* tsdgeos kills them07:13
Saviqtsdgeos, okies, I described everything I know about the crash in https://trello.com/c/lAudAZVp/40-13-shell-rotation-mp-review-iteration-on-feature-and-autopilot-tests08:57
Saviqtsdgeos, will you please try and see what you can do about it08:58
tsdgeosfinishing a bug report and patch to qt upstream08:58
tsdgeosand will hop into that08:58
tsdgeoswell, let's see what they think of https://codereview.qt-project.org/#/c/110348/09:09
tsdgeosfixes the testcase but may not be the best thing ever :D09:09
tsdgeosSaviq: so those crashes only happen on the device? or desktop too?09:26
Saviqtsdgeos, didnae try on desktop09:27
tsdgeoslet me try it09:27
tsdgeosbecause it is defenitely nicer to debug in the desktop09:27
=== Malsasa_ is now known as Malsasa
=== marcusto_ is now known as marcustomlinson
tsdgeosnot much success trying to make it crash in the pc10:06
=== marcusto_ is now known as marcustomlinson_
=== MacSlow is now known as MacSlow|lunch
=== marcusto_ is now known as marcustomlinson_
=== MacSlow|lunch is now known as MacSlow
tsdgeosSaviq: http://paste.ubuntu.com/10821832/13:31
tsdgeosQV4_MM_AGGRESSIVE_GC=1 qmlscene it13:31
tsdgeosMacSlow: ↑↑13:32
Saviqtsdgeos, nice13:33
Saviqtsdgeos, I did see the regex bits in the trace, and remember we've had issues with that before13:33
MacSlowtsdgeos, oh ha13:42
MacSlowtsdgeos, Saviq: so since we now know what it is... looking for a home-grown fix against Qt 5.4 or try 5.5 first?13:42
tsdgeos5.5 doesn't crash with that example13:43
MacSlowtsdgeos, so 5.5 it is then?!13:43
tsdgeoswell 5.5. crashes with lots of other things :D13:43
MacSlowtsdgeos, *sigh*13:43
tsdgeosi mean it's just alpha13:43
MacSlowtsdgeos, hm... I'm using RegExp in notifications for sure.. but it sofar never blew up like that.... iirc13:44
tsdgeoswell it all depends on how/when the garbage collector kicks in13:45
MacSlowtsdgeos, I can give it a go to fix it... unless you've already an idea about the what and how for  a fix.13:45
tsdgeosMacSlow: fix 1-> remove the regexp :D fix 2->fix the js stuff in QtDeclarative13:46
tsdgeosthat's all i can think of for the moment13:48
MacSlowtsdgeos, I'll try to see if I can eliminate all RegExp from my unity8-shellRotation branch first13:49
MacSlowtsdgeos, just three uses of RegExp in all our qml-code... should not be too hard13:50
MacSlowtsdgeos, and the relevant are of course just in snap-decision notifications :/13:51
MacSlowtsdgeos, Saviq: I'll try to come up with RegExp-less work-around for my unity8-shellRotation branch, test it some more and then MP it... while waiting on review I'll turn to a fix for QtDeclaraitve. Ok?13:54
tsdgeosi'm having a look at qt at the moment13:54
MacSlowtsdgeos, ok13:54
MacSlowtsdgeos, on the side... is there perhaps a RegExp-fix in Qt 5.5 we could backport to 5.4 ourselves?13:55
* tsdgeos tips hat to MacSlow for the suggestion14:02
MacSlowtsdgeos, you found a fix in Qt 5.5?14:03
MacSlowtsdgeos, sweet!14:03
MacSlowtsdgeos, btw... http://pastebin.ubuntu.com/1082194214:03
* MacSlow just bricked his n4 *sigh*14:04
MacSlowwell not reall bricked... just re-flashing required14:04
tsdgeosso patch accepted by upstream :)14:06
tsdgeosMirv: can we get ↑ ?14:07
tsdgeosSaviq: ↑↑14:07
MacSlowtsdgeos, nice :)14:07
MacSlowSaviq, tsdgeos is the hero of the day :)14:08
Saviqthat was quick!14:08
tsdgeoscheers to MacSlow that convinced me to look at the diff14:09
tsdgeosi was going all crazy and thinking of fixing it myself14:09
greybacknice one14:10
MacSlowtsdgeos, well by now I've learned to take shortcuts as early as possible ;)14:10
MacSlowtsdgeos, on the other hand... I'm just fed up with that branch and want to get it in :)14:10
tsdgeosnow this is hoping14:11
* MacSlow high5's tsdgeos anyway14:11
tsdgeosthat this is actually that is making it crash in the run without QV4_MM_AGGRESSIVE_GC=1 scenario too14:11
* tsdgeos hi5 back14:11
tsdgeosand makes one wonder if we should run all our tests with QV4_MM_AGGRESSIVE_GC to find this stuff14:11
MacSlowtsdgeos, sounds to be worth a try... maybe we'll catch more non-determinstic issues14:12
Saviqtsdgeos, MacSlow, I'm building qtdeclarative packages with this patch, will report asap14:13
tsdgeosSaviq: awesomeness14:13
MacSlowSaviq, ok14:13
dandraderSaviq, did trunk get bad tags again or is the script outdated? http://paste.ubuntu.com/10822008/14:14
Saviqdandrader, the script doesn't get outdated any more14:14
Saviqdandrader, and yeah, we did bring those in the landing yesterday, but should be gone now14:14
MacSlowSaviq, what's the tag-"threshold"?14:15
tsdgeosyeah i killed them this morning14:15
tsdgeoswe forgot about running pot_file again :D14:16
tsdgeosanyone? ↑14:16
MacSlowtsdgeos, on it14:16
SaviqMacSlow, wdym "threshold"?14:19
SaviqMacSlow, if a tag does not point at a revision, it's bad14:19
Mirvtsdgeos: like, in vivid/vivid-rtm regardless of the qtdbus patches? ok, if you can please file a LP bug I can refer to14:20
tsdgeosMirv: yes, this is a "proper/clear" 5.4 bugfix14:20
tsdgeosMirv: sure. opening the bugfix now14:21
Mirvtsdgeos: wow! you mean that'd affect the 1421009?14:21
MacSlowSaviq, I see... but I think to remember we wanted tags clearned in the past, which were just "too old" (a few months)14:21
tsdgeosMirv: hmmm, what is 1421009 ?14:21
SaviqMirv, no14:21
tsdgeosMirv: ah no, sorry no, it's just a different bugfix14:22
SaviqMacSlow, no, it was always a case of "they're wrong"14:22
Mirvoh, no, ok. I thought some magic so that the million dbus patches wouldn't be needed :)14:22
Mirvok, I'll have a different landing for this one then14:22
tsdgeosMirv: unfortunately no :D14:22
MirvI guess it'll need to go to vivid-rtm though with final freeze on Thu14:22
SaviqMirv, OTOH this could cause relatively random crashes anywhere that's using RegExes in QML14:24
tsdgeosyeah this makes sense for vivid proper imho14:25
tsdgeosMirv: https://bugs.launchpad.net/ubuntu/+source/qtdeclarative-opensource-src/+bug/144398114:25
ubot5Ubuntu bug 1443981 in qtdeclarative-opensource-src (Ubuntu) "Protect the RegExp against early garbage collection" [Undecided,New]14:25
Saviqooh there's a W-Series14:26
Saviqdon't think there's an archive yet ;)14:26
Saviqthere's only a placeholder in LP14:26
MacSlowtsdgeos, approved your pot_file branch... will top-approve it when jenkins is happy too14:27
Mirvtsdgeos: thanks14:28
MacSlowbut we don't have a funky animal yet... or did I miss that?14:28
MacSlowwith w it's going to be interesting...14:29
MacSlowwe could get back to  :)14:30
davmor2MacSlow: Wascling Wabbit </elma_fudd_impression>14:30
davmor2MacSlow: Wonderful Whippet14:31
MacSlowdavmor2, "whippet"?14:37
davmor2MacSlow: http://en.wikipedia.org/wiki/Whippet think smaller, slower greyhound14:38
Saviqwombat FTW!14:39
Saviqheh, it takes LP 1h50 to build qtdeclarative... looks like some 25m on arale ;)14:46
tsdgeosSaviq: really?14:52
tsdgeoswe need a arale powered ci14:52
tsdgeoswith someone continously clicking the screen14:52
Saviqtsdgeos, it's not -j9 on LP I think14:52
tsdgeosSaviq: does the hardware even has that many cores?14:52
Saviqtsdgeos, arale has 814:53
Saviqtsdgeos, 2x414:53
tsdgeosSaviq: sure i mean the CI14:53
