/srv/irclogs.ubuntu.com/2015/04/27/#ubuntu-unity.txt

yaakurotrying to write an applicaton that is using _NET_WM_WINDOW_TYPE_DOCK06:10
yaakuroit seems like that Unity does not allow window drag operations on that window06:10
yaakurois that a feature or bug?06:10
Saviq@unity: see allhands about new IRC server bits08:06
davmor2Saviq: you forgot didn't you ;)08:24
Saviqdavmor2, *I* didn't, I just didn't see anyone there ;P08:26
* Saviq had his bouncer ready mid-last-week :P08:26
davmor2Saviq: I just let irssi die quietly and restarted it this morning with the right creds08:27
davmor2Saviq: jibel was having issues with his creds08:28
Saviqdavmor2, I had to talk some sense into my bouncer, too, it took quite some time to connect08:28
SaviqCimi, /msg *status listservers, no need to kill znc ;)08:29
=== Guest87721 is now known as huangzhiquan
CimiSaviq, ok but is not working :/08:54
sil2100Saviq: ping o/09:02
sil2100Saviq: hey, do you know if you could have someone from the unity8 team assigned to help Mirv identifying the cause of bug LP: #1421009 ?09:03
ubot5Launchpad bug 1421009 in qtbase-opensource-src (Ubuntu) "unity8 sometimes hangs on boot" [Critical,In progress] https://launchpad.net/bugs/142100909:03
sil2100Saviq: we would appreciate experts from all related components as this is a blocker for our vivid RC image09:03
sil2100So that's rather critical ;)09:03
Saviqsil2100, tsdgeos has been working with Mirv on this for the past three week09:04
Saviqs09:04
Saviqsil2100, Mirv, have we decided against the Qt patchset?09:05
tsdgeossil2100: we've talked about that, the cause for it has been identified already, qtdbus deadlocks09:05
tsdgeosthat's the cause09:05
sil2100tsdgeos, Saviq: we need a working workaround somewhere, tvoss had a proposition how to make it better09:06
MirvSaviq: the patchset seems to cause regressions and also tsdgeos voted against using it before it's merged in upstream 5.5, so we need a workaround09:06
sil2100The Qt patchset is not a solution as it causes serious regressions in other parts of the system09:06
sil2100Right09:07
tsdgeoshonestly if its my vote that's holding this09:07
tsdgeosi'm changing it09:07
sil2100tsdgeos: well, we'd need to get the regressions fixed in this case, for which we need people from various teams working on it09:07
Mirvthere was now a tip of using QDBusConnection::connectToBus(BusType type, constQString & name) instead of QDBusConnection QDBusConnection::sessionBus()/systemBus() to minimize sharing of dbus connections09:07
Mirvboth unity8 and libusermetrics seem to use only the latter ones so that could help09:08
=== cimi is now known as Cimi
Saviqtsdgeos, can you prep an MP replacing the shared objects from ↑ in u8?09:11
Saviqpete-woods1, same for libusermetrics ↑09:11
sil2100If this would be enough to at least work-around the problem, we could finally unblock the whole release again09:12
tsdgeosSaviq: i can, i guess it's easy enough to try it09:13
pete-woods1Saviq: sure, will get that MP toegether09:14
sil2100Thanks guys09:14
* Mirv stops trying to do it for libusermetrics since I'm sure pete will do it faster+better09:15
Mirvthe tip was from tvoss09:16
Mirvpete-woods1: in case there's anything else that could be tried to workaround the Qt bug, maybe the backtrace helps https://launchpadlibrarian.net/204606516/usermetrics-fullbacktrace.txt - dbus_bus_add_match ... sender='com.canonical.UserMetrics',path='/com/canonical/UserMetrics/DataSource/1',interface='com.canonical.usermetrics.DataSource',member='formatStringChanged'" or alternatively eg path='/com/canonical/UserMetrics/DataSource/2' .. member='em09:26
=== Malsasa_ is now known as Malsasa
pete-woods1Mirv: https://code.launchpad.net/~unity-api-team/libusermetrics/dont-share-dbus-connections/+merge/25749709:38
=== pete-woods1 is now known as pete-woods
SaviqMacSlow, you need to strip tags on your shellRotation branch09:44
MacSlowSaviq, *sigh* how did... ok :)09:44
SaviqMacSlow, they came from shellrotation09:44
SaviqMacSlow, remember to strip on your local checkout, too09:48
MacSlowyup09:49
MacSlowSaviq, wiped local and remote ?-tags10:02
tsdgeosMirv: https://code.launchpad.net/~aacid/unity8/useOwnDbusConnections/+merge/25750010:04
tsdgeosSaviq: mzanetti: ↑↑↑↑10:04
Mirvthanks pete & albert! I'll build them in silo 710:10
Mirvthen both can be tested first together to see if any improvement.10:12
sil2100Mirv: keep us up-to-date if it's helping or not o/10:17
Mirvsil2100: if you have again your arale free, prepare it with the latest instructions in the bug description (just added dbg packages - but you can leave out usermetrics as it'll be a different version now) - it'll be useful to get full backtraces if we get the bug again, to see if anything changed10:19
Mirvsil2100: since it may take anything from 5 to 100 reboots, running on two devices should get us "results" (even though unwanted) faster. likewise, if there's help from the MP:s, we can build better confidence on the changes working with two devices10:19
sil2100Mirv: let me power up my arale10:20
Mirvsil2100: thanks, just concentrate on other stuff otherwise, I'll ping when you add the PPA and start the reboot loop :)10:20
=== spy is now known as Guest31921
sil2100Mirv: I need to charge it up first so it might take a bit10:35
tsdgeosMacSlow: you doing https://code.launchpad.net/~dandrader/unity8/autoInstallTouchRegistry/+merge/256726 ?10:37
tsdgeospstolowski: can you confirm https://code.launchpad.net/~saviq/unity8/drop-scopes-env/+merge/256962 is good?10:38
pstolowskitsdgeos, done10:41
MacSlowtsdgeos, yup10:47
tsdgeosMacSlow: okidoki :)10:47
=== willcooke_ is now known as willcooke
=== olli is now known as Olli
=== MacSlow is now known as MacSlow|lunch
=== alan_g is now known as alan_g|reboot
=== Guest31921 is now known as huangzhiquan
Saviqtsdgeos, can you please look through inline comments in suspend_screenshoting (they only show in "Preview Diff" of r1742 into r1745), not all of them have been addressed yet11:38
tsdgeosSaviq: sure11:39
tsdgeosi'm fighting back the dbus stuff11:39
tsdgeosseems you can't really use the connectToBus like that all the time11:39
Saviqtsdgeos, mhm, that's higher prio of course11:39
=== alan_g is now known as alan_g|lunch
Mirvsil2100: ok you can add 007, make sure libusermetrics gets upgraded (unity8 started failing a test so couldn't be built) and start the reboot loop. if it hangs, get bt full with those mentioned debug packages installed.11:56
Mirvpete-woods: is https://code.launchpad.net/~timo-jyrinki/libusermetrics/dont-share-dbus-sessionbus-connections/+merge/257505 totally pointless or worth testing in parallel?11:57
pete-woodsMirv: you've made more or less the exact same change as me11:58
pete-woodsoh , wait11:58
pete-woodsno you've not11:58
Mirvpete-woods: you did for system bus, I did for session bus11:59
=== dandrader is now known as dandrader|afk
pete-woodsMirv: I think you may have accidentally uncovered a stupid mistake in my code there12:08
Mirvpete-woods: I'm always happy to help, even when I don't know how :) it's not blowing up at least, I'm running the reboot loop test with your branch (only) on mako at the moment.12:11
sil2100Mirv: will do after lunch12:40
=== MacSlow|lunch is now known as MacSlow
Mirvthis is probably one of those annoyingly long random times when the bootloop seems promising... at 45 and counting12:50
Mirvlet's see somewhere around 7512:51
=== alan_g|lunch is now known as alan_g
tsdgeosMirv: i'm pretty confident the tests that fail are the tests "fault" (or rather they need a bit or rearchitecturing if we change this)12:57
tsdgeosMirv: so i'll comment the tests for now so yu can get to a stage we have passing tests so you can actually live test it12:57
tsdgeosand meanwhile fix it properly12:58
Mirvtsdgeos: ok.12:59
tsdgeosMirv: pushed, please try rebuilding the silo13:12
=== dandrader|afk is now known as dandrader
Mirvtsdgeos: thanks, I'll have another silo so that we can test just the libusermetrics too13:20
tsdgeosMirv: cool13:21
tsdgeospstolowski: we're on mandrelbot14:14
pete-woodsMirv: it will conflict with your MR, btw14:20
Mirvpete-woods: yeah I don't care about my MR, I care about a working workaround :) currently just your first revision seems good for me on mako at least - 105 reboots without hangs. sil2100 is starting on arale, although I guess we might need to rebuild libusermetrics if you think your original branch (just replacing systemBus usage) was somehow wrong, not only partial? (partial is ok if it workarounds the problem and doesn't cause regressions)14:25
sil2100We're fine with workarounds ;p14:26
sil2100As long as it's a workaround without any regressions14:26
Mirvsil2100: so the story here is that pete's original branch (that we're testing in 007) only changed DBus system bus usage to use the tipped function. my branch does the same for session bus usage, and I guess Pete is now changing his MP to do both (and maybe something else?), but as said we don't currently care that much of how completely the workaround is implemented everywhere as long as it fixes the boot hang14:27
Mirvbut we might want to retest with Pete's updated MP too, once we test that current 007 is cool14:27
pete-woodsMirv: I think it's probably worth including the extra change, as it's making shared connections to the session bus, also14:28
Mirvpete-woods: probably so, it does not harm either and might prevent some other lock case.14:30
Mirvpete-woods: oh, now I see what you meant I uncovered something useful :)14:31
Mirvregarding removing the paths more properly14:32
MacSlowdandrader, approved your autoInstallTouchRegistry branch14:57
MacSlowdandrader, still some (unrelated) AP-test failures though14:57
dandraderMacSlow, ok14:58
MacSlowSaviq, btw... with the flake8 issues fixed... are you ok with lp:~macslow/unity8/shellRotation too? Just trying to get it top-approved too15:01
SaviqMacSlow, I'll have a quick last look tomorrow morning15:02
MacSlowSaviq, ok15:02
dandraderMacSlow, could you please also top-approve it?15:05
MacSlowdandrader, doh... sure15:06
dandraderthanks!15:07
=== karni is now known as karni-afk
=== elopio_ is now known as elopio
=== dandrader is now known as dandrader|lunch
tsdgeospstolowski: http://paste.ubuntu.com/10913181/16:29
jameshMirv: hi.  Do you have any estimates on when you QQuickAsyncImageProvider backport (https://bugs.launchpad.net/ubuntu/+source/qtdeclarative-opensource-src/+bug/1439263)16:29
ubot5Ubuntu bug 1439263 in qtdeclarative-opensource-src (Ubuntu) "Backport QQuickAsyncImageProvider from upstream-dev" [Undecided,In progress]16:29
jameshgar.  "when you'll have a chance to look at the"16:30
tsdgeospstolowski: components["art"]["noartsummaryshape"]16:31
mterrytsdgeos, can you explain bug 1421009 to me a bit?  Is there something wrong with sharing connection objects via QDBusConnection::systemBus()?16:44
ubot5bug 1421009 in qtbase-opensource-src (Ubuntu) "unity8 sometimes hangs on boot" [Critical,In progress] https://launchpad.net/bugs/142100916:44
tsdgeostsdgeos: other than it may deadlock if you're using it concurrently between various threads16:44
tsdgeosno16:44
tsdgeos:D16:44
tsdgeosdamn here i am talking to myself again16:45
tsdgeosmterry: ↑↑↑16:45
mterrytsdgeos, oh qtdbus isn't thread-safe?16:45
tsdgeosmterry: it is16:45
tsdgeosbut it has bugs :D16:45
mterrytsdgeos, ok  :)16:45
mterrytsdgeos, didn't know if systemBus() was something I should avoid using or not.  Sounds like it's a fine pattern, just not right this second then  :)16:46
mterrytsdgeos, cheers16:46
greyback_tsdgeos: "safe but has bugs" <- I hope my car manufacturer doesn't use that slogan16:46
tsdgeosgreyback_: well isn't all software like that?16:47
tsdgeosit works, except when it doesn't16:47
greyback_tsdgeos: yep. Just don't go into marketing ;)16:48
tsdgeospstolowski: https://docs.google.com/document/d/1NmiM4UCnJgf6IEawmfyTOHRNAA5ZGrqpyrPqPOibwc8/edit16:49
Mirvjamesh: hi, last week but then this critical bug started taking all my time. if it seems the bug would be now soon fixed, I'll assign a silo to myself tomorrow (spreadsheet line #51), backport the qtdeclarative in there and I can ping you. I guess the landing would happen only with the associated thumbnailer branch, but do you have something else to land with it too?16:53
jameshMirv: I haven't started porting thumbnailer over to the new API because I had nothing to test against16:55
Mirvtsdgeos: at the current rate it'd seem we have a working workaround with just libusermetrics. that'd mean that landing only that is the 1st priority and the unity8 landing can be lower priority.16:55
jameshtotally understand delaying the work to handle the unity8 hang though.16:55
tsdgeosMirv: if libusermetrics fixes it already, i'd rather not land the unity8 patch at all unless we find out we totally need it16:55
Mirvjamesh: right. so your branch would be based on / replace the wip https://code.launchpad.net/~aacid/thumbnailer/asyncprovider/+merge/253188 ?16:55
jameshMirv: it would replace it.  We've made a fair number of changes since then.16:56
Mirvtsdgeos: sure, if it stays like that, that libusermetrics is all we need, it's not needed to be landed at all. after all, the goal is to get all the upstream fixes eventually when they work.16:56
Mirvjamesh: ok, thanks!16:56
=== alan_g is now known as alan_g|EOD
=== dandrader|lunch is now known as dandrader
