=== chihchun_afk is now known as chihchun === chihchun_afk is now known as chihchun === chihchun_afk is now known as chihchun === mariogrip_ is now known as mariogrip === _iwc_ is now known as _iwc === Stskeepz is now known as Stskeeps === davmor2_ is now known as davmor2 === popey_ is now known as popey [11:14] anpok: is this likely to be another gtk-mir problem? https://bugs.launchpad.net/miral/+bug/1627697 [11:14] Ubuntu bug 1627697 in MirAL "gnome-terminal crashes on resize" [Undecided,New] [11:58] kdub: Is our IRC server down for you too, or is it just me? [11:59] hmm, launchpad is down for me, so i'd guess outage [12:00] #launchpad says "Launchpad temporarily unavailable due to a network failure" [12:00] * alan_g decides to try again after lunch [12:02] Firewall madness strikes again. === w00t_ is now known as w00t === chihchun is now known as chihchun_afk === mhall119_ is now known as mhall119 [13:14] http://pastebin.com/DtHEr0U8 [13:14] any idea what might be the reason for unity8 to fail? unity-system-compositor works though [13:15] the device is MTK Helio X10 [13:21] NeKit: kdub might have an diea.. [13:22] alan_g: I will take a look when the keymapping bug is in a better state [13:23] NeKit, it seems to fail to map the buffer on the server side [13:23] and then seems to have fencing problems [13:24] maybe the usage bits go wrong somewhere [13:24] not sure though [13:25] visually Ubuntu logo animation is playing fine though [13:25] but nested compositor won't start due to buffer error [13:26] anpok: thanks, I'm starting to feel that some spelunking tin gtk+ is needed. [13:28] Hi! How can I query screen resolution from Mir ? [13:30] om26er: as soon as you create a surface that is shown on screen you get surface_output_event [13:30] anpok, is there a commandline way to get that ? [13:30] or something pythonic ? [13:30] mirout [13:31] that even will tell you on which output you are displayed.. the via mir_connection_display_config you get access to the current setup... [13:31] *event [13:31] yes mirout [13:32] kdub, might it be related to https://code-review.phablet.ubuntu.com/gitweb?p=aosp/platform/frameworks/native.git;a=commitdiff;h=90c5a47b7c7f00c04b4b681f4b81aea90b606b6f? [13:33] broken link [13:35] kdub, https://code-review.phablet.ubuntu.com/gitweb?p=aosp/platform/frameworks/native.git;a=commitdiff;h=90c5a47b7c7f00c04b4b681f4b81aea90b606b6f [13:36] well, we don't really use the BufferQueue system in the ubuntu touch stack [13:53] any steps that might help to debug buffer mapping then? [14:05] eh, I suppose I would start tinkering with mir::graphics::android::Buffer::write or read, and see if you can figure out a reason why its failing [14:05] maybe tinker with the allocation bits passed in to gralloc [14:06] thanks === charles_ is now known as charles [14:23] alf_: hey, i'm back...but only here :) [14:24] but i was going to say, i added the core/kern snap to the drawing in order to [14:24] kgunn: (me too, only here...) [14:24] show that those my have facilities needed by the u8-session snap [14:24] to actually control the screen [14:25] which i think is exactly what you were indicating [14:25] was missing from the display control i/f that is there today [14:27] kgunn: that's one aspect, the other is actually allowing repowerd (or others) provide the dbus service over which apps can request keep-display-on [14:27] alf_: yes, exactly [14:28] alf_: so what may not be obvious, "an interface" (as in a single interface) [14:28] can actually provide the slot/plug to the system/server side & the plug/slot to the client side [14:28] alf_: mir interface is actually an example of this [14:29] alf_: in the mir interface anything under "permanent slot" is what the mir server side is needing from the system [14:29] kgunn: looking... [14:29] anything on the "connected slot" is what the mir server may need to do on the client side, and "connected plug" is what the client needs from the server [14:30] alf_: ^ [14:30] hope that makes sense [14:31] kgunn: so, an interface can contain multiple slot/plug pairs? [14:32] alf_: yes...well, specifically 2 [14:33] alf_: in my simple mind, i think of it as "top" pair and "bottom" pair architecturally [14:42] kgunn: thank, your explanation + reading the mir snap helps. [14:47] kgunn: my approach would be for the screen-inhibit-control interface to have a ConnectedSlot (allowing accepting dbus request) and ConnectedPlug (allowing sending dbus requests) [14:47] kgunn: but no PermanentSlot/Plug [14:48] alf_: well, i think you've gotta drive that thru some testing.... [14:48] e.g. can u8 session do what it needs to do ? [14:49] alf_: when AlbertA and tedg ge the u8-session to the point where you can launch an app we'd need to try on an all-snaps system [14:51] kgunn: yes, I need to test, but my point (and I could be wrong) is that screen-inhibit-control interface shouldn't know about how repowerd in particular implements this functionality (hence no PermanentSlot/Plug) [14:52] alf_: well...it's not that the plug/slot knows about implementation per se...it's just permissions === JanC_ is now known as JanC [14:53] kgunn: it leaks implementation details, e.g., PermanentSlot contains