[01:37] <mterry> Guh QML
[06:57] <Mirv> bregma: FYI Unity7 has a blocker, failing to build on armhf https://bugs.launchpad.net/unity/+bug/1209512
[08:05] <mhr3_> Saviq, could i get you to have a rough look at https://code.launchpad.net/~mhr3/unity8/scopes-category-overriding/+merge/179015 ?
[08:05] <mhr3_> just to make sure it's going in the right direction
[08:06] <tsdgeos> does the vpn work for you?
[08:06] <Saviq> tsdgeos, +1
[08:07] <tsdgeos> i can't connect :-/
[08:07] <tsdgeos> ah connceted now
[08:07] <tsdgeos> just took more than usual
[08:08] <Saviq> mhr3_, not liking the Repeaters in DashApps.qml - would probably go for a ProxyModel
[08:08] <mhr3_> you can transform the data with ProxyModel?
[08:08] <tsdgeos> lol
[08:08] <Saviq> mhr3_, sure you can
[08:08] <tsdgeos> we killed the X server in an autopilot test
[08:08] <mhr3_> good to know :)
[08:09] <Saviq> mhr3_, and we should be able to get rid of "Available for download", no? the click scope should be providing it already?
[08:09] <tsdgeos> http://paste.kde.org/~tsdgeos/p4fdedc31/
[08:09] <tsdgeos> i like the "mieq is *NOT* the cause.  It is a victim." debug line
[08:09] <tsdgeos> :D
[08:09] <mhr3_> Saviq, i think they're perfectly capable of removing one line once it's in :)
[08:10] <Saviq> mhr3_, ;)
[08:10] <Saviq> tsdgeos, :D
[08:11] <Saviq> mhr3_, other than that - yeah, looking good
[08:12] <mhr3_> Saviq, i'm unsure whether the running apps will work, couldn't test it on the phone
[08:13] <Saviq> mhr3_, re: reparenting - might be tricky, we probably should "undo" the override when it's destroyed, instead
[08:13] <Saviq> mhr3_, should work, no reason why it wouldn't
[08:14] <mhr3_> Saviq, noting that, will remind you of it in a few days when you'll scream at me :P
[08:15] <mhr3_> Saviq, so i'll port also home to use the real thing
[08:17] <Saviq> mhr3_, yup
[08:18] <mhr3_> Saviq, any pointers how to do the transform with ProxyModel?
[08:18] <Saviq> mhr3_, just implement your own data()
[08:19] <mhr3_> you can do that in qml?
[08:19] <Saviq> mhr3_, no
[08:19] <mhr3_> another good to know thing :P
[08:19] <mhr3_> eeek
[08:19] <mhr3_> then i don't like that
[08:19] <Saviq> mhr3_, I don't like the Repeater more ;)
[08:19] <mhr3_> Saviq, come on, it's mocked data
[08:20] <Saviq> mhr3_, not for running apps i isn't
[08:20] <Saviq> t
[08:20] <Saviq> mhr3_, when they get reordered
[08:20] <Saviq> mhr3_, or added / removed
[08:20] <Saviq> mhr3_, the Repeater approach will just bite you in the a$$
[08:20] <mhr3_> Saviq, it's not using the repeater for the running apps
[08:20] <mhr3_> just the mocked freq and downloadable
[08:21] <Saviq> mhr3_, ah, so we're targetting the "real scope will already be there" case?
[08:21] <mhr3_> well not exactly, that's why i said i'm unsure if the running apps will work :P
[08:22] <Saviq> mhr3_, they would, with a Proxy ;)
[08:23] <mhr3_> then i'll need to update also the renderer, right?
[08:23] <mhr3_> meaning "Apps/RunningApplicationsGrid.qml"
[08:24] <Saviq> mhr3_, yeah, to use the real scope roles
[08:24] <Saviq> mhr3_, but that would need to happen for the "real running apps" scope anyway
[08:24] <Saviq> mhr3_, so a good thing
[08:24] <mhr3_> right, see, i almost broke running apps :P
[08:25] <Saviq> mhr3_, I'd really go for a c++ ProxyModel or two, they would need to be maybe 50 lines each
[08:25] <Saviq> mhr3_, well, for the mock data
[08:26] <Saviq> mhr3_, just make the mocks have the "real scope roles" straight away
[08:26] <Saviq> mhr3_, no need for a Proxy or the Repeater - they're just ListModels with some ListElements inside them
[08:26] <mhr3_> also true, sounds easier
[08:26] <Saviq> mhr3_, so as long as they have the same roles, you just pass it straight in
[08:26] <tsdgeos> Saviq: in https://code.launchpad.net/~saviq/unity8/reenable-battery-drag-test/+merge/178911 when you say "make sure it works on devices", you mean running the autopilot test on the Nexus4?
[08:27] <tsdgeos> also the Nexus10 or one should be enough?
[08:28] <Saviq> tsdgeos, would be good for N10 as well
[08:28] <tsdgeos> ok, me tries in both places
[08:28] <Saviq> tsdgeos, should be just a case of "grab the output.zip; phablet-run-test -n unity8"
[08:28] <Saviq> installing the stuff from the .zip on the device first, of course
[08:29] <tsdgeos> i probably need to flash the Nexus10 first
[08:29] <tsdgeos> it's ooold
[08:35] <Twenty-three> hi, i think a recent kernel update was incompatible with my graphics card or something, someone Dr_willis suggested trying to create a new user and testing whether the problem persisted, i tried that and the problem persisted, i am not sure how to restore unity
[08:40] <tsdgeos> Saviq: obviously our autopilot packages miss some deps
[08:40] <Saviq> tsdgeos, apt-get -f install
[08:40] <tsdgeos> i'm getting: RuntimeError: Unable to instantiate any backends
[08:40] <Saviq> tsdgeos, but actually
[08:40] <tsdgeos> Saviq: did that already
[08:40] <Saviq> tsdgeos, phablet-test-run -i
[08:41] <tsdgeos> and after that i had to manually install python-autopilot
[08:41] <Saviq> tsdgeos, that will install all that's needed on the device for autopilot
[08:41] <tsdgeos> and now i'm getting that
[08:41] <tsdgeos> ahhh
[08:41] <Saviq> tsdgeos, yeah, phablet-test-run -i will do stuff
[08:41] <Saviq> tsdgeos, need ssh first, though
[08:41] <tsdgeos> got that
[08:41] <Saviq> bug #1205244
[09:02] <larsu> I've MRed the themeiconprovider to the sdk, is there a way to make another module depend on that?
[09:02] <larsu> I don't want to duplicate the code between unitymenumodel and the sdk
[09:02] <larsu> and I'm sure it might be problematic when two modules add icon providers with the same name
[09:06] <Saviq> larsu, why two with the same name?
[09:07] <larsu> Saviq: if I were gto leave the provider in unitymenumodel
[09:07] <Saviq> tsdgeos, it should've been 12 tests with the reenabled test
[09:07] <larsu> and somebody links against that and the sdk
[09:08] <tsdgeos> Saviq: :S
[09:09] <Saviq> tsdgeos, or waity
[09:09] <Saviq> tsdgeos, no, 11
[09:09] <Saviq> tsdgeos, it was skipped before, and now it passes, that's fine :)
[09:09] <tsdgeos> phew
[09:10] <dednick> Saviq: ping. we may have run into a bit of a snag with this glib event loop issue i was having with indicators.
[09:11] <Saviq> dednick, :/
[09:11] <Saviq> dednick, how bad?
[09:11] <dednick> Saviq: https://bugreports.qt-project.org/browse/QTBUG-32859
[09:11] <dednick> Saviq: "Do we even want to support calling glib directly?"
[09:11] <dednick> Saviq: "I'd say that running GMainLoop directly is out-of-scope."
[09:12] <larsu> dednick: please don't approve ted's indicator-signals merge yet. I've had an unresolved discussion with him and we've decided to continue this discussion later and focus on other work right now.
[09:12] <dednick> larsu: ok
[09:12] <tsdgeos> larsu: can you comment that on the MR itself ? i was roaming on that MR and almost approve it :D
[09:12] <larsu> thanks
[09:12] <dednick> larsu: think it needs some work anyway
[09:13] <dednick> Saviq: i do have a potential solution, but it involves work for every instance we have a glib callback calling back into qt.
[09:13] <larsu> dednick: if you ask me, it needs to die :)
[09:13] <dednick> Saviq: essentially need to send a custom QEvent every time
[09:13] <dednick> Saviq: which is how qt handles window events and such.
[09:14] <larsu> tsdgeos: it was quite a heated discussion, I wanted a low-profile way to stall this a bit until I have time to talk to him again
[09:14] <tsdgeos> ok
[09:16] <Saviq> dednick, re: the comments, I think we need to restate that the g_ thing is just the simplest way to reproduce, and not how we're trying to use it
[09:17] <dednick> Saviq: fyi, this problem is valid for anywhere else we use glib. which means probaly means everything from unitycore::scopes.
[09:17] <Saviq> dednick, yeah, I understand
[09:17] <Saviq> dednick, hmm but wait, the problem is that we actually *do* use glib directly in there?
[09:17] <tsdgeos> hmmm
[09:17] <dednick> Saviq: yes
[09:17] <tsdgeos> Nexus 10 is not flipped?
[09:18] <Saviq> tsdgeos, yes it is
[09:18] <dednick> Saviq: we're connecting directly to gmenumodel
[09:18] <tsdgeos> ah wait phablet-flash failed
[09:18] <Saviq> dednick, right
[09:19] <tsdgeos> well it' a bit lame they claim glib event loop integration
[09:19] <tsdgeos> if the event loop integration doesnt' work :D
[09:19] <dednick> Saviq:  I think their point is that we should be sending events back through the qt event system.
[09:19] <Saviq> MacSlow, I reproduced the autopilot failure when running the tests on device
[09:21] <MacSlow> Saviq, I'll look into it later today.
[09:21] <Saviq> MacSlow, thanks
[09:22] <MacSlow> Saviq, I already switch the branch back to "wip" to avoid load on the jenkins-metal
[09:22] <Saviq> MacSlow, yeah, saw that
[09:23] <tsdgeos> ha, i'm 0% of battery, that's why flashing fails :D
[09:23] <MacSlow> Saviq, currently back on the UI-side of things regarding the "extended snap-decisions"
[09:23] <Saviq> MacSlow, yup
[09:24] <Saviq> larsu, btw, did you have time to write the menumodel vs. notifications email?
[09:24] <larsu> Saviq: typing right now :)
[09:24] <Saviq> larsu, thanks
[09:25] <MacSlow> larsu, looking forward to it
[09:27] <Saviq> dednick, the QEvent thing would be a workaround, then, while incrementing the loop level would be a solution, right?
[09:31] <dednick> Saviq: I think we would need to increment the loop level in our code (in the glib callbacks). I'm not sure there's a way around it. Problem is that the both routes (to glib handler, and to delete) pass through the same code.
[09:32] <dednick> Because all the event handling goes through the glib event dispatcher.
[09:33] <dednick> Saviq: unfortunately the QEvent is the actual solution in this case it would seem. I think that's how it was intended to work. Unless i can find a better option.
[09:34] <Saviq> dednick, that sounds like a substantial amount of work, too?
[09:34] <dednick> Saviq: i'm looking into it now, but most likely quite a bit.
[09:34] <dednick> custom event types for each different callback.
[09:35] <MacSlow> nic-doffay, hey there...
[09:36] <nic-doffay> MacSlow, what's up?
[09:36] <MacSlow> nic-doffay, I'm trying to get an expanded OptionSelector into a flickable (with a scrollbar) ... without any success sofar... it that possible at all?
[09:36] <dednick> Saviq: the only other solution i can potentially think of is to pull all the custom glib stuff onto a different loop. But I'm not sure how possible that would be, or how much work.
[09:37] <seb128> MacSlow, where did you find "OptionSelector"?
[09:37]  * seb128 needs it but though that didn't land yet
[09:37] <nic-doffay> MacSlow, I haven't tried a scrollbar but I've put it in a flickable without issue.
[09:37] <Saviq> seb128, yeah, not landed yet
[09:38] <MacSlow> seb128, it's in a not-yet-merged branch... I'm trying to get a feel for this... as it'll be needed in the wifi-ap selection dialog/snap-decision
[09:38] <MacSlow> nic-doffay, ok... I keep trying
[09:39] <nic-doffay> MacSlow, I'm going to pastebin you some of my filters code.
[09:39] <MacSlow> nic-doffay, ok... thanks
[09:39] <nic-doffay> It has scrolling too.
[09:40] <nic-doffay> https://pastebin.canonical.com/95620/
[09:40] <nic-doffay> MacSlow, ^
[09:40] <Saviq> dednick, so if I understand correctly, we're fooked regardless which route we'll go, and it's not just something that Qt guys say they don't care about
[09:40] <Saviq> dednick, so there's no "easy" fix for Qt itself
[09:40] <nic-doffay> MacSlow, you'll see there there are three OptionSelectors in a column which is in a flickable.
[09:40] <MacSlow> seb128, if you want to give it a try... lp:~nicolas-doffay/ubuntu-ui-toolkit/list-item-option-selector
[09:40] <MacSlow> nic-doffay, great.. thanks!
[09:40] <dednick> Saviq: yes, from my understanding of the issue.
[09:41] <nic-doffay> MacSlow, np!
[09:41] <seb128> MacSlow, thanks
[09:42] <seb128> MacSlow, nic-doffay: when is that likely to land?
[09:42] <Saviq> dednick, :|
[09:42]  * seb128 wants to start using it in system settings
[09:42] <seb128> our current hacked valueselector suck
[09:42] <dednick> Saviq: i may have a very dodgy workaround for the cpu problem that we can use for the moment while i work on the issue. Will only work while we have the proxy to qmenumodel though.
[09:43] <MacSlow> seb128, when you use it... (without installing it) be sure to use qmlscene -I path/to/modules ... and "import Ubuntu.Components.ListItems 0.1" so you can use OptionSelector in your QML-code.
[09:44] <nic-doffay> seb128, got feedback from design yesterday.
[09:44] <MacSlow> seb128, with "path/to/modules" being where ever you put nic-doffay's list-item-option-selector/modules
[09:44] <nic-doffay> Basically I need to ensure there's a pause between switching the chevron and tick then it's landable.
[09:44] <seb128> MacSlow, thanks
[09:44] <nic-doffay> seb128, I'm unsure when it will be reviewed though, obviously I'll push for asap.
[09:44] <seb128> nic-doffay, great, soon then
[09:50] <Saviq> dednick, yeah, would be good to have something temporary, even
[09:55] <tsdgeos> the tablet has regressed
[09:55] <tsdgeos> can't seem to be able to run the sidestage apps if there are no mainstage apps open
[09:57] <tsdgeos> well, you can run it
[09:57] <tsdgeos> it just doesn't paint
[09:57] <tsdgeos> Saviq: want me to open a bug?
[09:57] <Saviq> tsdgeos, please do
[09:59] <tsdgeos> what was the file that has the build number?
[09:59]  * tsdgeos always forgets
[10:02] <MacSlow> nic-doffay, I can't get your example to work fully... I can expand option-selectors, but I can't scroll through them
[10:02] <nic-doffay> MacSlow, you mean with the flickable?
[10:02] <MacSlow> nic-doffay, yes
[10:03] <nic-doffay> MacSlow, weird, I didn't have any issues.
[10:03] <nic-doffay> You can check my whole branch if you want.
[10:03] <nic-doffay> MacSlow, it could be because it's disabled.
