[08:15] dednick: morning. We stumbled across https://bugreports.qt.io/browse/QTBUG-32859 again (with gsettings-qt). Do you know if there's any activity towards fixing that? [08:20] larsu: don't think there's any activity on it. [08:20] too bad - this might hit us in more places [08:21] larsu: ya. i don't doubt it. [08:22] I'll bump the bug report for an update. [08:22] thanks [08:22] do we have a downstream bug as well? [08:25] larsu: I don't think so [08:25] dednick: I'll make one [08:25] larsu: ta [08:25] * larsu wonders if launchpad knows about the qt bug tracker as well [08:26] it doesn't [08:26] larsu, https://bugs.launchpad.net/launchpad/+bug/157488 [08:26] Launchpad bug 157488 in Launchpad itself "Add bugwatch support for the JIRA bugtracker" [High,Triaged] [08:27] seb128: wow you're quick. This is almost mpt-level in bug-finding skill [08:27] haha [08:27] larsu, well, I knew about that bug/looked at it in the past [08:27] seb128: what's the launchpad project for qt? [08:27] lp:qt doesn't have bugs [08:28] larsu, what part of qt? [08:28] larsu, https://bugs.launchpad.net/ubuntu/+source/qtbase-opensource-src/+filebug [08:28] for qtbase [08:28] weird. Thanks seb128! [08:28] qtdeclarative is another source though [08:29] ya this is base [08:36] dednick: http://pad.lv/1460970 [08:36] Launchpad bug 1460970 in qtbase-opensource-src (Ubuntu) "DeferredDelete events not handled in time when created by GLib dispatched events" [Undecided,New] [08:41] larsu: ta [08:58] dednick, larsu: will adjust the merge request [08:59] ltinkl: which merge request? [08:59] dednick: https://code.launchpad.net/~lukas-kde/gsettings-qt/queued-processing/+merge/259883 [09:00] ltinkl: ah. [09:00] ltinkl: thanks (and sorry for being a bit padantic) [09:00] dednick: to mention the QT bug report [09:00] larsu: no problems [09:14] ltinkl larsu: i've commented on the MP [09:22] dednick: ah I wasn't aware that deleteLater is defined as delete-in-idle. mzanetti favors that approach as well. Fine by me [09:23] larsu: I think dednick is afk [09:23] I think dednick has an issue with IRC notifications :) [09:24] [11:15] * dednick brb. doc appt. [09:24] ah ok [09:24] this time not :) [09:24] meh, I'm commenting on the mr as well [09:24] didn't know you could force queued connections [09:24] that's pretty neat [09:25] larsu: pretty much identical to the first solution [09:25] ltinkl: no, because it fixes it for every user of qgsettings as well [09:25] the first solution was qml-only [09:25] right. yeah, +1 for doing it one level down [09:25] right, this is the C++ iface [09:26] ltinkl: people use that as well somewhere [09:26] dunno where, but I remember people asking for it early on [09:26] * larsu had planned a qml-only library before [09:26] didn't know that, the project says "qml bindings" ;) [09:26] yeah... launcher's model uses it in c++, just one example [09:26] also in QtMir [09:26] now it might stop leaking memory *cough* [09:27] haha [09:27] * larsu hopes it's not actually doing that [09:28] mzanetti, ltinkl: speaking of qt, do you know what could possibly be the cause for this library to need the QObjectPrivate() symbol? pad.lv/1426335 [09:29] need to dig the code [09:30] now that's odd... grepping through it, it only gives results in the binary, not in the code [09:30] ya [09:30] might be in one of those macros [09:31] ltinkl, can you have a look please? I need to jump to a meeting [09:31] otherwise I can do in half an hour [09:31] no worries, just thought this might be something you stumbled onto before [09:33] mzanetti: sure [10:00] mzanetti: i have a problem with noticing IRC notifications ;) [10:00] :) [10:00] i think i need a forhorn [10:00] lol [10:00] foghorn [10:00] or a vuvuzela [10:01] haha. good idea. unfortunately i normally have laptop sound muted so might not help [10:06] you could always use a USB rocket launcher to shoot you on IRC notifications [10:06] or a red alarm LED sitting on your laptop's lid [10:07] led is not intrusive enough [10:07] ye or some effect that will start shaking all the windows [10:07] wobbly on steroids [10:09] ouch. the foghorn is quite annoying. [10:12] larsu, dednick: review updated, I guess we'll also have to rollback the descriptions now :) [10:13] and I found that symbol problem [10:13] ltinkl: tests still work? [10:13] qqmlpropertymap dereferences a QObjectPrivate in its protected constructor [10:13] I don't even know what that is public?! [10:14] I could not use that protected constructor, but the docs explicitly say to do so [10:16] ah, that doesn't work indeed (tests fail because the type isn't registered correctly) [10:30] dednick: yup [10:31] ltinkl: ok. if you're ready can you mark as "ready for review" ? [10:32] dednick: ah right [10:37] larsu: is g_settings_set_value synchornous? [10:38] i'm concerned that set + get in same mainloop cycle won't work anymore. [10:38] (if it