[02:55] <rsalveti> robert_ancell: getting bug 1364725 on krillin
[02:56] <rsalveti> krillin + ubuntu-touch from proposed
[02:56] <robert_ancell> rsalveti, got the lightdm.log after it occurs?
[02:56] <rsalveti> can easily get that for you, 1 sec
[02:58] <rsalveti> robert_ancell: https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/1364725/comments/1
[02:59] <robert_ancell> hmm, that looks normal to me
[03:00] <rsalveti> the main difference when comparing with the older one is:
[03:00] <rsalveti> [+2.00s] DEBUG: Seat: Switching to Mir session session-0
[03:00] <rsalveti> [+2.00s] DEBUG: Activating login1 session /org/freedesktop/login1/session/c2
[03:00] <rsalveti> I don't get such lines
[03:00] <rsalveti> but unity8 is running fine
[03:01] <rsalveti> let me try getting a screenshot
[03:01] <rsalveti> no, screenshot is also all black
[03:01] <robert_ancell> Right. So it's like LightDM hasn't told u-s-c that that session is ready
[03:03] <rsalveti> also added the log diff to the bug
[03:04] <robert_ancell> rsalveti, could you attach both logs as files? The diff is a bit confusing due to the timing
[03:05] <rsalveti> sure
[03:05] <robert_ancell> the diff shows more information from the log than you pasted
[03:07] <rsalveti> all attached
[03:07] <robert_ancell> rsalveti, could you also run lightdm --show-config on the device and paste that?
[03:08] <rsalveti> robert_ancell: from 1.11.6 or 1.11.7?
[03:08] <robert_ancell> 1.11.7
[03:08] <rsalveti> or doesn't make any difference
[03:08] <rsalveti> sure
[03:08] <robert_ancell> it was only introduced then
[03:08] <rsalveti> robert_ancell: https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/1364725/comments/7
[03:09] <rsalveti> robert_ancell: this doesn't happen all the time though, sometimes unity8 shows up just fine
[03:09] <robert_ancell> oh, ok
[03:09] <rsalveti> maybe a race
[03:10] <rsalveti> let me try to get into a working state (harder than actually reproducing the bug)
[03:19] <rsalveti> robert_ancell: https://launchpadlibrarian.net/183907934/lightdm.log with 1.11.7 from a working boot
[03:20] <rsalveti> so yeah, didn't switch to mir session
[03:21] <robert_ancell> definitely a race of some sort then
[03:30] <robert_ancell> rsalveti, could you try lp:~robert-ancell/lightdm/unity-activate and see if that makes a difference?
[03:32] <robert_ancell> rsalveti, Also, looking at ~/.xsession-errors might give some clies
[03:32] <robert_ancell> clues
[03:33] <rsalveti> robert_ancell: sure
[03:45] <rsalveti> nothing useful in ~/.xsession-errors
[03:45] <rsalveti> building from your branch
[03:58] <rsalveti> robert_ancell: rebooted more than 10 times, worked fine at every time
[04:01] <robert_ancell> rsalveti, cool
[04:07] <robert_ancell> rsalveti, do you have a log from a successful boot with that change? It was just a hunch that suggested that code might have broken but it really shouldn't work...
[04:08] <rsalveti> robert_ancell: http://paste.ubuntu.com/8220528/
[04:09] <robert_ancell> ah, so I was right. It's actually failed to find the logind session id which is a very bad thing (TM). That was exposing a bug that my branch fixed
[04:09] <robert_ancell> I'm working on a proper fix now
[04:10] <rsalveti> awesome
[04:35] <robert_ancell> rsalveti, could you try lp:~robert-ancell/lightdm/logind-session-id-env-var and see if that works
[04:35] <robert_ancell> This is the good fix
[04:36] <rsalveti> robert_ancell: sure
[04:49]  * robert_ancell hates the feeling when you'd seen some flaky code at some point in the past and not bothered to fix it then it comes to bite you in the butt later on...
[04:57] <rsalveti> robert_ancell: yeah, seems fine as well
[04:57] <rsalveti> unable to reproduce the issue
[04:57] <robert_ancell> rsalveti, ok, cool. I'll ship that then
[04:57] <rsalveti> robert_ancell: awesome, thanks
[04:57] <robert_ancell> Have you got a log so I can double check?
