=== maclin1 is now known as maclin
=== nudtrobert1 is now known as nudtrobert
=== zbenjamin_ is now known as zbenjamin
tsdgeosah funny09:09
tsdgeosthe qmluitests in tableStage pass here09:10
tsdgeosmeh the autopkg tests branch has degraded into tests not passing again09:29
tsdgeosat least one i know what it is09:31
ltinkltsdgeos, https://code.launchpad.net/~lukas-kde/unity8/globalshortcuts/+merge/269608 merged and conflicts resolved09:31
tsdgeosltinkl: ok, i'll let daniel continue the review09:32
ltinkltsdgeos, it had been approved already09:32
tsdgeosah right09:32
ltinkltsdgeos, by mzanetti too09:33
ltinkltsdgeos, thx09:35
=== greyback__ is now known as greyback
=== alan_g is now known as alan_g|llunch
tsdgeoswops, testNotifications broke10:09
* tsdgeos fixes10:09
tsdgeosSaviq: was speaking with someone yesterday that had exactly the same issue with the cropped images as we do10:25
tsdgeosmaybe after a bit of testing we can propose the CroppedImageMinimumSourceSize up to the SDK?10:26
Saviqtsdgeos, I'd rather we propose a change to the Qt APIs10:26
tsdgeosSaviq: what would you propose?10:26
Saviqtsdgeos, not sure yet, sourceSize: minimumSize(200, 300) or something of the sort10:26
SaviqQt.minimumSize I mean10:26
Saviqwould be backwards compatible10:27
tsdgeosi guess it might work10:28
kgunnpstolowski: hey there, i'm trying to untangle a mess10:43
kgunnso i understand lp:unity-api and lp:unity-api/trunk-15.04 are different10:43
kgunnwas hoping to just merge trunk15.04 differences into lp:unity-api10:43
kgunnbut heard  you might not think that's kosher10:44
kgunnis there any reason these are being maintained seperately ?10:44
tsdgeos:/ qmltestrunner::ShellWithPin::test_longLeftEdgeDrags is unstable11:04
pstolowskikgunn, hey,11:41
pstolowskikgunn, that happened during gcc5.0 mayhem. tbh i'm not sure anymore why i branched off unity-api since it doesn't have symbols, we could use a single tree (but we can't have single tree for e.g. shell plugin)11:44
pstolowskikgunn, having said that, both trunks are equal feature-wise, may have different revisions and commits as i cherry-picked11:45
pstolowskikgunn, ah, afair it had to do with needing to land stuff for ota6 while dual landing was not possible due to gcc511:49
=== alan_g|llunch is now known as alan_g
Saviqpstolowski, so we could reconcile now? why is it not possible to have a single tree for the shell plugin? can't do some #ifdefs?12:23
Saviqpstolowski, they're built separately for wily and v+o after all?12:24
Saviqor is it about the .symbols file differing between the two?12:24
pstolowskiSaviq, due to symbols12:24
Saviqthat feels dumb, but I feel the pain12:25
=== MacSlow is now known as MacSlow|lunch
Saviqpstolowski, even if separate branches, landing can be done in sync, right? so it's not like there's actual code differences12:26
pstolowskiSaviq, we can probably try to merge unity-api trunks back (i'm not sure if it poses any problems wrt citrain, but they are probably easy to work out)12:26
SaviqI mean that the fact that you need two branches there doesn't mean any dependants need to have two12:26
Saviqpstolowski, that'd be ideal, I'll compile a list of tasks for this to happen12:27
pstolowskiSaviq, correct - but you need to have separate MPs for unity-api, meaning you cannot just dual land12:27
pstolowskiSaviq, okay. for shell plugin it's not easily doable though12:28
pstolowskiSaviq, we have similar problem in unity-scopes-api (symbols)12:28
Saviqpstolowski, as long as the two (wily and v+o) changes land in sync, we're fine12:28
pstolowskiSaviq, michi has been working on a single-tree beanch for unity-scopes-api, it's huge effort12:28
Saviqpstolowski, it's only a problem when one lags behind the other12:28
pstolowskiSaviq, yeah. we just need two MPs as before with rtm12:29
pstolowskifeature-wise all trunks should be the same12:29
pstolowskialecu, ho?12:33
tsdgeosanyone feels like reviewing https://code.launchpad.net/~aacid/unity8/autopilot_test_search_workaround/+merge/269769 ?12:48
tsdgeosmterry: https://code.launchpad.net/~aacid/unity8/fixTestNotifications/+merge/26987612:48
mterrytsdgeos, whoops, will review that  :)12:49
mzanettimterry, good morning12:50
mterrymzanetti, hello!12:50
mzanettimterry, when you get a chance, please merge your reboot branch12:50
mterrymzanetti, will do12:50
kgunnSaviq if i followed your convo with pstolowski correctly , i think this just needs to be landed in wily first12:59
kgunnthen...src sync that to vivid+o12:59
kgunnwell...plus qtmir trunk into wily with that unity-api mp ^12:59
kgunni'll leave it with you...12:59
Saviqkgunn, something like that, I need to see the what the actual direction (wily to v+o or the other way round) will be, but in truth we just need the next landing to be dual-targeted and then push to trunks and drop the additional branches13:00
tsdgeosgreyback: any idea of what may be causing the failures in https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-vivid/939/consoleFull ?13:00
tsdgeosgreyback: the tabletStage ones13:01
greybackSaviq: fyi seems tags have infected qtmir somehow13:01
tsdgeosgreyback: search for "fail!"13:01
Saviqgreyback, yeah, mzanetti told me13:01
mzanettistill not sure how tho :D13:01
Saviq"reached maximum number of OpenGL contexts supported by UbuntuShape"13:01
greybacktsdgeos: "fail!" ?13:01
tsdgeosgreyback: without the quotes13:02
Saviqand that in DDA test??13:02
tsdgeosSaviq: SDK13:02
tsdgeosSaviq: it's already fixed they just need to land it13:02
greybacktsdgeos: yeah, firefox not seeing it. Ah matching case, sorry13:02
Saviqah so that's not the fail you're looking for13:02
* greyback waves hand13:02
tsdgeosSaviq: no since that's not on the phablet stage ;)13:02
tsdgeosSaviq: this is the one for the DDA one https://code.launchpad.net/~aacid/ubuntu-ui-toolkit/betterConnectForAboutToBeDestroyed/+merge/26845413:03
greybacktsdgeos: we had this same problem like 2 weeks ago, no?13:04
tsdgeosgreyback: did we?13:04
greybackyeah. UbuntuShape doing something funky13:04
greybackI thought there was a fix somewhere, looking13:04
tsdgeosgreyback: yes yes, i'm not speaking about the ubuntushape13:04
Saviqgreyback, look for further fail!13:04
greybackah sorry13:04
tsdgeosgreyback: i'm speaking about the tabletstage fails13:04
tsdgeosgreyback: which obviously i can't reproduce here13:05
greybacktsdgeos: hmm  the test makes sense too13:07
greybacksomething racey maybe13:08
greybacktho this is all single thread I think13:08
greybacktsdgeos: only thing I can suspect is switchToApp - it looks quite solid tho13:11
tsdgeosyeah :/13:11
tsdgeosand is used in other places too i think13:11
greybackdandrader: good morning, there is a multimonitor update for you13:11
greybacktsdgeos: I see it used in 2 tests, both are the ones which fail13:12
tsdgeosmaybe it's that then13:12
dandradergreyback, ok13:12
greybackjust a guess tho, it runs fine here annoyingly13:13
tsdgeosi'll try to investigate from there13:13
greybackwould be lovely to get videos of test fails13:13
greybackilke AP does13:13
Saviqgreyback, I've a thing doing that in my pipeline13:15
Saviqas in mostly implemented13:15
greybackSaviq: glad to hear it. Would be extra super nice if it visualize the cursor/gestures13:16
Saviqgreyback, I'm just using recordmydesktop, should be possible to replace that with something smarter, if that doesn't do it13:16
Saviqgreyback, ah but for QML tests we'd need something in the qmlscene doing that13:17
Saviqsince there's no real input13:17
Saviqnext step13:17
greybacka custom qmlscene might be needed13:18
greybackanyhoo, video will help muchos13:18
mterrytsdgeos, you might have some insight on bug 1489076?13:18
ubot5bug 1489076 in unity8 (Ubuntu) "clicking on active call banner and indicators with mouse" [Critical,New] https://launchpad.net/bugs/148907613:18
tsdgeosmterry: hmmm think you want dandrader for that?13:19
tsdgeosi haven't really done much DDA nor windowed work13:19
mterrytsdgeos, yup, I think I mentally swapped you two  :P13:19
tsdgeosok :D13:20
mterrytsdgeos, which is weird because dandrader even has "DandD" in his name13:20
mterrydandrader, anyway ^13:20
Saviqmterry, dandrader, shouldn't we just have a MouseArea there for onClicked? DDA doesn't care about mouse events anyway?13:22
dandradermterry, I would have to investigate the bug to be able to provide any decent insight13:22
dandraderSaviq, define "there"13:22
Saviqdandrader, wherever the DDA is, really13:23
Saviqwherever we want the click to act ;)13:23
mterrySaviq, does it pass them through?  that might work then, to extend the MouseArea from the left of the panel to all along the panel, underneath the indicators13:23
dandraderthe indicators bar already has a MouseArea13:23
mterrydandrader, yeah but it doesn't extend underneath the indicators, as I recall13:23
dandraderwhich is why you're able show the indicators panel by simply clicking on the bar13:24
dandradermterry, there might be a MouseArea "input eater" over there, I don't lnow13:24
mterrydandrader, the indicators panel shows with a click?  ok, good...  Then *something* is eating the mouse event / my quick investigation was wrong13:25
Saviqmterry, in any case, DDA shouldn't be involved with mouse events13:25
Saviqthere's obviously the problem of mouse events being converted to touch, and vice versa...13:26
mterrySaviq, (why not?  it conceptually makes sense that it could handle a mouse drag -- just work we haven't done or is it something we think shouldn't be done?)13:26
Saviqbut as long as you accept the event, it shouldn't happen13:26
dandradermterry, what do you mean by "indicators panel"? it that this narrow bar at the top of the thing you get when you expand it13:26
Saviqmterry, mouse gestures need to be different than touch13:26
dandradermterry, dednick added a MouseArea in the indicator bar a while ago to handle this call thingy I think13:26
mterrydandrader, I specifically meant the IndicatorsMenu object13:27
mterrydandrader, yeah, but it only is on left of IndicatorsMenu13:27
* dandrader not very familiar with the indicator components terminology13:27
mterrySaviq, yeah but it would be nice if we had a class that abstracted that for us13:27
mterryand without looking at it's API, DDA strikes me as a possible fit for that13:28
mterrydandrader, then I meant "the row of icons"13:28
tsdgeosmterry: https://code.launchpad.net/~aacid/unity8/fixTestNotifications/+merge/269876/comments/679272 up to you, do i do it?13:29
Saviqmterry, sure, but it wouldn't be DDA but something above it, including a MouseArea, likely ;)13:29
* Saviq away otp13:29
mterrytsdgeos, I think it's a cleaner branch if we do it.  But I feel bad asking you to do it, since it's my mistake in the first place.  I can propose a merge into your MP today13:30
tsdgeosmterry: no no, i'll do it13:30
tsdgeosdon't worry13:30
tsdgeosi'm on test failure hunting today13:31
tsdgeosmzanetti: can you do https://code.launchpad.net/~aacid/unity8/stabilize_shellwithpin_test/+merge/269913 ?13:31
mzanettitsdgeos, sure13:32
tsdgeosmterry: setStatus pushed13:46
mterrytsdgeos, that was a few calls indeed  :)13:48
mzanettitsdgeos, this seems to have the ability to mess up with the state machine... doesn't it?13:51
tsdgeosmzanetti: not really i mean the problem is that for some reason we're using a timer to set the state13:52
tsdgeosso if you do13:52
tsdgeosstarttimer that changes state, change state13:52
tsdgeosyou really need to stop the timer first13:52
tsdgeosotherwise you may not get the state you wanted13:52
mzanettiyes, but we don't do "change state"13:52
mzanettido we?13:52
tsdgeosmzanetti: if you we do, i can move that down on the animation code if you prefer13:53
mzanettilet me understand it better...13:53
mzanettitsdgeos, we do... you're right13:53
mzanettiok... seems fine then13:54
tsdgeosmaybe it makes more sense to stop the timer just before the13:54
tsdgeosroot.state = "";13:54
tsdgeoson the ScriptAction { ?13:54
tsdgeosi don't mind either13:54
mzanettiprobably... so we don't run into this again13:54
tsdgeosk, will move it down there13:54
mzanettiif someone else starts the fadeOut() , which admittedly is unlikely13:54
tsdgeosi don't even think you can reproduce this in the real world13:55
=== MacSlow|lunch is now known as MacSlow
tsdgeosmzanetti: moved down14:18
mzanettitsdgeos, ta14:19
kgunngreyback: mzanetti how do twins currently work ? it's been a while...do i have to manually update watch file still ?14:27
kgunnor just have an mp for -gles and build it in the silo ?14:27
mzanettikgunn, yeah, no change14:28
greybackkgunn: yep14:28
mzanettikgunn, debian/watch for the silo number and debian/changelog for the version14:28
kgunnmzanetti: ack no change except those, got it14:28
greybackand keep eye out for new dependencies14:28
kgunngreyback: so compare the control files ? between the twins14:29
* kgunn always wants to think parent-child vs twins14:29
greybackkgunn: yeah, just in case the 'parent' has something added14:29
mzanettiif the branch you're releasing changes the control, copy the changes over14:30
kgunneasy nuff14:30
mzanetti@unity: standup14:30
kgunnmzanetti: i'll miss14:30
=== dandrader_ is now known as dandrader
Saviq@unity, kgunn, pstolowski: here's what I believe we should do to resync trunks between wily and vivid, anything I'm missing? http://pastebin.ubuntu.com/12253846/14:48
Saviqkgunn, vivid has mir 0.14.1, wily has 0.15.1, what's the plan there?14:48
pstolowskiSaviq, looks good overall, but i think citrain will reject the first step (land a src rebuild of lp:unity-api/trunk-15.04 to wily) without some changelog hackery?14:51
Saviqpstolowski, I think it shouldn't, the version is higher than what's there in wily14:52
Saviqpstolowski, ultimately we can skip the train altogether and just upload the package to proposed directly14:53
pstolowskiSaviq, the ...+15.04 part of version string will confuse it14:53
Saviqpstolowski, don't think it looks that closely, we were dual-landing to rtm and distro before14:54
Saviqwithout the -rtm suffix14:54
Saviqand it was fine14:54
pstolowskiSaviq, ok, we will see :), i fought with something like that not long ago14:55
pstolowskiSaviq, i guess we just need to try14:55
mzanettiSaviq, looks good to me14:56
Saviqpstolowski, yup14:56
Saviqback in ~2.5h14:57
tsdgeosSaviq: overlay has 0.15 too, no?14:58
tsdgeosSaviq: ah no, ignore me14:58
mzanettigreyback, kgunn: what's the ETA for silo0 things to land? I cannot reproduce bug 1488828 but I don't have slimport support and silo0 is out of date15:40
ubot5bug 1488828 in qtmir (Ubuntu) "incoming call nexus4 windowed mode goes nuts and reboots" [Undecided,Incomplete] https://launchpad.net/bugs/148882815:41
greybackmzanetti: silo0 more a demo silo than full of actually landable things15:42
greybackmzanetti: we're taking things a piece at a time15:42
mzanettigreyback, sure. I'm particularly interested in the slimport support15:42
greybackmzanetti: well that's probably a bit of https://code.launchpad.net/~gerboland/qtmir/multimonitor/+merge/26990615:43
mzanettiah ok15:43
greybackbut there's also some magic in silo0 to automatically use desktop mode if >1 screens available15:44
greybackwhich you'll not have15:44
greybackmaybe convert to desktop mode by attaching BT mouse & keyboard, then emulate the call?15:44
dednickdandrader, mterry: I've commented on 1489076, but need a UX decision.15:46
mzanettigreyback, that's fine... I can do that15:48
mzanettigreyback, I've done incoming calls in desktop mode very often lately and didn't have a crash15:48
mzanettiso the only thing that's different here is the missing slimport15:48
greybackmzanetti: then it must be silo0 specific15:48
mzanettilet me try to do the exact same phonesim steps first (I usually use a real phone call)15:49
tsdgeosmterry: in exchange for fixing the notification thing can you review https://code.launchpad.net/~aacid/unity8/autopilot_test_search_workaround/+merge/269769 ? ;)15:49
mterrytsdgeos, :)  sure15:49
tsdgeosmeh all the qmluitests jobs are stuck15:58
tsdgeoshttp://s-jenkins.ubuntu-ci:8080/job/unity-phablet-qmluitests-vivid/ :/15:58
tsdgeosmzanetti: anyone we can ping? ↑15:59
mzanettitsdgeos, cihelp in #ubuntu-ci-eng15:59
tsdgeosk, need to do paperwork16:00
* tsdgeos runs16:00
syehHi, newbie Unity question, how do I check out the code for 14.04?  I think the version I want is 7.2.5+14.04.20150603-0ubuntu116:38
syehI've done a "bzr branch lp:unity" and got a version of source16:39
syehBut for some reason that version won't start after I've make and "make install"16:40
dednicksyeh: if you want the source for the version which you are currently running, "apt-get source unity"16:48
syehdednick:  I am trying to fix a defect, and so I need to get the tree, test it, then push it16:49
dednicksyeh: "bzr branch lp:unity/7.2"16:51
dednicksyeh: https://code.launchpad.net/~unity-team/unity/trusty16:52
syehdednick:  Thanks!   Trying that now.16:52
=== dandrader is now known as dandrader|lunch
=== dandrader|lunch is now known as dandrader
syehChrisTownsend:  in unityshell.cpp, is UNITY_LOW_GFX_MODE supposed to be an override in the case that it is set to "0"18:35
ChrisTownsendsyeh: Right, so if you set UNITY_LOW_GFX_MODE=1 before starting unity, then unity will be started in low graphics mode.18:37
syehwhat if I set it to 0, can I force it to not be in low gfx mode?18:37
syehThe current code won't allow that, but I'm wondering if that's the intention18:37
ChrisTownsendsyeh: Oh, I see.  If it's being forced into low graphics mode without UNITY_LOW_GFX_MODE being set, it's due to your hardware configuration.  Setting UNITY_LOW_GFX_MODE=0 will not force non-low graphics mode.18:40
syehRight, so would it be okay if I make it so that a user can force it to *not* go into low gfx mode?18:41
syehA couple of drivers are now reporting "LLVM" in their renderer string.18:41
syehI'm working on a Unity patch to fix that, and maybe providing a way to override the behavior if necessary18:42
ChrisTownsendsyeh: Hmm, are they really LLVM drivers or is it a case they just happen to use that string and Unity is not accounting for that?18:43
ChrisTownsendsyeh: I just want to make sure 3d is not enabled for real LLVM drivers and the user experience will be terrible.18:43
syehGallium drivers have a fall back path for certain draw operations (if they need it)18:45
syehand the Gallium Draw module can use LLVM if it is available18:46
syehit is different from the llvmpipe driver, which is a SW driver18:46
syehSo I think Unity should check for "llvmpipe" not "LLVM"18:47
ChrisTownsendsyeh: Sure, that makes sense.18:48
syehok.  What about the override path for UNITY_LOW_GFX_MODE = 0?18:48
syehDoes that make sense or should I leave it the way it currently is?18:49
ChrisTownsendsyeh: I think you should do one merge proposal for the LLVM vs. llvmpipe change.  Then you could do a second one for the override and see if it be accepted.  I no longer work on Unity, so others will have to decide if it should be taken or not.  You may want to discuss that will Trevinho when he's available.18:56
syehok.  Will do.  Thanks.18:57
kgunnSaviq hey so i'm not getting my way on working around that little unity-api/qtmir19:09
kgunnany eta on when i might be able to dual land a qtmir again ?19:10
Trevinhosyeh: I'm currently travelling in China so I'll be more reachable next week. If you have a MP ready would be nice.19:21
syehTrevinho:  I can't get through the firewall to push my branch, so I've attached it here:19:22
ubot5Ubuntu bug 1491555 in Unity "Unity unnecessarily goes to low graphics mode" [Undecided,New]19:22
Saviqkgunn, I want to prepare a silo with what I wrote in the pastebin tomorrow19:36
Saviqkgunn, but that might very well be a dual-landing of qtmir, too19:37
Saviqor well, *will* be, we can add commits to it if you want19:41
kgunnSAviq i think so...just need to stay in sync with bregma and camako, but yeah, it prolly makes sense to use this20:19
kgunnand then there's one for usc20:20
kgunnor one would need to be made rather....20:21
kgunnbregma: ^ this is the unity-api sorting, since it's all just no change rebuilds for mir/usc/qtmir this could be done in one shot20:23
bregmaassuming unity-api is fixed properly20:24
kgunnbregma: precisely the point of Saviq's landing...."it just might work"tm20:30
bregmawe'll get it straightened out while you're gone20:32

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