[09:06] Hi! Is there an API to get current frame rate of unity8-dash ? [09:08] duflu, ...do you know ? ^ [09:08] om26er: Yes, I wrote it :) [09:08] Just a sec [09:10] $ restart unity8-dash MIR_CLIENT_PERF_REPORT=log [09:10] $ tail -f ~/.cache/upstart/unity8-dash.log [09:10] Although vsync creates confusingly limited numbers so you can also try: [09:11] $ restart unity8-dash MIR_CLIENT_PERF_REPORT=log MIR_CLIENT_FORCE_SWAP_INTERVAL=0 [09:11] $ tail -f ~/.cache/upstart/unity8-dash.log [09:11] om26er ^ [09:12] om26er: Also with results: https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1564721 [09:12] Launchpad bug 1564721 in Canonical System Image "Unity8-dash render time is too high, causing frame skipping" [High,Confirmed] [09:17] duflu, so I have to grep fps out of that log file, hmm. [09:17] I am planning to write a test for https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1563287 [09:17] Launchpad bug 1563287 in unity8 (Ubuntu) "[regression] Right edge switcher stutters badly with only a few apps opened" [High,Confirmed] [09:18] om26er: Yes you do. Qt/QML also has graphical numbers, but those are disabled in code. The Unity8 team is working on turning on the feature this week [09:18] which I can use to narrow down there the regression happened. Like flashing the device with increment of +1 then running the test for fps etc [09:18] Qt/QML also has *on screen* numbers [09:19] om26er: I reflashed my krillin today but unfortunately could not get it to stutter. Only smooth in the switcher [09:19] duflu, which image number ? [09:19] I would also like to add on-screen reporting in Mir itself. But there's resistance within the team to let me do that (only because some people say text rendering doesn't belong in Mir) [09:20] I have r305 and the lag is quite visible. I have six apps running. [09:20] om26er: 308 ... and it seems smooth [09:20] Although I probably never ran more than 4 or 5 apps [09:22] duflu, having something for fps in Mir sounds like a good thing, given we know there may be clients other than unity8 as well. [09:22] I will update to latest. [09:23] om26er: Well we've had built-in performance logging for all Mir clients since around late 2014 at least. It's only to the log though [09:23] That's also a good thing, as the reporting does not affect the performance itself [09:24] You can use MIR_CLIENT_PERF_REPORT=log on unity8 too ... since it's a nested server it is a Mir client [12:11] so what's next for surface-wm? [12:11] greyback, ^ [12:12] dandrader: you fixed all the tests? [12:12] greyback, yes [12:12] dandrader: great. [12:13] then we should try landing [12:13] greyback, silo 013? [12:13] greyback, we should also have the prereq branches there [12:13] dandrader: I doubt it. I'm not the lander these days. Saviq around? [12:13] that was more a silo for testing convenience [12:15] wassup? [12:16] dandrader, last night there was one more failure [12:16] dandrader, https://unity8-jenkins.ubuntu.com/job/test-ppa-autopkgtest/90/ [12:24] Saviq, fixed [12:31] greyback, "Lander IRC Nicks: greyback" wdym "you're not the lander"? [12:32] Saviq: the usual, fobbing off responsibility ;) [12:34] greyback, would you say it's ready to land, then? [12:35] Saviq: I want to do a quick sanity pass, but my answer is yes [12:40] greyback, ok, then - Lander: Approved to get the autopkgtest machinery going - please describe a bit what to pay attention to during QA [13:25] dednick, https://requests.ci-train.ubuntu.com/#/ticket/736 can be abandoned probably? [13:25] Saviq: yep === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [14:53] tsdgeos, one small thing still, sorry: https://code.launchpad.net/~aacid/unity8/enable_debugging_flag/+merge/290969 [14:53] tsdgeos, but yes, I think this is much better like this, no need to force people setting 2 things just to enable 1 thing [14:54] mzanetti: i copied the code form qtbase [14:54] so i don't think it's wrong [14:54] mzanetti: see https://github.com/qtproject/qtbase/blob/dev/src/corelib/kernel/qcoreapplication.cpp around line 195 [14:55] or rather https://github.com/qtproject/qtbase/blob/5.5/src/corelib/kernel/qcoreapplication.cpp [14:55] hmm... strange [14:55] around line 190 [14:56] which is the 5.5 version [14:56] 5.4 version is the same https://github.com/qtproject/qtbase/blob/5.4/src/corelib/kernel/qcoreapplication.cpp [14:57] hmm... I mean... I didn't invent that debug print I posted... [14:58] mzanetti: i know [14:58] that's the parser being stupid [14:58] doesn't the thing still get enabled? [14:58] nope, didn't work to connect [15:00] let me check [15:04] tsdgeos, actually it does work [15:04] mzanetti: works here [15:04] not sure what went wrong before then [15:04] yeah [15:04] http://paste.ubuntu.com/15689587/ [15:05] ok... so it's just about the debug print that's bogus [15:05] wfm then [15:06] approved === dandrader is now known as dandrader|afk [15:29] dandrader|afk, testSpreadDelegate segfaulted [15:30] as did testShell... but that one went away with what(): std::bad_alloc [15:30] so OOM? [15:30] hmm it took it over ½h to throw that [15:31] but it did it twice... [16:18] dandrader, https://unity8-jenkins.ubuntu.com/job/test-ppa-autopkgtest/99/testReport/ testSpreadDelegate segfaults on vivid quite reliably (three times already) [16:19] I've started https://unity8-jenkins.ubuntu.com/job/test-ppa-autopkgtest/100/ to check again [16:20] Saviq, how can you tell it's a segfault from that webpage? [16:21] dandrader, usually when it's empty it's a segfault - only thing to do is to grep through the console output of the particular build [16:22] so you go to https://unity8-jenkins.ubuntu.com/job/test-ppa-autopkgtest/99/label=amd64,package=unity8,release=vivid+overlay,testname=qmluitests.sh/consoleFull and look for that test [16:22] they run in parallel, so output is a bit meh to read [16:22] but [16:22] 16:11:15 Segmentation fault (core dumped) [16:22] 16:11:15 tests/qmltests/CMakeFiles/xvfbtestSpreadDelegate.dir/build.make:52: recipe for target 'tests/qmltests/CMakeFiles/xvfbtestSpreadDelegate' failed [16:24] Saviq, yeah, there's a segfault somewhere. will run that on the phone to see what I get (no vivid box here) [16:25] Saviq, console is a bit useless since outputs are interleaved. tests are being run in parallel, aren't they? [16:30] hmm, I wonder if it's even possible to run qmltests on the phone... [16:31] I think I'm better off with a vm [16:33] Saviq, is silo 76 landing before surface-wm? [16:42] dandrader, I was planning that we land surface-wm first - we're close enough [16:42] Saviq, great [16:44] Saviq, you think a Ubuntu 15.10 image would be close enough to 15.04? [16:44] Saviq, can't seem to find a vivid image [16:44] dandrader, doubt it, why not just a chroot? [16:45] it should be enough to repro - that's what jenkins is doing [16:46] Saviq, ok, will try that [16:49] any guidance on how to use capslock as control in unity8? === dandrader is now known as dandrader|afk [16:53] hallyn, don't think it's possible atm, but try and ask in #ubuntu-mir [16:54] kind of a dealbreaker [16:54] will do thx [16:55] you can always swap the keys on your keyboard, will that help? ;) [17:02] no i can't they're different sizes [17:02] i did used to build a custom kernel with different map, before i used ubuntu... [17:02] maybe it's time to bring that back [17:03] also, the browser (on amd64) is still busted? [17:04] ... i'm... i'm gonna do it [17:04] (kenrel) === alan_g is now known as alan_g|EOW [17:06] is the browser crash deemed low prio bc noone should be usnig unity8 on desktop yet? [17:14] hallyn, wouldn't say it's low prio, just there's always things with higher prio... but yeah, outside of the devices we ship u8/mir is still pretty experimental [17:14] dandrader_, got it to crash here [17:14] will run under gdb [17:20] Saviq: so atm the only app which works for me (which, really, is sufficient -i'm using it right now) is the terminal. should any other apps work? [17:27] hallyn, anything that you grab a x86 click for... [17:28] hallyn, unfortunately store is a bit broken, but you can go to the store, click install (it will fail due to pkcon - working on it) and then you can "click install..." it [17:28] after you find it in ~/.cache/... [17:34] Saviq: is there a url where i can see the list of x86 clicks? [17:37] hallyn, https://uappexplorer.com/apps?arch=x86_64 [17:40] hm, empty page [17:42] hallyn, hmm? loads 66 pages of apps for me... === dandrader_ is now known as dandrader [17:43] Saviq: http://paste.ubuntu.com/15694247/ [17:44] hallyn, if you go to https://uappexplorer.com/ ? [17:44] Saviq, got anything? I was away [17:44] dandrader, yeah, just installing symbols [17:44] i'm sure i'm supposed to be able to do something with that, but since i have no gui browser working in unity8, i'm trying wget and edbrowse, and nieth like sit [17:44] hallyn, right, that page is JS [17:45] edbrowse does js. this is not just js [17:45] all right i'l lhave to look at it later then - thanks [17:46] hallyn, or just browse the Ubuntu Store - it will only display clicks working on your arch [17:47] dandrader, it does segfault maybe every 3 runs here in a vivid+overlay chroot [17:48] Saviq: i should be able to browse the ubuntu store from the scopes somehow? [17:48] hallyn, yeah, do you not have a huge orange Ubuntu Store icon below... the 3 or so apps? [17:49] nope [17:49] snappy scope is there... [17:49] or do you mean on the launcher? [17:49] if so then no [17:50] no, on Scopes [17:50] * hallyn installs the unity-scope-scopes pkg [17:51] do i need to log out and back in? :( [17:51] no [17:51] and unity-socpe-scopes is descoped :P [17:51] did you install unity8-desktop-session-mir? that should've brought unity-scope-click with it [17:51] gr [17:51] yeah i have that [17:51] so i should *un*install unity-scope-scopes? [17:52] doesn't matter [17:52] ok [17:52] i hope i'm being informative rather than wasting your time [17:53] yeah it's fine, I just don't have an idea why you wouldn't see the store :/ [17:57] dandrader, can't get any symbols out of it :/ [17:57] great [17:57] not even QV4_FORCE_INTERPRETER=1 helps [17:58] what's even weirder can't get gdb itself to run the test [17:58] had to rely on corefile, which might be why I'm not seeing symbols [17:58] aah dumb [17:58] qtchooser [17:58] Saviq, I'm still prepparing my vivid chroot... [17:58] much better [17:59] dandrader, got symbols, pasting in a sec [18:01] dandrader, http://paste.ubuntu.com/15694690/ [18:03] Saviq, nice, thanks [18:04] dandrader, lemme know if I can help with something [18:26] Saviq, I got an assertion failure on the secon drun [18:26] Saviq, did you build with debug on? [18:31] Saviq, pushed a change (revision 2352). See if it still crashes for you [18:31] Saviq, I'm running it in a loop here and no crash or assertion failure so far [18:55] dandrader, no, no debug - it was packages as built