[04:58] <rsalveti> robert_ancell: http://paste.ubuntu.com/8220784/
[04:58] <robert_ancell> that looks better
[05:45] <robert_ancell> rsalveti, brb, just smoke testing
[05:46] <robert_ancell> rsalveti, ok, all good. Uploaded as 1.11.8-0ubuntu1
[05:46]  * robert_ancell -> dinner
[05:46] <rsalveti> robert_ancell: awesome, thanks!
[06:23] <duflu> RAOF: Oh, the problem with SIGPIPE happens with SIGHUP too. It's just some pre-existing unreliability in the test framework - https://bugs.launchpad.net/mir/+bug/1364772
[06:23] <duflu> So we could have used SIGPIPE. Need to fix the test infrastructure tho
[06:26] <RAOF> Well, sigPIIIIIIIIIIIPE is up for review :)
[06:29] <duflu> RAOF: I know. Haven't got around to it yet
[06:33] <RAOF> That's fine; it's got a prerequisite branch anyway :)
[10:26] <Saviq> hmm guys how do you think would that be possible:
[10:26] <Saviq> unity-system-compositor: symbol lookup error: /usr/lib/arm-linux-gnueabihf/libmirserver.so.24: undefined symbol:
[10:26] <Saviq> _ZTIN7android7RefBaseE
[10:26] <Saviq> seems libmircommon1:armhf got upgraded as I was installing unity8-autopilot
[10:26] <Saviq> (that's RTM btw)
[10:28] <Saviq> but that must've not pulled in new mirserver or so
[10:28] <ogra_> new mir isnt in rtm yet
[10:29] <ogra_> it landed in 218 in devel-proposed but didnt migrate over yet
[10:29] <Saviq> http://pastebin.ubuntu.com/8222494/
[10:29] <Saviq> ogra_, well, yeah, but package deps should not have allowed me to get into that state
[10:29] <Saviq> ogra_, should force upgrades where required
[10:30] <ogra_> then you need a versioned dep
[10:30] <Saviq> ogra_, as now I ended up with a mix of mir 0.6.1 and 0.7.0
[10:30] <ogra_> defining a minimal version in unity
[10:30] <Saviq> ogra_, it's not that, unity doesn't even depend on mir directly
[10:31] <ogra_> well, it also seems like the new packages migrated but arent in the image
[10:31] <ogra_> (new rtm image is currently building, so i guess that will have it)
[10:31] <Saviq> ogra_, sure, still ;)
[10:31] <Saviq> ogra_, http://pastebin.ubuntu.com/8222504/
[10:32] <ogra_> right, there is some dependency screw up somewhere
[10:32] <Saviq> I only asked apt to install unity8-autopilot - that pulled in libmircommon1 0.7.0, that should've pulled others with it, as there seems to have been an ABI break
[10:32] <Saviq> because u-s-c doesn't start
[10:32] <Saviq> complaining about missing symbols
[10:33] <ogra_> right
[10:33] <Saviq> *something* should've forced upgrades of the whole of mir I'd say
[10:33] <ogra_> file a bug so a dep gets added
[10:33] <Saviq> yeah, will do, just wanted some feedback from the guys here
[10:34] <anpok> hmm
[10:34] <anpok> why does qtmir-plugin only depend on mircommon?
[10:35] <anpok> i would have thought anything of qtmir needs mirserver
[10:36] <anpok> is qtmir dloading mir?
[10:37] <anpok> hm ok qtdeclarative5-qtmir-plugin only pulls in mircommon while qtmir-android adds mirserver..
[10:38] <Saviq> bug #1364890
[11:16] <greyback> anpok: qtmir-{android,desktop} needs mirserver, as that is the code which creates a mir server. qtdeclarative5-qtmir-plugin just allows Qt/QML to interact with the mir server, so needs some of the common mir stuff (mostly enums I think)
[14:17] <racarr> Good morning!
[14:21] <alf_> racarr: Hi!
[14:33] <racarr> alf_: :) How goes?
[14:34] <anpok> hm I was curious what was happening there: https://code.launchpad.net/~raof/mir/dont-kill-the-poor-clients/+merge/232971/comments/568627 but I cannot reproduce locally
[14:34] <anpok> has anybody else looked at that yet?
[14:45] <racarr> not yet