did before) [10:39] dednick: mh? this is about change events, no? [10:39] larsu: the qml componnent for gsettings updates the internal value only when the change event comes in. [10:43] dednick: I thought qqmlpropertymap set the value from the return value of updateValue() [10:44] dednick: the change event always comes in later, because _set_value() is async (and it might not even come at all for the gsettings object that originated the change) [10:45] larsu: possible. ok, it's fine in that case. [10:46] +1 [10:46] ltinkl: you confirmed the deleteLater is being effected with the queued invoking? [10:47] perhaps we should add a unit test for it. [10:47] dednick: I think we have a test that sets/gets during the same iteration [10:47] at least i hope so... :) [10:47] yup, there is one [10:47] function test_changed() [10:48] and test_write() [10:48] ltinkl: how about a test for the delete later? [10:48] dednick: using the loader+component reproducer? [10:48] or "posted events" even [10:49] dednick: this one http://paste.ubuntu.com/11494214/ [10:50] ltinkl: as long as it uses deleteLater in the loader. Not sure the garbage collection for qml is entirely reliable. [10:51] it does [10:51] (for me *cough*) [10:52] :) might be better as a Qt C++ test [10:52] ye === MacSlow is now known as MacSlow|lunch === mzanetti is now known as mzanetti|run === MacSlow|lunch is now known as MacSlow === mzanetti|run is now known as mzanetti [12:56] larsu, dednick: added the requested C++ testcase for the MP [12:57] thanks, will have a look after lunch [12:58] ltinkl: ta [12:59] ltinkl: can you change the file name? test2.cpp isn't very descriptive! :) [12:59] test_gsettings or something [13:00] dednick: well the qml test is called test.cpp so... :) [13:08] ic. do you need the other pro file? [13:08] ltinkl: ? [13:09] dednick: umm, I guess so; it's a different target [13:09] dednick: QML vs. C++ testcase [13:16] ltinkl: ok. can you rename the target and testfile to cpptests at least? [13:17] dednick: ok === dandrader_ is now known as dandrader|afk [13:39] dednick: done [13:41] yes... yes... yes... [13:41] * MacSlow does _the_ dance [13:41] ...and wonders how he can avoid having to dring alcohol [13:42] dandrader|afk, we have to do the dance I guess :) [13:43] ummm, MacSlow i thot the bet was before the end of sprint :) [13:44] altho...you could blame mir guys for sabotage :-P [13:44] kgunn, technically yes... thanks for reminding me :) [13:44] kgunn, nah... they were just worried about my health ;) [13:47] brb [13:54] ltinkl: approved [13:57] dednick: ty === dandrader|afk is now known as dandrader [14:01] MacSlow, what? it has landed? [14:01] dandrader, well... approved at least... not fully landed yet... still I feel like funky-dancing :) [14:02] dandrader, tutorial available here https://www.youtube.com/watch?v=Zdz88MBWomo [14:03] :) [14:46] So re yesterday: what is the correct way to generate a .qmltypes file using qmlplugindump... I keep getting an error === Malsasa_ is now known as Malsasa === mzanetti is now known as mzanetti|food [15:09] It's been ages since I last used qmlplugindump. I think dednick used it recently [15:20] josharenson: you should just be able to call make qmltypes [15:21] if you're using unity8 i mean [15:21] it updates everything [15:22] josharenson: otherwise, take a look at the unity8/cmake/modules/QmlPlugins.cmake file for the correct usage. === mzanetti|food is now known as mzanetti === dandrader is now known as dandrader|lunch [17:00] kgunn: is there a list of video drivers that unity8/mir supports? (specifically I am using https://wiki.ubuntu.com/Unity8inLXC ) [17:01] popey: so you can think of unity8/mir running on any android driver set and any mesa driver set [17:01] nouveau? [17:02] popey: in theory yes [17:02] :) [17:02] popey: well, it does work, but we also hear alot about a variety of bugs that come and go [17:03] as nouveau isn't really backed by nvidia.....the result is kinda just that === alan_g is now known as alan_g|EOD [17:03] ok, just tried and it doesn't work at all, drops back to a login screen [17:03] oh, no, it froze [17:04] popey: so is this unity8 desktop session ? [17:04] yes [17:04] * balloons listens [17:04] before doing that, you could just install mir demos, and do the vt dance just to see if the mir+nouveau is ok [17:04] the vt dance? [17:04] sorry, I'm new to this [17:05] popey: under "running mir natively" http://unity.ubuntu.com/mir/using_mir_on_pc.html [17:05] ok [17:05] might be better to pop over to #ubuntu-mir also [17:06] ok [17:09] kgunn, ohh so someone is still using unity.ubuntu.com.. interesting [17:10] balloons: only b/c someone helped with some automagic to upload our .md files to that.... [17:15] probably an old CI job [17:15] good to know however === dandrader|lunch is now known as dandrader === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader