=== salem_ is now known as _salem
kgunntsdgeos: hey, do you have a galaxy nexus?08:03
tsdgeoskgunn: no, a nexus408:07
kgunntsdgeos: mm, thanks...i might come back...but i'll try not to :)08:08
tsdgeoscome on, can someone ban morphis for a while?08:09
Saviqtsdgeos, mzanetti, could one of you guys please try and investigate bug #120699108:58
ubot5bug 1206991 in Unity 8 "Frequent CPU hogging" [Critical,Triaged] https://launchpad.net/bugs/120699108:58
tsdgeosSaviq: sure, which image, today's one? or?08:59
Saviqtsdgeos, all of 'em, more or less ;)09:01
Saviqtsdgeos, it just gets stuck at 30-40% CPU at times (although I have no real way to reproduce)09:01
mzanettiSaviq: ah... happens on desktop all the time for me09:01
Saviqmzanetti, good, you're it09:02
mzanettiI just run unity8 and my fans start running.09:02
mzanettiok. I'll check it out09:02
tsdgeosmzanetti: ok, yours it is09:02
Saviqmzanetti, it seems to settle after a minute or so here (on desktop)09:05
Saviqmzanetti, "14,08%  unity8  libQt5Quick.so.5.0.2                [.] QSGNodeUpdater::isNodeBlocked(QSGNode*, QSGNode*) const" is the biggest one for me on desktop, per perf09:07
mzanettiSaviq: I'll check it out in a few. still discussing about LVWPH with tsdgeos right now.09:09
Saviqmzanetti, yup, thanks09:09
tsdgeosnooo, dandrer is on holiday already?09:15
tsdgeosah, ignore me :D09:15
Saviqtsdgeos, yeah, and out for the whole month!09:42
kgunnwho authorized that?09:48
seb128pete-woods, hey, just check, is the new glibmm in saucy working for you now?09:48
dednicklarsu: ping09:49
seb128dednick, he's at GUADEC for a week, he might not be around a lot (just mentioning it so you are not surprised if he doesn't pong) ... maybe let some question/context so he can reply when he reads backlog09:50
seb128dednick, or use emails09:50
dednickseb128: ok. thanks09:51
pete-woodsseb128: perfect, thanks!09:52
seb128pete-woods, excellent, yw ;-)09:53
seb128pete-woods, thanks for the updated patch!09:53
pete-woodsseb128: not a problem!09:54
tsdgeosoh may09:58
tsdgeosi mean09:58
tsdgeos    Unassigned "09:58
tsdgeosit'll be fixed fast!09:58
mzanettitsdgeos: otoh. we're only using it for the fake video data10:00
mzanettiSaviq: KWin has a plugin which shows all paint events on the screen. unity8 flickers like crazy. also callgrind shows like 47% cost in ligGL and mesa stuff10:05
mzanettiSaviq: seens something is triggering paints all along10:05
mzanettidednick: ping10:06
dednickmzanetti: pong10:06
mzanettidednick: if I run unity with the fake plugins, the indicators have some colored rectangles rotating all the time10:06
SaviqCimi, can you join https://plus.google.com/hangouts/_/calendar/Y2Fub25pY2FsLmNvbV85ODZnYWZucHJ2cmU5OGRscjgyMm9zYzg2c0Bncm91cC5jYWxlbmRhci5nb29nbGUuY29t.7i0fmrimvi8oactkg0q3s1e44o please10:06
mzanettidednick: mind changing that to non-rotating rectangles?10:07
mzanettidednick: it makes it hard to profile because that stuff constantly calls paints.10:07
dednickmzanetti: can do.10:07
mzanettidednick: and also, seems you should stop painting alltogether when indicators are collapsed. maybe even unload stuff10:08
dednickmzanetti: they unload after awhile10:08
mzanettidednick: define "a while"10:08
=== shiznix_ is now known as shiznix
dednickmzanetti: 20 seconds10:09
dednickmzanetti: https://code.launchpad.net/~nick-dedekind/unity8/indicators.remove-fake-animations/+merge/17803110:14
mzanettidednick: thanks. approved.10:16
mzanettidednick: and indeed my CPU is not hogging any more whenever I do ./run -p10:16
dednickmzanetti: :)10:16
mzanettidednick: but I have some bad news10:17
mzanettidednick: it didn't stop after 20 mins. so the unloading doesn't seem to work10:17
sil2100andyrock: hi!10:17
mzanettierr, 20 secs10:17
dednickmzanetti: ah. right. i think the fake ones dont actually10:18
sil2100andyrock: I'm currently *trying* to dogfood the new unity with compiz 0.9.10, but I think I'm having problems10:18
dednickmzanetti: they don't really "unload". they "stop" which is intended to stop the dbus shizzle.10:19
mzanettidednick: you sure that works? :P10:19
dednickmzanetti: er. last time i checked it did10:19
dednickmzanetti: and as far as i remember there was a test.10:20
mzanettidednick: because I see this all the time: https://bugs.launchpad.net/touch-preview-images/+bug/118306510:20
ubot5Launchpad bug 1183065 in touch-preview-images "Occassional severe battery drain" [High,Confirmed]10:20
mzanettidednick: and reading through the comments it seems to be related to network10:20
mzanettidednick: so unity8 going crazy because of networking can only be related to the network indicator, I'd think10:20
dednickmzanetti: i'll have to look into it10:22
dednickmzanetti: may have something to do with the network indicator continously crashing...10:22
mzanettidednick: hmm... that sounds indeed interesting10:23
mzanettidednick: this bug is the only really bad issue I'm suffering when dogfooding the device. Whenever I leave home my pocket starts heating up and battery is empty in no time10:24
dednickmzanetti: ever notice if the network indicators content is flashing?10:24
mzanettidednick: hm... I think so, yes. But I can't tell if there's a directo connection to the CPU hogging10:25
dednickmzanetti: well if the indicator dbus stuff isnt stopping, then the model wiill most likely be reloading itself continuously. i'll have to check if the stop is working.10:26
mzanettidednick: but unless networkmanager or the indicators backend have some issues, that shouldn't cause 80% cpu usage, no?10:27
sil2100andyrock: ok, scratch that10:27
dednickmzanetti: in which process? unity8?10:28
mzanettidednick: but actually... the fact that strace shows unity8 trying to access a file descriptor which isn't available any more, kind indicates a crash of something unity relies on10:28
mzanettidednick: yeah, unity8 is the one hogging the cpu10:29
mzanettiSaviq: now that we fixed the fake plugins, I can't reproduce it any more :/10:30
* mzanetti keeps on trying10:30
dednick:/ i can't really say where the process cycles are going. other than maybe redrawing the indicator.10:31
dednickwhich shouldnt be done if it's not showing10:32
dednickmzanetti: can you catch redraws of items in qml?10:32
mzanettidednick: yes10:33
mzanettidednick: well, seems we always paint the full qmlscene10:33
dednickmzanetti: how? :)10:34
mzanettidednick: so not single items, but I can see everything flicker if something repaints10:34
mzanettidednick: KDE's KWin has a "desktop effect" called "Show paint". That colorizes every painted frame differently10:34
jameshsil2100: hi.  I was looking at your lucene++ packages, and ran into a problem compiling the media scanner with it10:34
Saviqmzanetti, but can you confirm that's not the case for simple qmlscene apps?10:34
Saviqmzanetti, I don't think we have damage support in QML10:35
sil2100jamesh: hi! What was the problem?10:35
jameshsil2100: it looks like it has been compiled with a custom allocator, which trips up the configure checks in the mediascanner10:35
mzanettiSaviq: what, the CPU hogging?10:35
sil2100jamesh: we're configuring it with -DLPP_USE_ALLOCATOR:BOOL=OFF, so hm10:35
MacSlowmpt, ping10:36
dednickmzanetti: yeah, but you cant see that while your phone is off ;)10:36
dednickmzanetti: i meant is there a signal you can connect to for redraw requests on items.10:36
jameshsil2100: changing "-DLPP_USE_ALLOCATOR:BOOL=OFF" to "-DENABLE_STANDARD_ALLOCATOR:BOOL=ON" in debian/rules should do the trick10:36
mzanettidednick: ah... hmm... I guess that would require you to subclass QQuickItem or some part of the Engine10:37
jameshsil2100: I think the lucene++ cmake flags have changed since the packaging you based your work on was written10:37
sil2100jamesh: ok, might be - but I actually see Ken change it from -DENABLE_STANDARD_ALLOCATOR:BOOL=ON to the current one, so maybe in the past he had a need for a custom allocator?10:37
* mzanetti needs to turn off the show paint effect before he gets an epileptic stroke10:37
sil2100jamesh: I'm prepping a new version now anyway so I'll change that10:38
jameshsil2100: "grep LPP_USE_ALLOCATOR /usr/include/lucene++/Config.h" should show whether it is set correctly10:38
jameshsil2100: we need it to be an #undef, while the current test package has it #define'd10:38
jameshsil2100: I'm not sure if the media scanner was ever built against Ken's updated package: its packaging still refers to the "luceneplusplus" package rather than "lucene++"10:44
sil2100jamesh: probably!10:45
sil2100jamesh: anyway, it will be fixed in the next version10:45
jameshsil2100: awesome.  Thanks for your help with this.10:46
larsudednick: hey, what's up?10:47
dednicklarsu: hey. sorry, figured out my issue10:48
dednicklarsu: wasnt getting dataChages for submodels. but i was looking on the parent.10:49
larsudednick: great :)10:49
=== MacSlow is now known as MacSlow|lunch
Saviqmzanetti, no, the fact that it redraws the whole thing10:53
Saviqmzanetti, if it redraws constantly - that's a different issue10:53
Saviqmzanetti, but I don't think we can be redrawing just parts of our UI10:53
mzanettiSaviq: I agree... as its one single opengls scene10:53
Saviqmzanetti, yup10:53
mzanettiSaviq: anyways... the issue I've seen was the fake indicators10:54
Saviqmzanetti, ah, so not applicable to device10:54
mzanettiSaviq: now that this is fixed, I'm having a hard time to reproduce Kaleo's bug10:54
mzanettiSaviq: no... been swiping the dashes now for 5 minutes... nothing happens in regards of CPU usage10:54
Saviqmzanetti, yeah, it's random unfortunately10:54
mzanettiSaviq: I can however reliably reproduce the other one: https://bugs.launchpad.net/touch-preview-images/+bug/118306510:55
ubot5Launchpad bug 1183065 in touch-preview-images "Occassional severe battery drain" [High,Confirmed]10:55
Saviqmzanetti, as I wrote in the bug - I couldn't reproduce either10:55
mzanettiSaviq: so its maybe this one that Kaleo has seen in combination with the mem leaks10:55
Saviqmzanetti, yeah, but the battery drain is usually caused by CPU hogging, so...10:55
mzanettiSaviq: yeah... I did an strace for that one...10:56
mzanettiSaviq: its unity8 stuck in an endless loop trying to read some fd which is "temporarily not available"10:56
mzanettiSaviq: I'm tempted to reopen that one for unity8 as even if some backend crashes away, we shouldn't hog CPU because of that10:57
=== hikiko is now known as hikiko|lunch
mzanettitsdgeos: hey, do you know of a way to get to animations in QML tests?11:38
mzanettithey don't seem to be proper childs and are not found by findChild()11:38
tsdgeoshaven't tried11:39
tsdgeosmzanetti: so you can't find them by ojbectName?11:39
mzanettitsdgeos: not right now, no11:40
tsdgeosi mean a QQuickAnimation is QObject11:43
tsdgeosdon't see why it should not get parented as usual11:43
mzanettihmm... I don't know why either11:44
mzanettimaybe thats the mem leak dandrader tries to find :D11:44
=== alan_g is now known as alan_g|lunch
mzanettitsdgeos: ok... they are parented... but not where they are written in the QML code11:51
mzanettitsdgeos: In this case, the anim has no parent: http://paste.kde.org/pd5712ba011:52
mzanettitsdgeos: in the Lockscreen, the parent is the next wrapping Loader {}11:52
mzanettiyeah... totally11:52
tsdgeoslunch time11:52
=== MacSlow|lunch is now known as MacSlow
mzanettiok... it gets even weirder: myAnim.parent == myLoader, BUT myLoader.children does NOT contain myAnim12:02
sil2100andyrock: running the new compiz, so far so good12:05
sil2100andyrock: do you know if there have been any performance improvements?12:05
=== hikiko|lunch is now known as hikiko
mzanettiCimi: MacSlow: you guys had some tests where you would have needed to access animations in findChild, right?12:28
MacSlowmzanetti, correct... that's in the qmltest for the  notification-frontend (lp:unity8/unity8/tests/qmltests/Notifications/tst_Notifications.qml)12:29
mzanettiMacSlow: Cimi: I found the cause for that not working and fixed our findChild(). will land in trunk soon12:31
mzanettitsdgeos: btw ^12:31
MacSlowmzanetti, well in the notification-qmltest is works atm... what issue did you fix?12:31
=== _salem is now known as salem_
mzanettiMacSlow: findChild() now also finds animations (and other non-visible items)12:32
MacSlowmzanetti, ah... good to know... although I can't jump in an fix/change the test right now (once your stuff is merged)12:33
mzanettiMacSlow: iirc you added some workaround like some __properties that refer to animation.running, right?12:33
MacSlowmzanetti, waitForRendering(...) is used atm12:34
mzanettiMacSlow: ???12:34
mzanettithats not related in any way12:35
MacSlowmzanetti, then I don't know12:36
mzanettihah! could reproduce the testRightEdgeDrag hanging as we've seen it on jenkins lately12:38
Saviqmzanetti, tsdgeos can you or someone https://code.launchpad.net/~saviq/unity8/add-fake-disclaimer/+merge/178059 please12:41
mzanettiSaviq: ok. can do12:41
Saviqmzanetti, under autopilot or real?12:41
mzanettiSaviq: qmltestrunner12:42
Saviqmzanetti, ah12:42
Saviqmzanetti, sounds like a candidate for a test if you can reproduce12:43
kgunnSaviq:would you mind documenting your findings in the mp after you test galaxy nexus w/ duflu's switch branch12:47
Saviqkgunn, will do12:47
Saviqkgunn, unfortunately gn is *slow* for building...12:48
=== alan_g|lunch is now known as alan_g
Saviqkgunn, your leg is twitching12:51
Cimimzanetti, nope12:56
mzanettiCimi: I know you did...12:56
mzanettibut anyways... just wanted to let you know that as of now it works12:56
kgunnSaviq: :)12:59
kgunnonly way to stay awake12:59
tsdgeosSaviq: do we want tr() for that text?12:59
Saviqtsdgeos, nah12:59
mzanettiSaviq: tsdgeos: any idea on that one? http://paste.kde.org/p6ab7029213:01
mzanettithis is the result of a findChild(shell, "launcherPanel")13:01
mzanettisomewhere in responsiveGridViewGridView we have some parent-child loop13:03
tsdgeosyou sure?13:03
mzanettitsdgeos: the test hangs forever here... admittedly after my change in findChild which does not only search visible children but also invisible ones (e.g. animations)13:04
tsdgeosyeah 0x282bbd0 seems to be there too often13:04
mzanettitsdgeos: to reproduce, edit UnityTestCase and change "obj.children" to "obj.data" which also allows finding animations13:05
mzanettitsdgeos: I could now just create a second one, findInvisibleChild or the like and get around this issue. but it feels fishy enough to investigate13:05
tedgdednick, So it seems that the switch items require a target type to be set?13:07
tedgdednick, I'm a bit confused on why that is.  It seems like they should just be using the state of the action.13:07
dednicktedg: you mean based on the action state being of type boolean?13:09
tedgdednick, Yes, I think that is how it should be13:10
dednicktedg: hm. interesting point.13:10
dednicktedg: at the moment we dont look into the type until we know what it is.13:10
tedgdednick, That is, for instance, how check items work in GTK.13:10
dednicktedg: what if we have 2 items with the same format state?13:10
tedgdednick, What do you mean by format state?13:11
dednickeg an float for progeress/slider.13:11
tedgYes, it should be.  That should be the state of the action.13:11
dednicktedg: the gvariant format of the action state13:11
tedgTargets are really only used for radio groups.13:11
dednicktedg: what i mean is. how do we know to render a slider, or a progress bar?13:12
tedgdednick, Won't that be the x-canonical-type ?13:12
dednicktedg: what is the target type?13:12
dednicktedg: that's what i though you ment of 'type'13:13
dednicktedg: mumble?13:13
tedgdednick, Sure13:13
Saviqtsdgeos, mzanetti any word on the disclaimer? we would like to get it in so we can get into distro asap13:20
mzanettiSaviq: didn't test yet... can do now if urgent13:20
Saviqmzanetti, yes please13:21
mzanettiSaviq: this sucks13:23
Saviqmzanetti, that we get in on desktop all the time?13:23
Saviqmzanetti, think we should have an env var that would disable it?13:23
Saviqlike I_KNOW_ITS_ALPHA=1?13:24
mzanettiSaviq: probably add something into the ~/.unity8-greeter-demo even if the filename is probably not the right one, but its already there and easy to adjust13:24
mzanettiSaviq: and how does this go with planet ubuntu tellig people they should start dogfooding it?13:25
Saviqmzanetti, on phone13:25
Saviqmzanetti, it's not going to be there on phone13:25
Saviqmzanetti, only with the fake Ubuntu.Application13:25
mzanettiah... oh. that's different then13:25
mzanettilemme try on the phone13:26
Saviqmzanetti, it will be there on desktop and during autopilot runs, that's it13:26
mzanettiSaviq: ah ok... I'm fine with it then...13:26
mzanettiSaviq: just wondering. what's the resoning for this?13:27
Saviqmzanetti, so that when we land in distro and people see it13:27
Saviqmzanetti, it's obvious that it is so13:27
Saviqmzanetti, i.e. `apt-get install unity8; unity8` AAAH that's crap! let's blog about it13:27
mzanettiSaviq: right. makes sense13:28
mzanettiSaviq: still compiling for the phone, but I would expect it to throw a warning "no such property applicationManager.fake" if its the real one13:30
dednicktedg: having an issue with the current network indicator. it's not giving me a action state for the root menu item.13:30
mzanettiSaviq: yep: file:///home/phablet/shell/Components/ApplicationManagerWrapper.qml:32: Unable to assign [undefined] to bool13:31
Saviqmzanetti, hmm let me "cast" ;)13:31
dednicktedg: never mind. i think it's my fault.13:34
mzanettiCimi: standup?13:35
Saviqmzanetti, pushed13:39
mzanettiSaviq: happroved13:42
Saviqmzanetti, o/13:50
=== salem_ is now known as _salem
=== _salem is now known as salem_
andyrocksil2100, no idea about that14:05
andyrockthere are some branches that should improve the performance14:05
Saviqmterry, ping14:14
mterrySaviq, hello14:14
Saviqmterry, hey, would you have time to work on a quick'n'dirty solution for the powerd SysPowerStateChange signal?14:15
Saviqmterry, we identified that this would significantly improve battery life14:15
Saviqmterry, if we unfocus all the apps when locking, 'cause we'd disable accelerometer then14:15
Saviqmterry, I talked to sforshee that there is (some) plan for a common library to talk to powerd, but it's not on the roadmap14:16
Saviqmterry, so we need to hack something up :/14:17
Saviqdoesn't look like you will ;P14:19
mterrySaviq, my IRC connection seems awful14:20
Saviqmterry, I noticed14:20
Saviq<Saviq> mterry, hey, would you have time to work on a quick'n'dirty solution for the powerd SysPowerStateChange signal?14:20
Saviq mterry, we identified that this would significantly improve battery life14:20
Saviq mterry, if we unfocus all the apps when locking, 'cause we'd disable accelerometer then14:20
Saviq mterry, I talked to sforshee that there is (some) plan for a common library to talk to powerd, but it's not on the roadmap14:20
Saviq mterry, so we need to hack something up :/14:20
pete-woodsdoes anyone know what the plan is with regards to wiping the root partition on device flash? at the moment this is killing everything in /var/lib? is there somewhere more permanent that I should be putting things?14:20
Saviq<-- mterry has quit (Ping timeout: 248 seconds)14:20
Saviq<Saviq> doesn't look like you will ;P14:20
mterryMy internet doesn't appear to be this bad...14:20
Saviqpete-woods, with the system-image flashing, it won't anymroe14:20
Saviqpete-woods, or at least you won't need to flash14:21
pete-woodsSaviq: is there somewhere I can read about that? so as not to ask you a serious of dumb questions14:21
Saviqpete-woods, https://wiki.ubuntu.com/ImageBasedUpgrades/14:21
dednicktedg: ping14:22
Saviqpete-woods, lool would be your PoC for that while stgraber is away14:22
tedgpete-woods, You should talk to stgrabber there14:22
tedgdednick, howdy14:22
pete-woodsSaviq: basically the question is really, should I be stuffing data in another location than /var/lib, but it sounds like someone else has fixed the issue for me (which is always good)14:23
Saviqpete-woods, yeah, I think /var/lib is right14:23
dednicktedg: hi. ok, so i think i've found the code in unitymenumodel (at least) which is things up. it looks like the "action type" and "action state type" have to match to be deemed "activatable".14:23
pete-woodsSaviq: awesome, thanks!14:24
dednick*which is screwing things up14:24
dednicktedg: at least it's screwing it up for me because the network indicator root item has a action type, but no action state.14:24
tedgdednick, Hmm, is that a Unity menu model thing or a handling thing?14:25
Saviqmterry, want me to email?14:26
dednicktedg: its a gmenumodel handling thing. inbetween backend and unitymenumodel14:26
dednicktedg: so the item doesnt have a target type, but has a state type.14:27
=== alan_g is now known as alan_g|tea
dednickwhich unitymenumodel feels is bad.14:27
dednicklarsu: ^ when you're about.14:27
didrocksSaviq: http://bazaar.launchpad.net/~cupstream2distro-maintainers/cupstream2distro-config/trunk/revision/577 and (running the stack)14:28
Saviqtsdgeos, https://launchpadlibrarian.net/146390382/buildlog_ubuntu-saucy-amd64.unity-mir_0.1-0%2B201308011415~24~saucy1_FAILEDTOBUILD.txt.gz ;(14:28
didrockssil2100: FYI ^14:28
Saviqdidrocks, awesome :)14:28
dednicktedg: was your problem that you needed to set a target type? or a state type?14:28
dednicktedg: i'm confusing myself again i think14:29
Saviqkgunn, olli_, asac, unity8 is getting released into distro right now14:29
tedgdednick, I had to set the target type14:29
olli_Saviq, coolio14:29
didrocksnot "now", NOW! :)14:29
olli_Saviq, which string did you go with?14:29
mterrySaviq, heh, let's try one more time14:30
Saviqmterry, you've got @14:30
Saviqolli_, JohnLea's recommendation14:31
olli_Saviq, neato14:31
kgunnolli_: Saviq ...i still prefer "hold my beer...watch this"14:32
mterrySaviq, I see14:32
mterrySaviq, does SysPowerStateChange need implementing on Ubuntu or do we need unity8-side listening for it?  (or both)14:32
olli_kgunn, :)14:33
olli_kgunn, https://www.facebook.com/HoldMyBeerMemes?ref=stream14:33
Saviqmterry, we just need unity8 to listen for it14:36
Saviqmterry, it's there in powerd14:36
PedroGomesHey guys. When are you going to introduce a decent multiple screen support? I'm tired of this bugs that come since 12.0414:36
SaviqPedroGomes, did you file reports for those bugs?14:38
PedroGomesSaviq: AFAIK, one of them is widely known, the one where you take one of the screens and you windows go to random places.14:38
PedroGomesSaviq: I know that is better now, but even so there is no logic in the after location14:39
=== alan_g|tea is now known as alan_g
SaviqPedroGomes, if there's a bug filed, there's a better chance of getting it fixed14:40
SaviqPedroGomes, "widely known" is not really useful14:40
PedroGomesSaviq: widely known as the one of the guys here some months back that this was know14:42
PedroGomesSaviq: s/that this/ said that this/14:43
SaviqPedroGomes, did you look for the bug report?14:43
seb128mterry, hey14:43
Saviqseb128, mterry has huge IRC issues today14:44
seb128Saviq, oh, I saw that you tried to talk to you as well14:44
seb128Saviq, maybe he just doesn't like you and pretended having IRC issue to avoid having to reply :p14:44
seb128mterry, talk to me! :p14:44
Saviqseb128, yeah, it's almost a week here already, people start talking to themselves14:44
seb128Saviq, thanks ;-)14:44
Saviqseb128, nothing to point fingers about!14:45
PedroGomesSaviq: https://bugs.launchpad.net/compiz/+bug/76314814:45
ubot5Launchpad bug 763148 in Compiz Core "Adding/Removing an external monitor causes open windows to move to another workspace" [Medium,Fix committed]14:46
tsdgeosSaviq: what?14:46
PedroGomesSaviq: do you care to translate this to me? I see fix released/commited14:46
tsdgeosSaviq: mir broke unity-mir?14:46
Saviqtsdgeos, yeah14:47
Saviqtsdgeos, looks like http://bazaar.launchpad.net/~mir-team/mir/trunk/revision/90614:47
tsdgeosSaviq: ok, want me to have a look?14:47
SaviqPedroGomes, this bug is fixed, looks like it got reintroduced, or there's actually a different bug14:47
bregmaPedroGomes, what version of Ubuntu are you running?14:48
SaviqPedroGomes, if there isn't a newer / open one, file one with `apport-bug` ideally14:48
Saviqdidrocks, ugh, we need unity-api to get into distro, too14:49
Saviqdidrocks, will it get released with the whole stack or?14:49
didrocksSaviq: wholeeeeeee stack of course! :)14:49
didrocksunity8, unity-notifications and unity-api14:49
Saviqdidrocks, ok good, didn't know whether the target thing is per-stack or per-source14:50
didrocksSaviq: I have still enough coffee in my brain to have thought about it :)14:50
Saviqdidrocks, thanks ;)14:50
Saviqtsdgeos, what'd you have to interrupt for looking into it?14:54
mterryseb128, hi14:54
mterryseb128, that one was mostly just me grabbing a drink while autopilot ran :)14:54
olli_Saviq, when will I be able to apt-get it14:55
tsdgeosSaviq: well i'm doing some work on the LVWPH to support the expanding shell categories14:55
olli_minutes, hours, tomorrow14:55
Saviqolli_, hours14:55
tsdgeosSaviq: but almost done14:55
seb128mterry, hey14:56
olli_Saviq, coolio14:56
seb128mterry, you are the greeter man right? ;-)14:56
Saviqolli_, not even14:56
olli_k, thx14:56
seb128mterry, greeter/lock screen14:56
seb128mterry, I'm trying to figure out if the options listed on that panel are supported, and if they are, through what config/interface:14:57
seb128mterry, e.G the "when locked allow: launcher/camera/notepad", "stats on welcome screen", "messages on welcome screen"14:57
mterryseb128, Yup, greeter guy15:06
seb128mterry, can you reply to the questions then? ;-)15:13
Saviqolli_, indicators-client-autopilot is obsolete - it's built into unity8-autopilot15:20
Saviqtsdgeos, lp:platform-api/mir seems to be broken too15:22
Saviqtsdgeos, with the events refactor in mir15:22
tsdgeosSaviq: ok, i'll start updating the phone for unity-mir development again15:26
tsdgeosand have a look tomorrow morning15:26
tsdgeossince it's not going to finish that fast :D15:26
Saviqtsdgeos, thanks15:27
tsdgeosSaviq: so the LVWPH sectionHeaders now know to which index they belong with https://code.launchpad.net/~aacid/unity8/lvwph_sectionHeaderIndex/+merge/17809915:27
tsdgeosso i can map sectionHeader -> index without having to depend on the section name to be unique15:27
Saviqtsdgeos, cool15:28
tsdgeosthe best thing is that it's not much code15:28
larsudednick: there's no such thing as an "action type"15:37
larsudednick: an action can have state (which is typed), and when you activate an action, you can pass a parameter (which is also typed)15:38
larsudednick: the two do *not* need to match. In fact, they often don't15:38
larsudednick: the problem you're hitting is probably that the menu item's "target" (which is passed as parameter when activating the action) does not match the actions parameter type15:39
=== jhodapp is now known as jhodapp|lunch
larsudednick: unitymenumodel doesn't know what to do in that case, so it reports the menu item as not activatable15:39
Saviqolli_, you can install from ppa:ubuntu-unity/daily-build already15:50
Saviqolli_, it's building for armhf still15:50
dednicktedg: ^ so you needed to set the menu items target type to match the action parameter type?15:56
tedglarsu, But we're talking about for switch items.  Which seems they shouldn't have a target at all.  Just a state of type b.15:57
tedgI think that the switch item is trying to use the target type for the state.15:57
dednicktedg: but then shouldnt the action parameter also be empty? and you set the state?15:57
tedg(perhaps others, but that one in particular)15:57
tedgdednick, Correct, and if I set the target type to null (unset) the UI doesn't seem to lookup the state.15:58
tedgIf I set the target type to 'b' then it's happy.15:58
larsutedg: right, you have two options:15:58
larsu(a) set parameter_type to NULL on the action15:59
dednicktedg: probably because the x-canonical-type isnt set15:59
larsu(b) set the right target15:59
tedglarsu, What do check boxes do, I think that the switches should match check items.15:59
larsufor a switch menu item, I'd not set a target15:59
larsucause activating that item should simply toggle the switch15:59
larsuwhich in turn toggles the action state15:59
larsutedg: check boxes work the way I just described (no parameter, activate toggles)16:00
tedglarsu, dednick, Also (tangent) I added two proposed props to the root item.  Thoughts?  https://wiki.ubuntu.com/SystemComponents16:00
tedglarsu, I think that the problem dednick is having is that something is checking that target type, and if it doesn't match the state, is getting confused.16:00
tedgDo we have any item types that need a target?16:01
* tedg thinks no16:01
tedgThis might make the whole thing clearer if we just said code that used it was wrong (for now) :-)16:01
dednicktedg: i think the way unity8 currently gets around the switch is with setting the target=NULL, and it works because the menu is typed with "com.canonical.indicator.switch" which assumes a state.16:01
larsutedg: I like the "icons" property, but it should be av, not ay (g_icon_serialize may return any kind of variant)16:01
tedglarsu, Ah, okay, I wasn't sure what the type was there.16:02
larsutedg: I don't understand the pre-label one, isn't that the same semantics as a normal label?16:02
tedglarsu, So then icon should be 'v' as well?16:02
tedglarsu, No the normal label goes after the icon.16:02
larsutedg: yes16:02
dednicktedg: i think the chewie network needed a target16:02
larsutedg: ah! I misread it's for the "other" side. Sounds good to me16:02
larsutedg: thanks for documenting it16:02
tedgdednick, Hmm, I wonder why.  We don't anymore :-)16:03
tedgdednick, Should we have a new type?16:03
dednicktedg: my problem - ACTION ADDED :indicator.network-status  , action_target:(nil), parameter_type:0x269b37016:03
larsusounds like dednick has the right idea :)  /me needs to go, bl16:04
tedgdednick, That sounds right, why is that a problem?16:05
dednickbecause it doesnt fetch an action state for that condition.16:06
dednicktedg: why is the root item action parameter type set? we're never going to activate it are we?16:06
dednicktedg: shouldnt it just have a action state for the connection state info?16:07
dednicktedg: re root item proposal - do we need labels on both sides?16:08
tedgdednick, It should still get the state even if the target is null.  They're independent.16:09
tedgdednick, The type is just so we can extend it later, but sure, on the root item the type isn't that important.16:09
tedgdednick, Yes, we need the label on one side for power and the other for network.16:09
dednicktedg: target being null isnt the issue, it's that the target does not match the actions parameter type16:09
tedgdednick, And that's fine, as they're independent.  The state and the target shouldn't effect each other.16:10
dednicktedg: but i guess maybe it should get it anyway16:10
dednicklarsu: ^ should set state even if not activatable?16:12
dednicktedg: but anyway, if we extend, we can change both? i dont think it makes sense that they dont match.16:15
dednicktedg: since you cant activate if they're not anyway.16:15
tedgdednick, What do you mean by both?16:15
dednicktarget and action parameter type16:16
tedgYes, they could not match.16:17
Saviq+            print("fpppppppppppppppppppppppp1")16:20
Saviq             checkRightEdgeDragWithNoRunningApps();16:20
Saviq 16:20
Saviq+            print("fpppppppppppppppppppppppp2")16:20
Saviq             dragLauncherIntoView();16:20
Saviq 16:20
Saviq+            print("fpppppppppppppppppppppppp3")16:20
Saviq ;)16:20
Saviq+            print("baaaaaaaaaaaar1")16:21
Saviq+            print("trying to find child \"launcher\" from shell")16:21
Saviq             var launcherPanel = findChild(shell, "launcherPanel");16:21
Saviq+            print("baaaaaaaaaaaar2")16:21
Saviq             verify(launcherPanel.x = - launcherPanel.width);16:21
Saviq+            print("baaaaaaaaaaaar3")16:21
Saviq             swipeFromLeftEdge();16:21
Saviq+            print("baaaaaaaaaaaar4")16:21
Saviq             tryCompare(launcherPanel, "x", 0);16:21
Saviq+            print("baaaaaaaaaaaar5")16:21
mzanetti_Saviq: already removed16:21
Saviqmzanetti_, ;)16:21
mzanetti_the art of debugging16:21
mzanetti_Saviq: I think this commit fixes the few "Job killed after 60 mins" we've seen lately16:22
mzanetti_Saviq: it speads up searching Shell.qml to no-time while searching the whole tree took multiple seconds here and with the invisible childs enabled it wouldn't finish any more16:22
Saviqmzanetti_, cool16:32
mzanetti_Saviq: when you have a minute, please respond on my comment here: https://code.launchpad.net/~macslow/unity8/notification-autopilot-tests-dbus/+merge/17778016:57
=== alan_g is now known as alan_g|EOD
dednicktedg: ping17:30
tedgHowdy dednick17:32
dednicktedg: hi again :) just a quick question. the root item uses (sssb) type. will that be updated to a{sv} ? or it is to stay?17:33
dednicktedg: talking network indicator17:33
tedgdednick, I think many of the indicators have already gone to a{sv}, network will join.17:33
tedgNot sure how many are still using sssb17:34
tedgI'd consider that deprecated, but not forgotten.17:34
dednicktedg: cool. thanks. i think only session and network remain.17:34
tedgHmm, charles, do you have session open?  ^17:34
tedgWe could perhaps exterminate.17:34
dednicktedg: yeah, i have support for it. but the icons are strings instead of serialised data, so need to update17:34
charlestedg, session is still using sssb17:35
charlesimo all of them should be using a{sv}17:35
dednicktfb i'll leave in the support.17:35
tedgcharles, I remembered you had a couple session things on your TODO.  When you have it open, do you want to move it over?17:35
tedgThat should be a small MR :-)17:36
charlestedg, yes, I'll exterminate it17:36
charlestedg: you think you'll get a small MR out of me?! ha ha hahaha17:36
tedgNo reformatting or dropping returns!17:36
* charles looks for a way to bloat the header action type MR to 100s of lines17:36
charlestedg: I'll add "return;" everywhere17:37
charlesat the ends of functions17:37
charlesat the beginnings of functions17:37
charlestedg: if I add the to the beginning of both the code /and/ the tests, there won't be any regressions either17:38
tedgI need to get a better way to show exits from functions in VIM.  I like the return because there's a big yellow marker there.17:38
tedgPerhaps some marker that is universal to no matter how you exit.17:38
tedgWow, need a gtk_box.  I haven't done GTK in so long.17:39
tedgWait, grid.17:40
=== naee is now known as eeanm
=== jhodapp|lunch is now known as jhodapp
bschaefersil2100, ping18:47
sil2100bschaefer: ping!21:07
sil2100bschaefer: hi!21:07
sil2100bschaefer: sorry it took so long, but I have a massacre with the new flat21:07
sil2100bschaefer: any luck with the new ibus?21:09
bschaefersil2100, yeah, I was hoping to get someone to restart the daily build stuff to test it!21:12
bschaeferand now worries!21:12
bschaefersil2100, as I've pushed in a change to remove the gconf stuff we have in the ap tests, which should get around that error, but I wasn't able to reproduce on my machine21:13
bschaefersil2100, well restart the AP tests for the daily build21:14
sil2100bschaefer: stack restarted21:24
bschaefersil2100, awesome thanks!21:24
* bschaefer hates not being able to reproduce errors on his machine21:24
=== salem_ is now known as _salem

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