I'm pretty sure I didn't.. [Do Feb 7 2008] [17:50:19] ok, then I can wait pretty long for a response ;-) [Do Feb 7 2008] [17:50:24] I sent it on 01/30 [Do Feb 7 2008] [17:50:45] onestone: good that's what we want [Do Feb 7 2008] [17:50:52] maniac103: let me check again.. [Do Feb 7 2008] [17:51:30] david_r: if you can't find it: http://www.pastebin.ca/895164 - the patch is pretty simple, the basic idea is to make all session management callbacks wrappable via the core object [Do Feb 7 2008] [17:51:41] so a plugin can implement session save/restore [Do Feb 7 2008] [17:54:38] david_r: I played a lot with reparenting in compiz, and I think that we can do the "reparent the resized window into a bigger one" only if we have support for reparenting in core. And I think that I could create a system that would be able to do our current decorations, the faster window resize and "old style" reparented decorations, only with small changes to the plugin API [Do Feb 7 2008] [17:54:40] maniac103: yes, session management should definitely be made pluggable [Do Feb 7 2008] [17:54:56] maniac103: object-framework branch allows it (and many more things) to be made pluggable without any changes to the core [Do Feb 7 2008] [17:55:27] maniac103: your current patch can go into master for know if you want that, though [Do Feb 7 2008] [17:55:52] david_r: the problem is that you are the only that really understands the object framework [Do Feb 7 2008] [17:56:06] david_r: ok, great ... and yeah, what onestone said ;-) [Do Feb 7 2008] [17:57:30] onestone, maniac103: yes, so don't worry about the object-framework branch for now.. [Do Feb 7 2008] [17:57:59] I'll make sure you understands it all once it's ready.. [Do Feb 7 2008] [17:58:00] david_r, so we submit 0.7 packages for factory? [Do Feb 7 2008] [17:58:11] david_r: except we need it very soon for opensuse 11.0 :-) [Do Feb 7 2008] [17:58:12] cyberorg: sure [Do Feb 7 2008] [17:58:32] david_r: we already thought that we could even release a stable 0.8, because of the delays with the object framewoek [Do Feb 7 2008] [17:59:25] david_r: I fear that object-framework won't be ready for the next round of distro releases, so I think it makes sense to release a 0.7.x for those (with my patch) and give the distros an o-f version in 6 months [Do Feb 7 2008] [17:59:30] jpr: im not sure I'll make that anymore but Im thinking about using all of hackweek (next week) to try and get it done [Do Feb 7 2008] [17:59:35] jpr: so maybe.. [Do Feb 7 2008] [18:00:10] david_r: how much more time do you need? [Do Feb 7 2008] [18:01:27] additionally, we might need to consider that we require a certain amount of time to convert the compiz-fusion stuff [Do Feb 7 2008] [18:01:45] Betreten darix- hat den Kanal betreten (i=darix@alpha049.server4you.de). [Do Feb 7 2008] [18:02:16] hi darix :) [Do Feb 7 2008] [18:02:36] jpr: not a huge amount of work left but some important pieces. i might be able to get it done in one week. then onestone, maniac103 also needs to understand it all before we can merge it so that might take some time as well.. [Do Feb 7 2008] [18:03:21] * jpr wonders if this can take place at FOSDEM by any chance [Do Feb 7 2008] [18:03:36] or some other place in person [Do Feb 7 2008] [18:03:45] onestone, maniac103 : locations? [Do Feb 7 2008] [18:04:56] jpr: Germany Cologne/Aachen [Do Feb 7 2008] [18:05:12] maniac103: port existing plugins so they are properly designed and take advantage of everything that the of-branch provides will take a while but just making them run with minimal changes should be quick.. [Do Feb 7 2008] [18:05:36] jpr: Eastern Germany [Do Feb 7 2008] [18:05:45] Betreten afonit hat den Kanal betreten (n=afonit@72.243.23.170). [Do Feb 7 2008] [18:05:52] david_r: i don't suppose you are going to germany soon :-) [Do Feb 7 2008] [18:06:20] david_r: ok ... then we're left with compizconfig [Do Feb 7 2008] [18:06:45] jpr: no plans but I guess it can be arranged [Do Feb 7 2008] [18:08:26] hpj who will work on the config will be in berlin the week of March 10 [Do Feb 7 2008] [18:08:33] for the gtk hackfest [Do Feb 7 2008] [18:08:41] the config for opensuse that is [Do Feb 7 2008] [18:08:43] the question (from my POV) is if o-f brings us so many _immediate_ advantages so that we'd need to hurry up with implementation [Do Feb 7 2008] [18:14:24] david_r: I also got another patch to review which fixes outputDeviceForGeometry for completely overlapping output devices (see ML thread, problem arises with Xrandr 1.2), but I'll send it to you via mail as it's a bit more complicated than the session one [Do Feb 7 2008] [18:15:43] I'll be back in ~25 minutes [Do Feb 7 2008] [18:38:31] Betreten maniac103 hat den Kanal betreten (n=daba@pD95DF626.dip.t-dialin.net). [Do Feb 7 2008] [18:45:15] david_r: so what do you think about window reparenting support in core? [Do Feb 7 2008] [18:55:28] onestone: if would be great if it could be made as an addition, as a plugin. I'm not too happy about it if it's something that adds a lot of complexity to existing code and affects other plugins [Do Feb 7 2008] [18:55:44] onestone: the of-branch allows us to move everything except the object system itself into plugins [Do Feb 7 2008] [18:56:40] onestone: texture-from-pixmap support, GL support, X11 support.. can all be implemented as plugins [Do Feb 7 2008] [18:56:49] david_r: the problem is that doing it as plugin is more complex. And it would only need changes to core and the decoration plugin [Do Feb 7 2008] [18:57:11] david_r: the core is not able to handle reparented windows correctly [Do Feb 7 2008] [18:58:05] onestone: sure our X11 code should be made to better support reparented windows if that's a problem [Do Feb 7 2008] [18:59:09] onestone: you want the reparenting to support a non-composited desktop, right? [Do Feb 7 2008] [18:59:32] david_r: yes [Do Feb 7 2008] [19:00:22] david_r: but also to implement your idea of the faster window resize [Do Feb 7 2008] [19:01:07] onestone: I'd like to see non-redirected windows as an additional way of drawing windows [Do Feb 7 2008] [19:02:39] onestone: e.g. GL plugin allow us to draw windows using openGL. an additional plugin could use un-redirecting to render them... [Do Feb 7 2008] [19:03:28] onestone: so that's how I'd like to approach the support for a non-composited [Do Feb 7 2008] [19:04:20] david_r: my idea would be a xcomposite plugin that does the reparention and opengl/xrender plugins hook into it to have a composited desktop [Do Feb 7 2008] [19:05:25] david_r, just to clarify: you're ok with adding my session management patch until we support it through object-framework? it would give us a "safety net" if o-f doesn't make it in time until the next distro release round [Do Feb 7 2008] [19:05:43] (because session management is something a lot of people are asking for) [Do Feb 7 2008] [19:05:44] mw: i'm not clear about the spec copyright in those cases [Do Feb 7 2008] [19:05:46] onestone: pretty much the same thing. either way it seems like whatever X11 code we have it needs to properly support reparenting [Do Feb 7 2008] [19:06:27] onestone: so whatever you can do to improve the reparenting support is good [Do Feb 7 2008] [19:08:11] david_r: I think that I have a good plan to do it, I hope that I will get something working next week. But I'm sure we will need your X11 knowledge to handle all Xevents then correctly [Do Feb 7 2008] [19:08:16] maniac103: yep, the merge of the of-branch will just be about introducing it the minimal changes to existing code [Do Feb 7 2008] [19:08:59] david_r, ok, I will add it ... thanks [Do Feb 7 2008] [19:15:14] david_r, also, as I just read it on the ML: is there a better solution for edge actions than the invisible windows? it seems like more and more people keep complaining about them breaking Fitt's Law [Do Feb 7 2008] [19:16:39] david_r, I guess once we have a software cursor, we can use its coordinates, can't we? [Do Feb 7 2008] [19:18:02] maniac103: you would want them to be able to handle enter-leave events but let button presses through to window beneath, right? [Do Feb 7 2008] [19:18:46] maniac103: yes, with a software cursor and tracking of the cursor position we'll be able to handle this much better [Do Feb 7 2008] [19:19:11] david_r, indeed, those people complaining want clicks to go through [Do Feb 7 2008] [19:19:22] david_r: I think the only simple solution for now is to activate the top/bottom/left/right edges for edgeflip only if we really move a window [Do Feb 7 2008] [19:19:42] onestone, don't forget DnD ;-) [Do Feb 7 2008] [19:20:20] onestone, and stuff like dragging group tab bar slots [Do Feb 7 2008] [19:43:47] david_r, what also came to my mind is that IMO we should put the non-public function declarations into a separate header (e.g. src/compiz-internal.h) - stuff like the core instances of the wrapped functions or e.g. initSession() ... are you ok with me doing that after the o-f merge? [Do Feb 7 2008] [19:55:22] maniac103: the object system takes care of this naturally. all public functions are part of the object vtable (except a small amount of object system utility functions). it's going to be rare with private functions that need to be declared in headers. only plugins that use multiple source code files will need this. [Do Feb 7 2008] [19:56:40] maniac103: so moving declarations into internal headers will naturally be taken core of as we port code over to the object framework.. [Do Feb 7 2008] [19:56:52] david_r, ok then ... so stuff like the declaration of the base functions for the wrapped pointers will vanish automatically? [Do Feb 7 2008] [19:57:00] if yes, that's perfect :-) [Do Feb 7 2008] [19:58:00] maniac103: yes :) [Do Feb 7 2008] [20:22:05] david_r: Thx for being here, and I think that we have cleared up everything for now. [Do Feb 7 2008] [20:23:32] david_r, i am doing 0.7 packages on BS, will mail you the changes once i am done [Do Feb 7 2008] [22:42:21] Beenden david_r hat den Server verlassen ("Leaving").