=== dandrader is now known as dandrader|afk
kgunnChrisTownsend: hey, so after the update to oxide and everything, i still get apps seemingly bailing out no unity8 desktop19:15
kgunni was going to file a new bug, unless you already had19:16
ChrisTownsendkgunn: Hmm, I haven't tried recently.  Lemm try it now.19:16
ChrisTownsendkgunn: Yeah, I get the same behavior.  I'd say file the bug.19:18
kgunnChrisTownsend: also, is there a bug for any keyboard input not working...it works for some, but not for all19:19
kgunne.g. if you go to the store, and then get prompted for credentials19:19
kgunnit won't take input19:20
ChrisTownsendkgunn: Yes, that is long outstanding bug.19:20
kgunnChrisTownsend: also do you guys have any plans to expand the apps available ?19:20
kgunnterminal would be so convenient for debug :)19:21
ChrisTownsendkgunn: Well, there is a terminal app, it's just doesn't work very well...at least the last time I tried.19:21
ChrisTownsendkgunn: But I'm not sure what you mean by us expanding the apps available.19:23
kgunnChrisTownsend: meaning preloaded19:25
kgunnatm, i only see browser, sys settings, and "system testing"19:25
kgunnand that's all19:26
=== dandrader|afk is now known as dandrader
ChrisTownsendkgunn: Yeah, you have to install it.  The Ubuntu Next ISO is seeded to have more apps (not the terminal).  If you load the recommended dependencies for unity8-desktop-session-mir, you'll get more apps.19:27
kgunnah19:27
ChrisTownsendOr maybe it's suggested, but you know what I mean.19:28
kgunnyeah :)19:28
ChrisTownsendkgunn: I'm purging the overlay PPA and seeing if the oxide crash happens since there is a new official version of oxide in the archive.19:29
kgunnChrisTownsend: sure, i think it's unrelated....cause sys settings also does the same thing19:30
ChrisTownsendkgunn: Yep, same thing.  I'll file a bug since it's the official package.19:30
kgunnor is sys settings a webapp19:30
ChrisTownsendkgunn: Hmm, not sure.  It might use it to render, but I'm just guessing.19:31
=== karni-afk is now known as karni
kgunnif you file a bug lemme know...i have a unity8 log...seems its got some cgroup complaints19:31
kgunni don't think this is oxides fault(?)19:31
ChrisTownsendkgunn: Ok, I'll run apport on it here in a bit.19:31
ChrisTownsendUgh, cgroups19:32
kgunni've got it here...i don't mind doing19:33
kgunni'll share bug # in a sec19:33
ChrisTownsendkgunn: Are you uploading the crash file?19:35
kgunnChrisTownsend: well i was going to...but apport-cli seems to just spew dots when i say 'view'19:36
ChrisTownsendkgunn: I'm uploading my crash now.19:37
kgunnholy cow, it just enormous19:40
kgunn(and kind of on an old machine...)19:40
ChrisTownsendkgunn: https://bugs/launchpad.net/bugs/144923119:40
ChrisTownsendkgunn: Whoops, bad copy19:40
ChrisTownsendkgunn: We'll see what the retracer can do, but symbols may be no good.19:41
ChrisTownsendkgunn: But the crash is definitely occurring in Oxide.19:42
kgunnChrisTownsend: interesting, so i cleared out all my crash files and logs, just to be sure...interesting, i don't see any crash file for oxide/browser...unless i'm looking in the wrong spot ?20:00
kgunn/var/crash/20:00
ChrisTownsendkgunn: That's where it should be.20:01
kgunni only have an ibus crash file20:01
ChrisTownsendkgunn: Hmm20:01
ChrisTownsendkgunn: I definitely get a crash in webbrowser.20:02
kgunni'll keep fiddling20:02
ChrisTownsendkgunn: You cleared out ~/.cache/upstart as well?20:02
kgunnyep20:02
ChrisTownsendNothing regarding webbrowser in there?20:03
kgunnlemm check20:03
ChrisTownsendkgunn: And maybe something called application-failed.log?20:03
kgunnnope just the ibus20:03
ChrisTownsendSo webbrowser is not even starting for you.20:04
kgunnnope...lemme take a quick video, maybe i see something diff20:04
ChrisTownsendok.  I get webbrowser starting for a bit, but nothing is rendered where web content should be and then it crashes.20:05
ChrisTownsendkgunn: Ok, retracer was able to do it's magic.  Hopefully someone with Oxide knowledge will look at this.20:07
kgunnChrisTownsend: cool...here's what i see20:09
kgunnhttps://www.youtube.com/watch?v=3HMNhqrTbgI20:09
kgunni'm on a relatively clean machine...wonder if i'm missing something20:10
ChrisTownsendkgunn: Yeah, you definitely have a different symptom.20:10
ChrisTownsendkgunn: Yours looks like u-a-l doesn't like something.20:11
justAn8hi21:51
justAn8will unity 8 use compiz?21:52

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!