[06:43] <duflu> RAOF: When would you estimate the majority of transitions to yakkety would finish?
[06:43] <RAOF> What do you mean?
[06:44] <duflu> RAOF: I mean when do you reckon the floor and mountains would stop moving
[06:44] <RAOF> Oh, roughly now.
[06:44] <duflu> Hmm, OK. Might switch soon
[06:45] <RAOF> Oh, I don't think gcc-6 is default yet. That's the last piece of outstanding foundation change I know of.
[06:45] <duflu> Sounds like a big one
[06:45] <duflu> Usually is
[06:46] <RAOF> Well, mir already builds with it :)
[06:46] <duflu> Cool, but we also depend on a whole archive that needs to work too
[06:47] <RAOF> Well, my laptop is fine with it (at the moment).
[06:47] <RAOF> The everything-stages-in-proposed switch has shaved *a lot* of the sharp edges off the development release.
[06:48] <duflu> Yeah I can put it on a laptop any time. But if my main work dev machine stops working, that's a problem
[06:49] <duflu> RAOF: What's that? We just use proposed more now?
[06:49] <duflu> Or use it properly
[06:50] <RAOF> Every upload gets diverted to proposed, and only migrates when (a) the migration does not break installability, and (b) all autopkgtests pass.
[06:50] <duflu> Awesome
[06:51] <RAOF> This is why having a autopkgtests for Mir would be valuable; every mesa upload would sit in proposed until the Mir autopkgtests completed successfully.
[06:51] <duflu> Except I'm sitting on an installability regression that's a major pain
[06:51] <duflu> Which we don't test for
[06:52] <RAOF> By “installability” I mean “every package in the archive can be installed”, not “can be installed on a PC with a new nvidia card ☺”.
[06:52] <duflu> Ooh
[06:52] <duflu> RAOF: I meant https://bugs.launchpad.net/ubuntu/+source/unity-settings-daemon/+bug/1546641
[06:53] <duflu> Which admittedly is not installability as much as acceptability
[06:54] <RAOF> Indeed.
[06:55] <RAOF> Also, isn't a bug in the development branch, but in the release :).
[06:55] <duflu> It was in development, before it got released :)
[06:55] <duflu> Two months before
[06:55] <RAOF> Right, but this is less a question about yakkety.
[06:56]  * duflu continues searching for unofficial EGL extensions
[11:27] <kdub> alf__, are you back today?
[11:28] <alf__> kdub: yes
[11:28] <kdub> alf__, welcome back
[11:28] <alf__> kdub: thanks!
[11:29] <kdub> alf__, just wanted to point out https://bugs.launchpad.net/mir/+bug/1577357, wondering if you remember anything about that bug/workaround where we try to force mir_demo_server to link to pthread
[11:30]  * kdub will keep chipping at it after i've got 1578159 squared away
[11:31] <kdub> probably would be nice to have in 0.23 if its simple (my experiments with it monday werent sucessful though)
[11:32] <kdub> so if its not, we'll just tag to 0.24
[11:43] <alf__> kdub: The original issue was seen only when cross-compiling with clang+ld.gold (IIRC). The packaged binaries are built with gcc+ld.bfd, right?
[11:44] <kdub> yes, they should be
[11:45] <alf__> kdub: In any case, it's very strange that the executable doesn't link with pthread even though we have the work around (i.e. explicitly use pthread symbols)
[11:45] <kdub> right, and I've traced through enough to see that -pthread is there in the options