the paths that repowerd needs to access to provide the display-on functionality. [14:54] kgunn: mir can get away with this because the interface is 'mir' not 'compositor' [14:59] hmm, login.ubuntu.com isn't working for me, which means no hangouts === chihchun_afk is now known as chihchun === chihchun_afk is now known as chihchun [16:17] anpok: when you get headspace to look at gtk-mir, could you check all the miral bugs tagged with gtk-mir - AFAICT they are all in the toolkit [16:20] alan_g, do you want me to tag vs put [ gtk ] in the name for bugs? [16:20] * bschaefer wasnt really sure how you wanted to do the bug reports [16:21] bschaefer: tags are easy to search [16:21] right, ill make sure to add those for future bugs [16:22] But really once we're sure they're toolkit related they should be logged upstream [16:22] i agree [16:22] I've not got around to that either [16:22] (not sure where to log gtk mir issue?) [16:22] is there a launchpad page for it? [16:23] same with sdl2/sdl1.2 i suppose you *could* log it upstream for SDL2 (which is fine since mir lives in upstream SDL2) [16:25] 22 MP's up for review :/ [16:26] kdub, partly my fault :| [16:26] * bschaefer made a bunch of sub MPs [16:26] but now depend on the ABI changes (which is blocked atm) === chihchun_afk is now known as chihchun [16:27] partly everyone's fault :) [16:27] :) [16:27] * bschaefer starts reviewing [16:31] alan_g, how do I tell miral that no window should have focus? ie, I opened a indicator panel, so the foreground window should lose focus and it should not be given to any other window [16:32] when indicator panel closes, focus return to foreground window [16:32] select_active_window(Window{}) should do it [16:32] alan_g, if I understood its code, it will give focus to the previous window [16:33] Hmm. Let me look... [16:34] select_active_window(Window{}) should do it. [16:35] alan_g, right, that prev_window variable confused me.... [16:35] It needs to notify focus loss to the previous window [16:38] bschaefer: anpok did point me where to log gtk bugs, butr I can't find it in my IRC log today [16:39] alan_g, :) i can poke anpok later [16:39] as if i can figure out if its miral (ill still report it for miral) then report it for gtk as well [16:39] just so i can mark it as a dup of that (so others can follow it) [16:39] That sounds good === dandrader is now known as dandrader|afk [16:58] ... i am out for a bit .. but will continue for a few hours later tonight [16:59] bschaefer: bugzilla.gnome.org product:gtk+ component: Backend:Mir [16:59] anpok, sweet thanks [16:59] alan_g, ^ [17:00] 8) [17:01] * alan_g has lp:1625401 in gdb, with debug symbols [17:02] o nice [17:02] its not just X? [17:02] * bschaefer only saw it on miral on x [17:03] but doesnt mean it didnt happen on kms :) [17:03] o strange on mir 0.24... ive not seen this on mir_demo_server (wonder if its something strange happening specifically in miral?) [17:04] though ive not tried xmir on mir_demo_server [17:04] I hope to find out [17:08] thats a good hope [17:34] bschaefer: found it [17:35] alan_g, o nice, what was it? [17:35] https://bugs.launchpad.net/mir/+bug/1625401/comments/3 [17:35] Ubuntu bug 1625401 in MirAL "Miral seems to spin at 100% and steals focus from everything until the server shutdown" [Critical,Confirmed] [17:35] o interesting [17:35] alan_g, just it just loop forever trying to find a not null session? [17:36] Trying to find a session that's either null or doesn't have a null default_surface [17:36] interesting [17:38] I wish I could reproduce more than twice a day. But I've already stretched EOD, so I'll finish off tomorrow. [17:38] yeah :( i was randomly hitting [17:39] alan_g, o also the reason we couldnt remove that API [17:39] is it would break ABI? (from my understanding?) [17:39] (the public client API) since it was released in 0.24 [17:40] No worse than changing the function to not work [17:40] but we cant break public client ABI? [17:40] Both cases break clients that use it [17:40] (the C abi) [17:41] the function [17:41] any other takers? https://code.launchpad.net/~kdub/mir/eglimage-from-mirbuffer-android/+merge/305868 https://code.launchpad.net/~kdub/mir/passthrough-ipc-plumbing/+merge/305698 https://code.launchpad.net/~kdub/mir/fix-1626503/+merge/306480 ? [17:41] kdub, ill attempt to review but a lot of the context is lost :) [17:41] alan_g, hmm right but ... i would love to remove it but if we are going to break client ABI ... shouldnt we just go ahead and break it for a bunch of things we want to remove? [17:41] bschaefer, thanks (and thanks to the other vogons who have reviewed those ones too) [17:42] alan_g, ill ask you tomorrow, dont want to question to much here :) [17:42] (since you're EOD) [17:43] bschaefer: I've pushed an (untested) fix, if you have time to try it... [17:43] alan_g, yeah i can try to reproduce it all day [17:43] and let you know [17:45] bschaefer: breaking the function affects exactly the same clients that removing it does [17:45] * alan_g has to go === alan_g is now known as alan_g|EOD [17:46] alan_g|EOD, ...very true === dandrader|afk is now known as dandrader === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader