[07:47] Saviq: i can't change the dash while it's rebounding :S [07:54] o/ [07:54] \o [07:54] Saviq: any high priority issue to tackle or should I continue with fixing the preview stuff? [08:03] can any of you guys try to repro https://bugs.launchpad.net/unity8/+bug/1238094 ? [08:03] Ubuntu bug 1238094 in unity8 (Ubuntu) "Header in LVWPH gets confused when rebounding" [Undecided,New] [08:03] i don't seem to be able to sweep horizontally while it's rebounding [08:06] oh [08:06] tsdgeos: I managed something [08:06] i can if the scope doesn't fill all the screen [08:06] interesting [08:06] not anymor [08:07] and now i can again [08:09] the question is, do we always want to simply disable h-swiping while rebounding or we always want to enable it? [08:11] we are not caching thumbnails ? if I expand the 'more suggestions' category everytime there are blank icons and they load after a few seconds. that looks ugly [08:12] the same happens in the Music scope or video [08:12] s/or/and [08:16] tsdgeos, any ideas about https://bugs.launchpad.net/unity8/+bug/1238302 ? [08:16] Ubuntu bug 1238302 in Unity 8 "Category reordering causing column "shifting" in overridden result model" [Undecided,New] [08:16] hmmm [08:16] nope, but it's interesting [08:17] * tsdgeos has a look while someone who can decide what we want to do with #1238094 comes up (i.e. Saviq) :D [08:18] mhr3: how do you reproduce exactly? can't seem to unerstand the bug text :-S [08:19] tsdgeos, yea.. my bug reporting skills are low at midnight :/ [08:19] let me try again [08:22] tsdgeos, perhaps now [08:22] bug #1238094 [08:22] bug 1238094 in unity8 (Ubuntu) "Header in LVWPH gets confused when rebounding" [Undecided,New] https://launchpad.net/bugs/1238094 [08:23] tsdgeos, yeah, me neither [08:23] mhr3: should the application category even be there? [08:24] tsdgeos, well it shouldn't be overridden, but we're probably not going to fix that for 13.10 [08:24] Saviq: thing is, you can only swipe horizontally if you drag from an "empty space", i.e. if there's a category you can not drag horizontally to change the scope while it's vertically rebounding [08:24] tsdgeos, ah [08:25] Saviq: so basically i'd say "disable horizontal scrolling enterely" when we are rebounding [08:25] saves us a lot of pain [08:25] tsdgeos, huuh, btw, do you have like 6 horizontal dividers below "Apps" in home scope? [08:25] tsdgeos, +1 - when it's moving - no moving to the sides [08:25] Saviq: nope [08:26] Saviq: but i saw that regularly before we fixed the bug in qsortproxymodel [08:26] * Saviq flashes [08:26] Saviq: maybe there's still something that needs more kicking in there? or your qt is old? [08:26] tsdgeos, flashed yesterday... [08:26] ok, then it's defenitely not your qt [08:26] tsdgeos, either way - I can't see any glitches when I *do* manage to swipe the dash while it's moving [08:27] Saviq: sometimes the header gets in the middle [08:27] it's maybe even easier if you do in applications scope [08:27] scroll it up, then quickly left from the botto [08:27] come back [08:27] the header is in the middle [08:28] tsdgeos, I can see it behaving weirdly while it moves - but never did get it to end up in the middle [08:29] i can repro quite easily here [08:29] but tbh i agree we should just disable h-swipe if it's moving vertically [08:29] tsdgeos, anyway +1 on disabling ←→ while ↓↑ [08:29] since we kind of half do that already [08:29] * tsdgeos gets on it [08:31] veebers, thanks for digging for input [08:33] veebers, dashboard, fortunately, runs just one test at a time ;) [08:43] Saviq, good morning. Anything you want me to test or look into for u8? [08:44] tvoss_, veebers did some digging for https://bugs.launchpad.net/mir/+bug/1238417 [08:44] Ubuntu bug 1238417 in Mir "Unity does not process events from evdev device created before unity is restarted (autopilot tests)" [Critical,Confirmed] [08:45] tvoss_, and what's more - found a workaround [08:45] woot [08:45] tvoss_, I filed some crashers bug #1238287 bug #1238116 bug #1238107 - some of them you saw already [08:46] bug 1238287 in unity8 (Ubuntu) "unity8 crashed with SIGABRT in raise()" [High,Confirmed] https://launchpad.net/bugs/1238287 [08:46] bug 1238116 in unity8 (Ubuntu) "unity8 crashed with SIGSEGV in QIcon::~QIcon()" [Medium,New] https://launchpad.net/bugs/1238116 [08:46] bug 1238107 in maliit-framework (Ubuntu) "maliit-server crashed with SIGSEGV in __GI___pthread_mutex_lock()" [Medium,New] https://launchpad.net/bugs/1238107 [08:46] mzanetti, there? [08:46] Saviq: yes [08:46] mzanetti, re: https://code.launchpad.net/~mzanetti/unity-mir/fix-appid-parsing/+merge/190419 [08:46] mzanetti, we need to make sure everything works as it did [08:47] mzanetti, i.e. launching with --desktop-file-hint=/full/path [08:47] erm _ _ [08:47] Saviq: yes. this is exactly what it fixes [08:47] mzanetti, or --desktop_file_hint=basename.desktop [08:47] oh... is that supposed to work? [08:47] as well as application:///full/path.desktop and application:///appid.desktop [08:47] mhm... ok. need to check again [08:47] will do now [08:48] mzanetti, we did enable it for *some* reason [08:48] mzanetti, obviously won't be able to tell you what the reason was [08:48] mzanetti, but while we have this, let's not break it [08:48] sure [08:48] mzanetti, we also need to make sure it all works on sflinger [08:48] Saviq: this code isn't used with SF at all [08:49] mzanetti, right, of course ;) [08:49] Saviq: and is only used when calling something from cmdline with --desktop_file_hint. but yeah. It probably breaks the two above [08:49] will fix [08:51] mzanetti, truth is, maybe we should adapt sflinger's appmanager to store the app id and not the desktop file path, but I'm not entirely sure we want to open that pandora's box [08:51] Saviq: not sure I understand... we DO use the appId. that's exactly the issue it fixes === iahmad is now known as iahmad|afk [08:52] mzanetti, aah, but we didn't in unity-mir? [08:52] mzanetti, k [08:52] Saviq, asked pitti to help with https://bugs.launchpad.net/mir/+bug/1238417 [08:52] Ubuntu bug 1238417 in Mir "Unity does not process events from evdev device created before unity is restarted (autopilot tests)" [Critical,Confirmed] [08:52] Saviq: current code has an issue that it uses the full .desktop path when calling from cmdline [08:52] Saviq: but uses the appid everywhere else [08:52] mzanetti, right, 'stood [08:52] mzanetti, anyway, I'll stop now :) [08:53] lemme add some better description to the MR [08:54] Saviq, got a better stack trace for https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1238287 [08:54] ? [08:54] Ubuntu bug 1238287 in unity8 (Ubuntu) "unity8 crashed with SIGABRT in raise()" [High,Confirmed] [08:55] tvoss_, I *did*, but f*cking retracing service removed it apparently [08:55] tvoss_, when it decided to mark as dupe [08:55] Saviq, damn it [08:56] tvoss_, but I could easily reproduce it - let me try again [08:56] yup, there it goes [08:56] will be with you in 5 [08:57] Saviq, thx [08:57] Saviq, see #ubuntu-mir [08:58] Saviq, https://launchpadlibrarian.net/153403487/StacktraceSource.txt indicates that maliit is trying to start, not to stop [08:58] which contradicts the bug description [08:59] tvoss_, thing is this happens when unity8 is being stopped sometimes [08:59] tvoss_, maliit spins the CPU [08:59] tvoss_, and prevents unity8 from exiting [08:59] tvoss_, but maybe the trace isn't right [08:59] Saviq, sure, but the stack trace is not the one of a spinning maliit, but the one of a restarting maliit wihtout u8 [09:00] running [09:00] btb [09:00] brb [09:03] Cimi: https://code.launchpad.net/~nick-dedekind/ubuntu-settings-components/qmltest.deps/+merge/190557 [09:04] when you get a minute :) [09:09] Saviq: hmm... just checked with the currently released code. none of those examples you made works right now [09:10] Saviq: and frankly, they seem wrong :D [09:10] dednick, ok [09:14] Saviq, mzanetti so the root cause of https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1238116 [09:14] Ubuntu bug 1238116 in unity8 (Ubuntu) "unity8 crashed with SIGSEGV in QIcon::~QIcon()" [Medium,New] [09:15] seems to be that an icon owns the icon engine it has been loaded from and deletes the engine in its private dtor, which in turn makes the engine delete all its associated icons ... [09:16] Saviq, mzanetti this happens after main, when destructing globals, i.e., the icon cache, which has already been cleared by a post routine [09:16] * mzanetti reads through the bug [09:17] mhm... I see [09:19] Saviq: are we not using DashVideos and DashMusic anymore? [09:19] or desktop != phone in this regar [09:19] d [09:21] tsdgeos: afaik at least DasVideos is obsolete [09:21] mzanetti, I don't understand why QIConPrivate should delete the engine pointer [09:21] tsdgeos: and the whole Dash directory seems to desperately need a cleanup [09:21] mzanetti: +1 :D [09:22] mzanetti, a quick fix would be: make the engine pointer a QSharedPointer in https://qt.gitorious.org/qt/qt/source/e709077eff4d8b05cc9022d85dcb48587d96c720:src/gui/image/qicon.cpp#L109 [09:22] mzanetti, with that, we do not need to worry about deallocation at all, and the last icon will automatically take the engine down [09:23] tvoss_: hmm... not sure... but I don't think that's something new. [09:23] tvoss_: because it's how it works :D [09:23] tsdgeos, no, we're not using them [09:23] tvoss_: it's a QIconEngine not a QIconsEngine, i.e. it's for this particular icon [09:24] tsdgeos, not according to https://launchpadlibrarian.net/153404403/Stacktrace.txt [09:25] mzanetti, wtym "none work"? :D [09:25] Saviq: ok, so i'll propose a merge to kill them it's pretty confusing to still them have around [09:25] tsdgeos, +1 [09:25] Saviq: log in your mir enabled phoen and try to launch some binary with --desktop_file_hint holding something else than an absolute file path [09:25] Saviq: won't work right now [09:26] tvoss_: what makes you say from that backtrace that there are two icons with the same icon engine? [09:26] Saviq: and imho --desktop_file_hint=application:///appId.desktop is plain wrong. same with application:///full/path.desktop [09:26] mzanetti, `webbrowser-app --desktop_file_hint=webbrowser-app` worked for me [09:27] mzanetti, right, of course, /me stupid [09:27] mzanetti, the application:/// things were about using them in activation [09:27] mzanetti, not with desktop hints [09:27] mzanetti, like from the launcher or dash home apps [09:27] Saviq: right... true. the correct appId should work [09:27] Saviq: I can reenable that one as it doesn't seem totally wrong (just a little) [09:29] tsdgeos, hang on, reading through source code [09:29] tvoss_: i agree it could be that, but is not necesarily that [09:30] tsdgeos, looking what the engine is doing [09:35] tsdgeos, yup, circular delete. QIconLoaderEngine has multiple entries (ScalableEntry, which have a QIcon member), so we have a one-to-many-relation, which contradicts every icon exclusively owning the engine [09:35] tvoss_: not really, it's icon has it's own engine [09:36] tvoss_: check the code on how can you end up with different icons having the same engine [09:36] you can't [09:36] there's like 5 places where the engine is set, and it's always "new fooEngine" [09:37] Saviq: "fixed" [09:39] mzanetti, not happy? ;) [09:39] tsdgeos, but https://qt.gitorious.org/qt/qtbase/source/6c06e14a49773ce5572935864ed6b9be219c6103:src/gui/image/qiconloader.cpp [09:39] mzanetti, I'm generally of the opinion that we need to take a step back and stop parsing all that in 5 different places [09:39] tsdgeos, clearly says that the loader engine can have multiple entries ... which makes sense [09:39] Saviq: it feels still wrong to give some else than a desktop file in --desktop_file_hint :) [09:39] mzanetti, *all* apps should be launched via url-dispatcher [09:40] which are the id I can access from one component of the shell? [09:40] I know I can access shell [09:40] Saviq: yes. I agree [09:40] can I access the lock screen as well? [09:40] mzanetti, as I said - I'm not even sure we need it, but we did enable it somewhere - maybe you can find the commit? [09:40] with id lock screen? [09:40] Saviq: well, of course we should still support running them in the command line [09:40] mzanetti, through upstart [09:40] mzanetti, or url-dispatcher [09:41] tvoss_: what line exactly? [09:41] tsdgeos, 332 [09:41] mzanetti, that's going to be the only way to launch apps - we need a single point where that happens - and match PIDs to app ids [09:41] mzanetti, otherwise we'll end up with BAMF again [09:42] Saviq: wasn't there something new in the f.d.o spec that would solve that issue? [09:42] mzanetti, yeah, launching via application:/// [09:42] mzanetti, or something [09:42] mzanetti, either way - on Ubuntu it'll end up going through upstart [09:43] Saviq, why do we still need the desktop file hint? [09:43] tvoss_, legacy [09:43] tvoss_: sure, a QIconLoaderEngine can have multiple m_entries which are QThemeIconEntries, some of them are QIcons and those will have different QIconEngines, that is all fine still, i don't see how it will end up in circular deletion [09:43] tvoss_, just a workaround until we say "now it's over" [09:43] Saviq, ack [09:44] Saviq: well, anyways, my branch now shouldn't break anything existing any more and still fix the --desktop_file_hint thingie (and with it the autopilot tests) [09:44] mzanetti, \o/ [09:53] greyback, o/ === alan_g is now known as alan_g|vt [09:53] tsdgeos, hmmm, just found a ocmment: simply reuse svg icon engine [09:53] greyback, o/ [09:54] tvoss_: where's that? [09:54] tsdgeos, https://qt.gitorious.org/qt/qtbase/source/6c06e14a49773ce5572935864ed6b9be219c6103:src/gui/image/qiconloader.cpp [09:54] line 531 [09:55] Saviq: tvoss_ hi! [09:55] greyback, see, and we don't even want anything from you! [09:55] at least not straight away [09:56] I find that hard to believe [09:56] :) [09:56] tvoss_: the comment looks scary indeed, but then the code doesn't seem like it does anything scary :D [09:57] tvoss_: btw i'm not saying there's no bug, i'm saying i just don't see how it could happen by reading the code [09:57] tsdgeos, there has to be an engine being reused behind the scenes, otherwise, no icon would be loaded [09:57] tsdgeos, yup :) trying to convince you [09:57] tvoss_: why do you say the engine has to be reused? [09:57] tsdgeos, and I have a suspicion that a global instance is reused somewhere [09:58] there is a global cache of QIcons, that is true [09:58] tsdgeos for the icon to load anything it needs an engine instance: https://qt.gitorious.org/qt/qt/source/e709077eff4d8b05cc9022d85dcb48587d96c720:src/gui/image/qicon.cpp#L109 [09:58] yes, that's the global icon cache [09:59] QCache IconCache; [10:00] and yes every QIcon has an engine to load stuff, but still from what i can see, every QIconPrivate has it's own engine [10:05] mzanetti, can you see what you get with your fix and https://code.launchpad.net/~saviq/unity8/workaround-lp1238417/+merge/190574 on the device? [10:06] mzanetti, the only remaining issues it seems would be crashes/hangs [10:07] Saviq: ack [10:08] mzanetti, other than that - fix switching previews please! [10:09] Saviq: ok [10:09] mzanetti, would be awesome to get them n [10:09] in [10:10] Saviq: I agree it would be nice, didn't seem like top priority to me though. but if we're mostly good otherwise, I'm happy to fix them [10:10] Saviq, did you get a change to try out the sim-unlocking? [10:10] MacSlow, yeah +1'd [10:10] MacSlow, and it's merged, AFAIK [10:10] Saviq, :) [10:10] MacSlow, notifications look awful on top of indicators though [10:11] MacSlow, bug #1238174 [10:11] bug 1238174 in Unity 8 "notifications / snap decisions over indicators blend into background" [Undecided,New] https://launchpad.net/bugs/1238174 [10:11] Saviq, *shrugg* [10:11] MacSlow, also bug #1238182 [10:11] bug 1238182 in Unity 8 "Should close indicators when triggered anotification or an application" [Undecided,Incomplete] https://launchpad.net/bugs/1238182 [10:11] MacSlow, but not critical of course [10:12] greyback, can do https://code.launchpad.net/~mzanetti/unity-mir/fix-appid-parsing then? [10:12] Saviq, I'm in review/testing mode still... but if you want me to chase that I can [10:12] greyback, didn't get to it [10:12] MacSlow, no no [10:13] MacSlow, just filed them so that we don't forget [10:13] Saviq: please do [10:13] Saviq, ok... looks certainly like something for Design to look over [10:13] greyback, that was a "can you... do?" ;) [10:13] Saviq: oh, sure, yes I'll look after it then [10:14] Saviq, maybe it'll become less of issue once the much updated UbuntuShape lands in tookit trunk [10:14] Saviq: https://code.launchpad.net/~aacid/unity8/dash_disable_hswipe_on_vswipe/+merge/190576 [10:14] tsdgeos, I'm not referring to the global cache, the qiconloader engine has got multiple entries carrying a QICon, too [10:14] Saviq: https://code.launchpad.net/~aacid/unity8/remove_unusued_dash_videos_music/+merge/190588 [10:15] Saviq, with that outter shadows will be possible... probably providing some more contrast against the background... we'll see [10:16] tvoss_: yes, but for the crash to happen, the QIconPrivate of those QIcon should have as engine the same QIconLoaderEngine that is trying to delete them, and as far as i can see i see nowhere were QIcon*Engine are reused in different QIconPrivate [10:17] MacSlow, yup [10:19] tsdgeos, well, it is already weird that the global dtor is actually finding icons in the cache, as the post cleanup function should already have cleared the cache [10:21] tvoss_: that is also true [10:21] wonder if it's just crashing there because of bad luck [10:21] tsdgeos, it's weird that there is a cleanup function and a global static deleter wrapper thingy [10:21] and it's just that memory is just broken already before reaching there [10:22] tvoss_: maybe one predates the other :D [10:22] s/maybe/probably [10:22] tsdgeos, yup [10:22] actually it's not until qt5 that they have static deleters afaik, copied/inspired by kde having them === _salem is now known as salem_ [10:23] tsdgeos, perhaps removing the cleanup handler would already work [10:24] work as in "fix the crash" or work as in "still do what it's supposed to do"? [10:30] tvoss_: i guess we don't have the core file that caused that backtrace, no? [10:30] tsdgeos, that's a question for saviq [10:30] tsdgeos, just start unity8 and stop unity8 [10:30] tsdgeos, under Mir [10:31] tsdgeos, if it's not attached to the bug, that is [10:31] Saviq: crashes all the time? some? half? [10:31] tsdgeos, all [10:31] tsdgeos, although sometimes differently [10:31] ok [10:31] tsdgeos, but if you just stop unity8; unity8; ^C [10:31] tsdgeos, you should get it [10:31] ok === hikiko is now known as hikiko|lunch [10:34] Saviq, I thought you said https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1238287 had a better stacktrace now? [10:34] Ubuntu bug 1238287 in unity8 (Ubuntu) "unity8 crashed with SIGABRT in raise()" [High,Confirmed] [10:35] tvoss_, didn't retrace it yet [10:35] Saviq, ah [10:35] tvoss_, or, the retrace failed [10:35] tvoss_, but easy to repro, so I'll get it [10:35] Saviq, great, thx [10:46] MacSlow, suggestion. When the password dialog for wifi network appears. the focus should be on the password field so that OSK automatically appears :) [10:47] om26er, true [10:55] MacSlow, om26er might be kinda tricky when there's more input fields [10:55] MacSlow, om26er, maybe the indicator should mark one of the fields to be focused [10:55] Saviq, in that case focus the first input box [10:55] om26er, yeah, we don't know how many there are ;) [10:55] om26er, or the order they come in [10:55] Saviq, don't be the devil's advocate... more than two text-entries?! :) [10:56] MacSlow, two is enough to not know :) [10:56] IMO indicator should mark one that's supposed to be focused on creation [10:57] ah [10:57] no ah [10:57] tvoss_, uploade a retraced .crash file to bug #1238287 [10:57] bug 1238287 in unity8 (Ubuntu) "unity8 crashed with SIGABRT in raise()" [High,Confirmed] https://launchpad.net/bugs/1238287 [10:57] tsdgeos, rofl [10:57] MacSlow, slow if the password box appears and I tap 'Cancel' the dialog should vanish. right now it asks the password again [10:57] *also :) [10:58] om26er, that's the triggering app not the notifications [10:59] reponsibility [10:59] aah [10:59] responsibility even [11:00] om26er, remember... notifications are not domain/context-aware === shentey_ is now known as shentey === salem_ is now known as _salem [11:07] om26er, there's a bug [11:08] om26er, bug #1236386 [11:08] bug 1236386 in Network Menu "Wi-Fi selection offers two opportunities to connect" [Medium,Triaged] https://launchpad.net/bugs/1236386 [11:08] Saviq, cool. I have very small testing on a secure wifi network. I just happened to be at a place where the network was password projected so faced that issue === MacSlow is now known as MacSlow|lunch [11:09] om26er, scary, as mentioned ;) [11:09] om26er, you know securing networks is not about people not being able to "steal your internet" but "steal your data" instead? ;) [11:10] Saviq, I think no one in the neighborhood have those skills but I get your point :) [11:11] om26er, "skills"? like connect to your network and access the services you left unintentionally open? ;) [11:12] Saviq, I don't mind they use some of the free internet ;). also this house is big even I don't get signals in some rooms.. [11:12] i should enable password still.. [11:13] :D [11:16] didrocks, can we get the Qt fix in today? [11:17] Saviq: which qt fix? we have 4 in flight :p [11:17] didrocks, ah, the looping [11:17] didrocks, bug #1236765 [11:17] bug 1236765 in qtdeclarative-opensource-src (Ubuntu) "[PATCH] Fix infinite loop in QQmlIncubator::forceCompletion" [Undecided,In progress] https://launchpad.net/bugs/1236765 [11:17] Saviq: already in proposed [11:17] Saviq: this is with our branches combined: http://paste.ubuntu.com/6221959 [11:17] didrocks, awesome [11:18] mzanetti, how about the rest of the tests? :D [11:18] mzanetti, but yeah, means it's working [11:18] mzanetti, approve/merge, then? [11:18] Saviq: yeah... exactly... not sure, did tsdgeos fix the others yet? [11:18] Saviq: yep, can approve yours [11:18] mzanetti, there were no other real failures [11:18] ah ok [11:18] mzanetti, only crashes and stuff [11:18] right... I had one crash too [11:19] Saviq: ah I think I'm close in reproducing the edge drag crash [11:19] Saviq: can reproduce it by frequently tapping 2 edges (e.g. left + right) [11:19] mzanetti, oh good, if you get a .crash at any point [11:19] mzanetti, send it to me [11:19] tvoss_: you know what's interesting? in the desktop that cache is not destructed if you kill -15 the process, just if you shut down it properly (i.e. alt+f4) [11:19] Saviq: well, I do have a quite useful crash trace. [11:20] Saviq: but reading through the code I'm not sure how we could end up in that situation [11:20] mzanetti, ah [11:20] tvoss_: my current guess is that the thing is being executed when it should not, and the svg lib is already unloaded and thus can't delete properly the QSvgIconEngine because it has no clue on how to do it [11:20] tvoss_: but not sure if it can be possible that the svglib is already unloaded and no idea how to check if it has happened [11:21] anyone knows enough about C++/elf app shutdown to say how/if libs are unloaded on shutdown? === hikiko|lunch is now known as hikiko [11:21] tsdgeos, why kill -15? kill -9 should be enough [11:22] Saviq: this is it btw: http://paste.ubuntu.com/6133905/ [11:22] tsdgeos, I would think getting rid of the cleanup should help [11:22] of course would help [11:22] but it's the wrong thing to do :D [11:22] Saviq: so what happens is that we get an invalid touch event in touchEvent_recognized(event *) [11:22] tsdgeos, not really, because the cache is destructed anyway in qt5 [11:22] tvoss_: wait, with cleanup you mean that [11:22] mzanetti, :/ [11:22] tsdgeos, post main, that is [11:22] tvoss_: no it won't help, that code is not executed on the phone [11:22] Saviq: would be easy to fix by adding a check there. but doesn't feel like its the right place to fix it [11:23] tsdgeos, ? [11:23] because that code only executes on "correct" shutdown [11:23] and we're killing it [11:23] not doing correct shutdown [11:23] tsdgeos, we are really just kill -9'ing it? [11:24] tvoss_: not sure what "stop unity8" does, but i guess something along the lines [11:24] i mean how do you stop it otherwise? [11:24] tsdgeos, I would assume it sends a friendly sigterm first, before it gets out the gun [11:24] sure [11:24] -15 [11:25] that's what i said [11:25] tsdgeos, ah, sorry [11:25] anyhow [11:25] when we -15 it [11:25] tsdgeos, but the cleanup handler should run with -15, too [11:25] it doesn't execute QCoreApplication deletion [11:25] so the postroutines are not executed [11:26] only the static deletion ones [11:26] no idea what I'm doing wrong [11:26] I'm trying to connect the genericscopeview with the shell or greeter [11:26] I have this connection with proper target, nothing works [11:34] Cimi: code? [11:35] oh interesting [11:35] mzanetti, can you try: stop maliit-server; autopilot run unity8.shell.tests.test_lock_screen.TestLockscreen.test_can_unlock_passphrase_screen [11:35] mzanetti, and do the same after starting maliit again [11:35] Saviq: sure [11:35] tsdgeos, with surprise I realised the unlock signal of lock screen and greeter is not what I thought it was [11:36] I don't think it's emitted when I slide the lockscreen [11:36] mzanetti, for me I'm not getting keyboard input unless maliit is up [11:38] Saviq: confirmed [11:38] which seems really weird tho [11:38] indeed [11:39] Saviq: I think there was some WIP to actually use maliit to inject the events. no idea how far that got [11:40] I thought we'd still use /dev/uinput [11:40] mzanetti, not that it should matter anyway [11:40] mzanetti, yeah we are [11:41] Saviq: must be something in the the qpa. as it obviously works on the desktop without maliit being up [11:41] Saviq: probably some if (!maliit.connected()) return [11:45] https://code.launchpad.net/~om26er/unity8/header_height_dash_5gu/+merge/190622 [11:48] thanks om26er [11:48] Saviq: hey, I've 2 branches attached to https://bugs.launchpad.net/unity8/+bug/1237850 - who can review? [11:48] Ubuntu bug 1237850 in Unity 8 "Apps started via upstart don't participate in window management correctly" [Critical,In progress] [11:48] Saviq, is that assigned to me? https://bugs.launchpad.net/unity8/+bug/1226221 [11:48] Ubuntu bug 1226221 in Unity 8 ""Recent searches" are not persistent and are not common between scopes" [Undecided,Triaged] [11:52] mzanetti, you gonna merge the autopilot workaround or am I? [11:53] Saviq: this? [11:53]