tsdgeosSaviq: it's actually 4+4 no? 4 big 4 small14:53
Saviqtsdgeos, oh no, that's part of why they're not -j9 ;)14:53
Saviqyeah 4+414:53
Saviqtsdgeos, but I mean I don't think they parallelize at all14:53
* Saviq checks to see he's not full of sh*t14:54
=== dandrader is now known as dandrader|afk
Saviqgreyback, should there not be a 1:1 dependency between qtdeclarative5-qtmir-plugin and qtmir-{android,desktop}?14:59
* Saviq sees the plugin updated but not qtmir-android14:59
greybackSaviq: think there should be yes15:00
Saviqgreyback, dandrader|afk, how about the shell-rotation-aware qtubuntu, should anything pull it in?15:01
mzanettiMacSlow, Wollmilchsau15:01
MacSlowmzanetti, well... I would like that :)15:01
mzanetti+1 :D15:01
MacSlowmzanetti, to be honest Linux is that almost... there's hardly anything it cannot do or become :)15:02
greybackSaviq: usual thing which pulls it in is ubuntu-touch-session - can't have unity8 depend on it explicitly as unnecessary for desktop15:02
mzanettiMacSlow, which is why that name came to my mind15:02
Saviqgreyback, well, we could have -android | -desktop15:02
greybackSaviq: true15:02
MacSlowmzanetti, how could we convince sabdfl of that ;)15:03
Saviqgreyback, I'm just asking because I can install rotation-aware unity8 and only qtmir plugin gets pulled in with, right now15:03
mzanettiMacSlow, with this image perhaps: http://www.forexfactory.com/attachment.php?attachmentid=255957&stc=1&d=124415050115:03
MacSlowmzanetti, there have to be better examples :)15:03
greybackSaviq: a stronger package dependency chain would do no harm. Thought it kept relatively weak to stop androidy stuff installing on desktop15:08
greybackno other reason15:09
Saviqtsdgeos, MacSlow, https://owncloud.sawicz.net/index.php/s/0afe9e8a599ef8fef2f2a105e9db357d#15:17
MacSlowSaviq, grabbing...15:18
=== dandrader|afk is now known as dandrader
tsdgeosMacSlow: Saviq: looks good here, have you had any issue?15:47
Saviqtsdgeos, no crash in sight15:48
MacSlowtsdgeos, just about to finish the setup (re-flashing, pulling packages etc)15:49
Saviqwe really need to fix the on-exit crash...15:50
tsdgeosdo we have an on exit crash?15:50
Saviqtsdgeos, it's an abort in mir15:51
SaviqMacSlow, 6 subsequent successful runs of the two rotation tests15:54
Saviqand counting15:54
Saviqgot it to fail now, no crash15:58
MacSlowSaviq, what kind of failure?16:00
SaviqMacSlow, http://pastebin.ubuntu.com/10822593/16:00
MacSlowSaviq, ah... that's new16:00
MacSlowSaviq, compilation almost done here16:01
SaviqMacSlow, kk16:01
=== Malsasa_ is now known as Malsasa
MacSlowSaviq, hm... first run and it fails with http://pastebin.ubuntu.com/1082270216:27
MacSlowSaviq, might be I overwritten some of the debs out of your zip with other apt-gets16:33
MacSlowSaviq, currently it ran fine a couple of attempts16:33
MacSlowSaviq, 10 in a row without failure yet.16:34
MacSlowSaviq, hm... I still get crashes16:37
MacSlowSaviq, got crash files written for malit, unity8-dash and webbrowser-app16:38
MacSlowSaviq, at least no unity8, but then it seems the qtdeclaritive fix isn't enough16:39
MacSlowSaviq, I'll look into the crashes and send you the gathered info.16:42
=== alan_g is now known as alan_g|EOD
ZehCnaS34Is this where I go to talk about deb packages?17:05
ZehCnaS34talk about *improving* dep packaging17:06
=== dandrader is now known as dandrader|lunch
=== Malsasa_ is now known as Malsasa
=== dandrader|lunch is now known as dandrader
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader

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