=== _salem is now known as salem_ === salem_ is now known as _salem [11:28] boiko: it needs rerun (now is okay), there was a faulty UITK in overlay for a while that needed a no-change rebuild === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [12:47] Mirv: nice, thanks! :) === _salem is now known as salem_ === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [16:09] attente, how do I know if gdk backend probing is at work? and how can I force it when launching from commnad line? looks like it stopped working for me... I'm confused [16:12] bregma_, looks like gdk backend probing stopped working for me [16:13] bregma_, I did the GDK_BACKEND="" as you suggested [16:13] dandrader: which backend are you trying to use? mir? [16:13] bregma_, worked the first fime or so (saw client trying consecutive connections on unity8.log) [16:13] bregma_, but not anymore [16:14] dandrader: gtk from z archive checks for mir first before x11 [16:14] bregma_, now, if I do GDK_BACKEND="mir" on terminal. it all works. but with GDK_BACKEND="" I don't see any connection attempt at all [16:14] dandrader: you'd have to specify GDK_BACKEND="*", but that's a no-op [16:14] attente, yes, mir. I'm trying to [16:14] get it to try mir via backend probing [16:15] attente, ah so it's "*" no just "". [16:15] dandrader: if you want to run this under u8, you can either try ubuntu-app-launch [16:15] dandrader: you can also try adding '-- --desktop_file_hint=' after the binary name too [16:16] attente, UAL will now launch with GDK_BACKEND="x11" I believe - that's what dandrader's trying to avoid [16:16] we weren't ready for the connect, disconnect, connect dance [16:16] attente, yeah, I do the hint as well. btw, now you can use a DESKTOP_FILE_HINT env var as well. more convenient [16:16] ah, my bad [16:16] dandrader, both of which we should do away with ;P [16:17] dandrader: there's also "X-Ubuntu-XMir-Enable=false" which you can add to the desktop file of the app you want to test to disable xmir and allow gtk to skip to using mir [16:19] but yeah... GDK_BACKEND='*' should get you the original upstream probing order (mir first, then x11), assuming u-a-l doesn't override that [16:19] attente, what I want to do is try out gdk's backend probing [16:19] attente, now my app started from commandline using GDK_BACKEND="*" \o/ [16:20] dandrader: great! [16:20] attente, although It only created one connection/session. is that how it works? [16:21] dandrader: yes, it'll only connect once to the mir socket [16:21] attente, I thought bakcend probing would cause a gtk app to create a mir connection, then drop it, then finally create a new one and use it for real [16:21] attente, so probing doesn't waste connections then? awesome [16:21] dandrader: no, it just tries it, and if it works, keeps it, and moves on to the next if it fails [16:21] Saviq, ^ [16:21] Saviq, now I'm puzzled :/ [16:22] * Saviq too [16:22] dandrader: what's the issue? does it look like gtk is creating multiple connections? [16:23] attente, I was trying to fix a issue where probing seemed to create a mir connection, drop it, and then create a new final one. that caused unity8 to consider that the app stopped on the first connection drop [16:23] dandrader, one of those connections could've been xmir http://pastebin.ubuntu.com/24319827/ [16:23] Saviq, attente, a so xmir would be the one probing for mir? [16:23] but that should've worked then with your original approach, 'cause xmir stays on, even if app connects to Mir [16:24] dandrader, right now, unless X-Ubuntu..., xmir just gets launched [16:24] and should connect first, and stay connected [16:24] i thought the problem was xmir and gtk both connecting to the mir socket [16:24] or is this actually supposed to be ok? [16:24] attente, tjat [16:24] also, libSDL does the double-probe dance for whatever reason [16:24] yes, we'll need the double-double for SDL anyway [16:25] attente, yes. multiple simultaneous connections will work now with the qtmir about to be release [16:25] heh, double-double makes it sound like a joke in Candian patois [16:26] * Saviq wonders what GDK_BACKEND="" results in, then ;P [16:26] bregma_, any simple sdl app in ubuntu you recommend me to use as a test subject? [16:27] bregma_, also, can we unset GDK_BACKEND somewhere to verify GTK works? [16:27] dandrader, we user neverball for testing [16:28] bregma_, great, any env vars I should set? [16:29] Saviq, I'm not sure where GDK_BACK gets set (doesn;t seem to be in trunk UAL code), I'd just change the Exec line in a .desktop file to start with "env -u GDK_BACKEND ..." to remove that variable from the app's environment [16:29] ah -u [16:29] dandrader, er, I'm not sure of the probing order gets set properly in libSDL after recent merged... bschaefer do you know? [16:30] bregma_, SDL2 is still x11 first [16:30] dandrader, but doesn't http://pastebin.ubuntu.com/24319827/ suggest the approach in to-be-released qtmir didn't work? [16:30] ive not made that patch yet to that lib [16:30] bschaefer, what's the env var to force the libSDL probe order? [16:30] * bschaefer has plans for it [16:30] SDL_VIDEODRIVER=mir [16:30] or x11 [16:30] well, yeah, that's not "probe order' ;) [16:31] haha yeah thats not [16:31] dandrader, xmir connects, app connects, gets rejected, app dies, xmir exits [16:31] IIUC [16:31] libSDL doesn't let you specify the order by env var [16:31] just the back end [16:31] we should get that patched upstream [16:31] does u8 require upstart for anything? [16:32] attente, far as I know it still requires upstart to start itself and its dependencies up [16:32] bregma_, upstream does not want that [16:32] bschaefer, upstream needs to be slapped silly [16:32] the order is to do X11 first until Mir or Wayland take over more [16:33] bregma_, yeah ... i asked a year or so ago [16:33] bregma_, but for now we can carry it in a patch [16:33] yes, it's pretty much a one-liner patch [16:33] sdl1.2 does check for mir first though (once that lands ... hopefuly ~2 weeks) [16:33] bregma_, yup, but need to get that sdl2 patch in for mir deprecations [16:33] soo might as well add it then [16:33] bregma_: do you know what dependencies rely on that? jbicha was asking in #ubuntu-desktop [16:34] bschaefer, doyou have that patched sdl in a usable silo? [16:34] dandrader, sdl1.2? or sdl2? [16:34] i dont have either in a silo [16:34] dandrader, i can make a ppa though [16:35] bschaefer, I don't know. whatever neverball uses (that's the preferred sdl app for testing, right?) [16:35] yeah thats sdl2 [16:35] sdl1.2 gives you a bunch of other games [16:35] plus mplayer [16:35] which is awesome to watch some videos on mir [16:36] dandrader, let me get a ppa up though with that in the correct order [16:38] bschaefer, actually it should be fine without the patch. I just have to change neverballs' desktop file to not use xmir and then it will do probing and find mir, right? [16:38] yeah if it cant find the DISPLAY env [16:38] it'll fail the x11 check and move to mir [16:39] bschaefer, then it will make a probe mir connection followed by a real mir connection? [16:39] dandrader, actually that double connection part is not in the current ubuntu archive for sdl2 [16:39] that was upstreamed a month or so ago [16:40] sooo sdl2 atm only does 1 mir connection in zesty [16:42] dandrader, bregma_, ok I've confirmed now that with and without `env -u GDK_BACKEND` in the Exec line both gnome-softwares start up fine (with silo 2668) [16:43] bregma_, so we can undo GDK_BACKEND=x11 [16:43] tedg, ^^^ [16:43] dandrader, and that means it's not so much a priority any more [16:43] sorry for the noise, I wouldn't have expected ="" to behave differently than ="*" or unset altogether... [16:44] Saviq, ok, will go back to multi-instance support work then [16:46] dandrader, if you're not far away with this one, might as well complete it - will be needed anyway [16:46] but your call at this point [16:47] Saviq, was just doing the debugging so far. no real code written [16:47] ack === db` is now known as db === db is now known as Guest94484 [17:09] dandrader, did you still want that ppa (with the upstream SDL2 changes?) [17:26] bschaefer, nah [17:26] alright! Let me know if you need any testing for those changes [17:30] jeez just read the news. gutted as you all must feel [17:40] I take it this was just as much a bombshell for the Canonical devs as it is for us. Soo sorry for you guys. This should have been something awsum. [17:47] is mir development going to be discontinued as well? [20:03] personally I saw this coming months ago...