[00:01] I like when commit messages come off like plot twists [00:02] "The DispatcherController is the InputFocusSelector" (audience gasps) [00:19] racarr, haha :D [00:20] RAOF, use_dma_buf landed \o/ congrats! [00:20] robert_ancell: Even better, I'm about to mark enable-inprocess-egl for landing, too ;) [00:20] RAOF, woohoo! [00:33] yay coed landing [00:33] code [02:03] robert_ancell: autolanding & CI should be working for lp:lightdm - do you have a MP we can use to test it? [02:03] thomi, not currently but I'll make something [02:03] robert_ancell: cool - if/when we know that's working I'll enable it for the mir branch as well [02:04] robert_ancell: ping me when you have a MP link and I'll make sure everything's hunky-dory on the backend [02:04] thomi, ok [02:13] thomi, https://code.launchpad.net/~robert-ancell/lightdm/whitespace/+merge/158264 [02:13] awesome, I'll follow that and make sure everything works as expected [02:13] do you mind if I approve it once the CI stuff has done it's thing? [02:17] thomi, sure [02:56] robert_ancell: can you please add 'ps-jenkins' to the lightdm team? This allows the CI task to comment on MPs. [02:56] thomi, done [02:56] thomi, do you need a second MP? [02:56] robert_ancell: nah [02:57] the first one will get picked up again [03:30] robert_ancell: CI worked (second time around, anyway): https://code.launchpad.net/~robert-ancell/lightdm/whitespace/+merge/158264 [03:30] merging it now [03:31] thomi, weird error on first attempt!? [03:31] robert_ancell: lightdm doesn't build on quantal chroots without addinga build-dep [03:31] thomi, ah [03:31] the first time around we had it running ona quantal box [03:31] so we change the config and set it running again [03:32] autolanding is running... [03:51] robert_ancell: Re performance issues... Usually it is std::string construction/destruction that's the biggest problem. Because event a string on the stack will indirectly be allocated to the heap. And heaps are slow. [03:51] But string comparison is usually very fast. With Rabin-Karp (and better) algorithms in use in all the STLs [03:52] -event +even ;) [03:53] Hmm, Rabin-Karp is for searching. But still, string comparison is relatively fast [05:13] duflu, protobuf support fixed-sized arrays, too. Might be an option to look into that, too [05:13] tvoss|sick: Hmm, yeah good idea. Assuming some profiler tells us that's the problem [05:14] duflu, agreed [05:14] tvoss|sick: Though seriously, we should have lots of room to do lots of things 60 times per second and stay around 0% CPU. There must be more to it [05:15] duflu, I think running it through cachegrind again should reveal other hotspots [05:15] tvoss|sick: Would love to. Once my other work is done [05:15] tvoss|sick: Go rest! [05:16] duflu, feeling better already :) [05:16] thomi, still around? [05:16] tvoss|sick: And if not resting then please look at normalize-MirEvent :) [05:16] tvoss|sick: Careful to not confuse cachegrind for callgrind [05:17] duflu, I meant callgrind :) [05:17] Which reminds me, we should try helgrind. It's bound to tell us our locking sucks [05:17] duflu, that's its purpose, yes :) [05:58] duflu, approved [05:58] tvoss|sick: Thanks. You can rest now ;) [06:02] duflu, nice :) [07:57] tvoss|sick: Hi [07:57] tvoss|sick: what's up? [08:00] * thomi heads off to read his book. [08:31] * robert_ancell notices 58 or so users here. This channel has got a lot busier! [09:20] hi, this is a bit offtopic, but I was told you guys probably have the best idea on how to go about this, I want get kwin + X + hardware acceleration working on the nexus 10 [09:20] can I just plop the graphics driver from here https://developers.google.com/android/nexus/drivers#manta [09:29] shadeslayer: Yes that's quite off-topic. You might get an answer in #ubuntu-desktop or #ubuntu-unity [09:29] ack [14:28] racarr: \0/ ....in process egl landing === francisco is now known as Guest35830 [15:08] good morning folks, working on cleaning up my nativewindow branch, i think i've finally overcome the big hurdles [15:09] kdub: good morning [15:09] status: Iterated VT switching branch, started investigation of pending "High" priority bugs [15:10] kdub: I am still getting a frozen screen (only the first frame) on the nexus 4, have you seen that? [15:11] no :/ havent seen that yet [15:12] maybe you could send me your phone image using the recovery mode? [15:13] kdub: well, it's just a normal ubuntu image... I will investigate a bit more and as a last resort I will send you the image (or try updating again first) [15:14] kdub: could the problem (hypothetically) be due to a different hardware revision? [15:15] alf_, could be, although i don't think its likely... could you send the 'adb logcat' of when it starts? [15:19] kdub: http://paste.ubuntu.com/5698763/ [15:21] alf_, i think surfaceflinger is running, based on that [15:21] mv /system/bin/surfaceflinger /system/bin/sf [15:21] then reboot. the system won't be able to find it :) [15:22] kdub: yes, I haven't yet stopped it (it's a logcat of device startup, did you mean a logcat of mir startup?) [15:22] right of mir startup [15:35] kdub: http://paste.ubuntu.com/5698810/ (this is render_to_fb, surfaceflinger certainly not running) [15:35] kdub: Like before, I can only see the first frame [15:38] kdub: hmmm, I think that my cross-compiler may have been inadvertently upgraded to the raring version at some point... [15:38] kdub: let me downgrade and try again [15:39] kdub: (I have a feeling that this is the cause) [15:39] alf_, yeah, i think i got hangs last time that happened to me too. something in chrono if i recall [15:49] kdub: yes, that was it, sorry for the false alarm [15:50] alf_, cool :) [16:06] 189 character-long function declaration, i think thats a new record :) [19:24] yay not having to juggle mesa trees between my branches anymore XD [19:57] morning [19:58] Morning [20:40] robert_ancell: gotta bug out early for baseball today....i'll be on later [20:41] kgunn, bye [20:59] robert_ancell: Ready when you are [20:59] *announcer voice* The 1-1 Sync-Up hangout everyone has been waiting for! Robert - Robert. Live on G+ at 9:00 UTC. [20:59] :p [22:17] racarr, do you know the correct way for a shell to know when a session is connected? [22:34] robert_ancell: Hmm...I don't think there is a most correct way at the moment [22:34] I mean you could listen to the SessionMediatorReport but that seems incorrect [22:34] racarr, yeah, I'm trying to find *any* hook at the moment to know when sessions appear/disappear [22:34] robert_ancell: ^ [22:34] yes, I thought so too [22:34] But um. [22:35] I mean also you could replace various parts of the Shell configuration [22:35] we should add a SessionManagerListener [22:35] I ant to do this also to invert [22:35] well, change, the control flow for focus setting [22:35] robert_ancell: It will be really easy to add a SM listener with https://code.launchpad.net/~robertcarr/mir/ease-shell-configuration [22:36] racarr, yes, that's what I want [22:37] I think I'll just wait for that to land [22:41] Ok. hopefully tomorro [23:01] yay android jenkins for xkbcommon-mapper fixed itself [23:08] robert_ancell: got a second? [23:08] thomi, sure [23:09] robert_ancell: I'm just looking through https://code.launchpad.net/lightdm/+activereviews to make sure that the CI process is working as expected before enabling it for the mir branch [23:09] robert_ancell: I notice that lots of CI jobs are failing due to a lack of commit message on the MP [23:10] robert_ancell: I wanted to make sure you were aware that this is a requirement for the autolanding process to work - it uses the commit message in debian/changelog [23:10] thomi, yep [23:10] ...although we can turn that requirement off, but it's not recommended [23:11] other than that, it looks like everything is running flawlessly, so I'm going to add the mir branch now [23:11] thomi, awesome, thanks [23:11] thomi, I'll fix up the MPs now [23:11] robert_ancell: you mentioned there was an extra build-dep for the mir branch - is that libmirclient-dev? [23:11] thomi, yes [23:12] coolio [23:12] thomi, oh, it there going to be some sort of fix for Jenkins giving links to 's-jenkins'? [23:13] robert_ancell: I keep asking Martin to fix that, but he won't - his "solution" is to add s-jenkins to your /etc/hosts file, which... ewwwww [23:13] thomi, but you also need VPN access right? [23:13] robert_ancell: exactly [23:13] robert_ancell: and you can specify DNS servers to be used when you connect to the VPN [23:14] thomi, so it should at least say "YOU NEED VPN ACCESS" because these branches are public and it will confuse people [23:14] robert_ancell: right, we can also publish the jenkins results publically, which will chanegt he links to point to jenkins.qa.ubuntu.com [23:14] which I'll look into doing today as well [23:15] thomi, oh, the other thing that confuses me is. I enable the VPN, then I click the link and I can't work out what to do to make it rebuilde [23:15] thomi, do I also need to log in? [23:15] robert_ancell: it will rebuild automatically when the MP changes [23:15] so, if you push a new revision, for example [23:16] thomi, so in the case of https://code.launchpad.net/~mterry/lightdm/enums/+merge/156695 I need to rebuild because the commit message hadn't been set [23:16] It even tells me I have to trigger the rebuild myself :) [23:16] you can trigger a rebuild manually, which is sometimes needed, and you do need to be logged in for that [23:16] right [23:16] do you have an account on that jenkins box? [23:16] if not, I can create one [23:17] thomi, that's where I'm not sure - what would my login be? [23:17] let me see if I can find you [23:19] you don't exist :-/ [23:19] I'll create you [23:19] robert_ancell: what username would you like? [23:20] thomi, robert-ancell so it matches my LP one and I don't forget it :) [23:20] thomi, any chance that will be linked so SSO? [23:20] robert_ancell: hahaaaaa... no :) [23:27] robert_ancell: lp:lightdm/mir should be released into ppa:mir-team/staging correct? [23:28] thomi, yes [23:29] robert_ancell, if i want to add hwc1.0 support to the blueprints somewhere, which one would be the one to do that to? [23:30] kdub, I think you'd attach that to a blueprint for the particular device that requires it [23:30] robert_ancell, that would be the galaxy nexus... can we re-open blueprints, or should we start new ones? [23:35] kdub, just reopen it