[00:01] nall: So, the chances are that this is a problem somewhere in the EGL stack. It might be time for two-machine debugging - start up mir_demo_server under gdb, and break on... [00:01] Something or other. What would that be... [00:05] Probably mf::SessionMediator::create_surface [00:11] Yeah there are I think 6 definitions for create_surface in various classes. That's the first one called so I'll start there. I've never used gdb before, didn't even know it existed. I've just been grepping the source and putting puts() in various places. Thanks for the tip. [00:12] Yeah. [00:12] Single-stepping is pretty awesome. [00:15] might be interesting to test render_to_fb and render_surfaces first [00:16] s/interesting/easier/ [00:28] I'll try that too. Thanks [00:30] racarr, still here? [05:35] RAOF, ping [05:35] tvoss: Pong. [05:35] Up early? [05:35] RAOF, yup :) how is it going? [05:35] Well. === duflu_ is now known as duflu [07:06] alf_, ping [07:06] tvoss: pong [07:10] Saviq: tvoss: At what point in the lifecycle of a session does the shell need to take a snapshot of a session/surface? Just before it switches away from it? [07:10] alf_, or to it [07:11] Saviq: why when to it? [07:11] alf_, two cases: session is born, the app slides from the side [07:11] alf_, or we switch between apps [07:11] alf_, in both cases there are transitions for both the currently and the next focused app [07:12] alf_, at the moment basically whenever an app appears not in its final position (fully on screen, bar the panel) [07:13] alf_, is when a snapshot is used [10:04] I'm in the process of updating to new images and find that I can't run mir on my updated N4 [10:04] # ldd /bin/mir_demo_standalone_render_surfaces | grep not\ f [10:04] libhardware.so.1 => not found [10:04] Am I doing something stupid? === mmrazik is now known as mmrazik|lunch [10:30] alan_g: It's all part of the various branches/bugs I'm working on [10:30] And shall have to stop soon [10:32] duflu_: ack [10:34] I give up... === mmrazik|lunch is now known as mmrazik === mzanetti is now known as mzanetti|lunch === alan_g is now known as alan_g|lunch === mzanetti|lunch is now known as mzanetti === alan_g|lunch is now known as alan_g [13:56] Hi folks, anyone using the mir staging ppa on saucy at the moment? [13:56] * alan_g thinks that is an ominous question [13:57] I added the PPA yesterday, but on reboot lightdm failed to start. Seemed Mir did start to launch ok, as I got the cursor, but then that died. [13:57] :) [14:00] I updated a machine to saucy yesterday - but not tried using Mr as system compositor [14:16] alan_g: that's a no then? :) [14:16] no worries, just a head's up [14:17] well, I used the ppa - but only for building mir [14:19] * alan_g isn't likely to use system compositor until mir can be nested === mmrazik is now known as mmrazik|afk === alan_g is now known as alan_g|tea [15:06] good morning === alan_g|tea is now known as alan_g [15:06] kdub: good afternoon [15:09] status, mp'd a cleanup for mesa/gbm, probably will tackle the egl symbols bug 1189938 in mir this morning [15:09] bug 1189938 in Mir "android links to EGL extension functions directly" [High,In progress] https://launchpad.net/bugs/1189938 [15:17] Morning [15:22] afternoon [15:34] Just spilled my morning coffee halfway through....my "habit brain" is revolting [15:35] :) [16:26] racarr, rofl :) [16:30] tvoss: What happened with that app-centric world meeting [16:30] yesterday when I woke up I saw that it had been rescheduled [16:30] in to the past? [16:32] and everyone but you attended ! [16:33] oh did it actually happen yesterday? [16:33] :) [16:33] I was at the dentist anyway [16:33] nah, kidding [16:33] racarr, it happend, it is a weekly one now [16:33] oh ok [16:33] ill come next week [16:34] racarr, yup [16:34] kdub: Is there a fast way to get the pixels of an android buffer? glReadPixels() from an FBO backed by our buffer as a texture takes >50ms on Nexus4 :/ [16:34] kdub: (for a 512x512 surface) [16:34] alf_, yeah, glReadPixels on adrenos is reallly slow [16:34] you can access the buffer's pixels via gralloc [16:35] we have some tests that do it right? [16:35] example code at tests/draw/android_graphics.cpp [16:37] adreno is tile-based gpu, so glReadPixels is tricky to do quickly [16:38] kdub: thanks, I think that we will need to have a Buffer::pixels() method, and use that to get the pixels, fall back to glReadPixels()/FBO if ::pixels() fails (e.g. on GBM we can't really get the pixels of a hardware buffer, they may be tiled). [16:39] alf_: If we can peek at the buffers, I don't understand why we need software [16:39] well :) another alternative is to have anothe class that we can plug the native handles into, and it can provide us with the pixels [16:40] copy [16:40] kdub: sure [16:46] racarr: That's the ubuntu API for now, but there are also some synchronization issues with making buffers available to the shell outside of the compositor. [16:49] racarr: One idea discussed is for mir to give out a buffer to the shell when an app closes [16:52] oh when it closes, of course, sorry [16:52] alf_: ^ === alan_g is now known as alan_g|EOW [17:02] Bye all! [17:02] Bye! [17:03] alan_g|EOW, bye [17:12] * racarr enters the "Q" === seb128_ is now known as seb128 [17:27] this half a cup of coffee thing just isnt working out for me [17:27] brb [19:36] hey racarr, updated the doc in https://code.launchpad.net/~mir-team/mir/saucy-android/+merge/169365 if you could take another look? [19:37] i have a egl-extensions branch that can land right after that one [19:47] hmm, actually... its probably just easier to land all the fixes at onec [20:14] kdub: Should we merge it/ [20:14] yeah, i think so [20:16] i'll mp the extensions branch within an hour [20:25] Ok [20:30] kdub: " [20:30] apparently if you bzr commit, change, save commit message, it needs a second commit [20:30] " :) [20:30] haha the more you know [20:30] yeah :P [20:40] hello everyone, i've uploaded my current mir support for sdl1.2 on the mir ppa if anyone wants to try it out :) [20:42] (games are work :) [20:42] working* [21:09] bschaefer: Excellent [21:09] ill try it out the next time im looking for something to play with [21:10] racarr, thanks :), and cool! Still need to fix some odd thing with SW rendering...but I've a work around for now [21:10] (copying the entire screen over to mir...) [21:10] hmm [21:10] ill try and look at that [21:11] racarr, im making a different ppa to show you what I mean, its strange, I copy the changes for that frame over to mir, and swap buffers [21:11] one day we should port it to the platform-api which should mostly be find and replaceish stuff [21:11] it doesn't really matter now, but might as more application description stuff, etc comes in [21:11] yeah, that would be nice [21:12] racarr, its as almost mir doesn't paint the alpha? Or assumes its double buffered...im not very sure :)