[10:03] <nic-doffay> MacSlow,  interactive: false
[10:03] <nic-doffay> ?
[10:03] <MacSlow> nic-doffay, I changed those of course
[10:04] <nic-doffay> MacSlow, then I have no idea, sounds weird.
[10:05] <MacSlow> I tried to add it as an element in my dialog... but there it just didn't show up... I 'm currently using it stand-alone (also set a fixed width and no longer parent.width)
[10:06] <MacSlow> nic-doffay, also disabled (commented out) the anchors to be able to use it stand-alone.
[10:23] <nic-doffay> Saviq, I've modified the CrossFadeImage a fair amount, think we'll have to get Kaleo in on this review now too.
[10:23] <nic-doffay> It might be best just to put all that functionality in the OptionSelector and not use CrossFadeImage, or create a new SDK component.
[10:29] <MacSlow> nic-doffay, so I went from this http://ubuntuone.com/3plX5yStNdhHqFS4yLqtO4 to this http://ubuntuone.com/1BZOL12QuErfCfwjq5gIBa which is already a bit closer to that mpt has in mind... but scrolling I could not get going yet... also the icons per entry are not supported in the option-selector as far as I can tell
[10:29] <MacSlow> nic-doffay, that scribble is all the visual-spec I have atm
[10:29]  * mpt applauds
[10:30] <nic-doffay> MacSlow, nope it doesn't support the icon.
[10:30] <nic-doffay> This is what will need to be amended.
[10:30] <MacSlow> mpt, well there's much much left to be done (on the UI-side alone)
[10:30] <mpt> sure :-)
[10:30] <nic-doffay> MacSlow, I think it's probably going to change a hell of a lot.
[10:30] <MacSlow> nic-doffay, yeah... I remember you mentioned there are still discussions gonig on with Design about details (such as this I assume)
[10:30] <nic-doffay> I'm wondering if it's worth even using an OptionSelector.
[10:31] <nic-doffay> Or at least just copying and pasting the relevant code.
[10:31] <nic-doffay> MacSlow, like the selection stuff etc.
[10:32] <MacSlow> nic-doffay, at least I can actually select elements in the list now... that didn't work with the first iteration http://ubuntuone.com/3plX5yStNdhHqFS4yLqtO4
[10:33] <nic-doffay> MacSlow, it looks awesome!
[10:33] <MacSlow> thx
[10:49] <tsdgeos> ouch, my X suddenly restarted out of nowhere :-/
[10:51] <mlankhorst> valgrind :-)
[11:20] <dednick> Saviq: https://code.launchpad.net/~nick-dedekind/unity8/indicators-cpuburn.QTBUG-32859/+merge/179161
[11:40] <Saviq> dednick, +1 on greyback's comment, and FWIW I don't think that's a huge hack
[11:40] <Saviq> nic-doffay, you should make the CrossFadeImage mods a separate merge request, though
[11:41] <Saviq> nic-doffay, and it's the sdk guys that need to review anyway, I'm only looking at it from a visual / functional perspective
[11:41] <dednick> Saviq: yeah, but it's kind of a sidestep just because we're using a proxy. the problem will re-appear when we change to unitymenumodel if not fixed.
[11:41] <Saviq> dednick, right
[11:44] <dednick> Saviq: comment added.
[11:45] <Saviq> cheers
[11:52] <Saviq> dednick, so that's bug #1206991 / bug #1183065 that should be helped with with that?
[11:53] <dednick> Saviq: yeah
[11:54] <Saviq> dednick, awesome
[11:55] <Saviq> dednick, less awesome that it's not a proper fix, but still awesome, and great investigation, thanks
[11:57] <dednick> Saviq: my pleasure. i have a much more detailed understanding of the qt event loop now. For the next few days until i forget it all at least.
[11:57] <Saviq> :)
[12:42] <Saviq> dednick, we seem to be getting more and more of https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-saucy/975/console
[12:42] <Saviq> dednick, which gets stuck in test_rightEdgeDrag()
[12:42] <Saviq> dednick, you were looking at Daniel's edge drags / velocity calculator
[12:43] <Saviq> dednick, could you look at the test and see if you can reproduce it locally, and ideally - fix?
[12:43] <Saviq> dednick, I've seen that in autopilot tests as well
[12:44] <Saviq> where the unlocking drag got stuck at ~ 1/3 of the screen from the left
[12:45] <Saviq> there's a suspicion that due to the VM's speed the gesture isn't completed properly, 'cause velocity drops down or similar
[13:11] <tsdgeos> and we're back to 695$ :-)
[13:15] <seb128> tsdgeos, waouh, quite a change
[13:15] <tsdgeos> yep :)
[13:19] <Saviq> wow, and refunds
[13:19] <Saviq> nice
[13:20] <Mirv> bregma: so the previous Unity release blocker was resolved, but what remains is the amount of failing autopilot tests (varying on each run) that makes it hard to release. do you see anything at http://pastebin.ubuntu.com/5962552/ not addressed by other bugs you're already working on?
[13:22] <bregma> Mirc, the pinyin problems are in the new ibus, we're investigating those, a couple of the menu problems are because of a new indicators library upstream being broken (bug filed, no progress from them)
[13:23] <bregma> the rest are vaguaries of the test system
[13:24] <bregma> for example the previous Unity release blocker was fixed by fixing the build environment because libhybris broke x.org (nothing to do with Uniy or Compiz)
[13:24] <tedg> dednick, Checking in on the indicator work.  Are we at the point that we can have old/new indicators in Unity?
[13:26] <dednick> tedg: yes, but not in trunk
[13:26] <dednick> tedg: need unitymenumodel to land, then my branch.
[13:27] <tedg> larsu, When is unitymenumodel landing?
[13:27] <tedg> Is it up for review somewhere?
[13:27] <dednick> tedg: https://code.launchpad.net/~larsu/qmenumodel/add-unitymenumodel/+merge/176857
[13:27] <dednick> tedg: there's still a bit missing which larsu is looking at
[13:30] <Mirv> bregma: yes, I know many are long known flakiness in the tests, just pointing out the ones that keep popping up now and make it difficult to release. http://pastebin.ubuntu.com/5962573/ contains a couple more from two earlier runs from today. if there's anything you can do to reduce the flakiness of those tests (the ones in the latter pastebin link didn't show up now in the latest run), it'd help
[13:31] <Saviq> greyback_, standup?
[13:31] <Mirv> not sure if there'd be an option to mark some tests as false alarms, but then the tests don't do much good in the first place if they just randomly fail and they'd be even ignored
[13:32] <greyback_> Saviq: coming
[13:34] <greyback_> tsdgeos: are you doing notes then?
[13:34] <tsdgeos> greyback_: i thought i was, but if you do, all yours :D
[13:38] <tedg> dednick, So talking with larsu he knows of one more bug for that branch.
[13:38] <tedg> dednick, After that's fixed, he thinks we could probably merge it.
[13:38] <tedg> dednick, Is that your assessment as well?
[13:39] <tedg> There'll be additional bugs, but this starts unblocking things.
[13:39] <dednick> tedg: yep
[13:39] <tedg> dednick, Cool.
[13:39] <tedg> larsu, Is it fixed yet?  :-)
[13:40] <larsu> tedg: no. I'm still have the occasional discussion ähere at guadec. I'll try to get it done today thouhg, as I'll be on a train tomorrow
[13:41] <tedg> larsu, Great, that'd be good.
[13:41] <tedg> dednick, ^
[13:42] <dednick> tedg: got it
[13:43] <xnox> tedg: do you know how could review https://code.launchpad.net/~joe-yasi/libdbusmenu-qt/fix-x-timestamp/+merge/174573 a fix against libdbusmenu-qt from community and it looks sane (optional dependencies and fallbacks)
[13:43] <xnox> sil2100 seems to be away.
[13:43] <tedg> xnox, Hmm, have you ping'd agateau ?
[13:43] <tedg> xnox, He'd be my first person to ask, but not sure if he'd do it.
[13:44] <xnox> tedg: nope. agateau is not here either?
[13:44] <tedg> xnox, Hmm, he doesn't seem to be in any channels on freenode.
[13:45] <mhr3> oh ffs, how do i use Regex::escape in qml?
[13:46] <Saviq> dednick, ping
[13:46] <dednick> Saviq: pong
[14:07] <Saviq> tsdgeos, looking at http://paste.ubuntu.com/5962716/ (and the console output)
[14:08] <Saviq> tsdgeos, then, reading http://qt-project.org/doc/qt-5.0/qtqml/qml-size.html
[14:08] <Saviq> tsdgeos, would you consider the error a bug?
[14:08] <tsdgeos> you get a qsize vs qsizef thing?
[14:10] <tsdgeos> Saviq: problem is 0x0 creases a qsize and then you can't assign it to qsizef
[14:10] <tsdgeos> i can "see" why it happens
[14:10] <tsdgeos> but yeah i don't think you should get a warning
[14:10] <mhr3> anyone seen SortFilterProxyModel provide one extra empty result?
[14:10] <mhr3> any ideas why it does that?
[14:10] <Saviq> tsdgeos, yeah, it should get converted automagically, right?
[14:11] <tsdgeos> Saviq: yeah
[14:11] <tsdgeos> mhr3: shouldn't, can you reproduce it?
[14:12] <mhr3> always
[14:12] <mhr3> therefore i guess i'm doing something stupid
[14:13] <tsdgeos> mhr3: not necessarily
[14:13] <tsdgeos> may be exposing a bug in the code
[14:13] <mhr3> tsdgeos, this does it
[14:13] <mhr3> http://paste.ubuntu.com/5962736/
[14:14] <tsdgeos> interesting
[14:14] <tsdgeos> let me see if i can repro too
[14:15] <mhr3> tsdgeos, lp:~mhr3/unity8/scopes-category-overriding
[14:15] <mhr3> apply there
[14:20] <tsdgeos> mhr3: would it be the 7th item in "Recently used Apps"?
[14:21] <mhr3> tsdgeos, yea, also in "More suggestions"
[14:21] <mhr3> tsdgeos, also, works fine it you revert the pastebin
[14:24] <tedg> dednick, Did you notice?  https://code.launchpad.net/~ted/unity8/indicator-signals/+merge/179028
[14:24] <tsdgeos> the DestkopFileModel.qml is weird :D
[14:24] <dednick> tedg: lars told me not to look at it. he said he wanted to discuss with you
[14:24] <tsdgeos> using a repeater to fill a list feels weird
[14:25] <mhr3> tsdgeos, works :P
[14:26] <tsdgeos> mhr3: but you use more memory than you need
[14:26] <mhr3> patches welcome :)
[14:26] <tsdgeos> i mean that repeater is there forever
[14:26] <tedg> dednick, Could you look to make sure t is correct though.  I'm not as familiar with the code, not sure I put things in the right places.
[14:30] <tsdgeos> mhr3: there's something really fishy
[14:30] <mhr3> oh really? :)
[14:30] <dednick> tedg: i was thinking we should have a upstarter object in the utils/unity library rather than embedding the libupstart code in indicators. If we're going to be signalling things from in process, we will problably have do it more than once.
[14:30] <tsdgeos> mhr3: if you add "onCountChanged: console.log("SearchableResultModel count", count)" inside id: frequentlyUsedAppsModel
[14:30] <tsdgeos> it works :D
[14:31] <mhr3> ehm
[14:31] <tsdgeos> yeah that
[14:31] <tedg> dednick, Perahps, but many of those are through libupstart-app-launcher.  And libdbus consolidates the connection.  So there's not too much benefit to abstracting away libupstart.
[14:31] <tsdgeos> mhr3: but it's totally reproducible :S
[14:31] <tsdgeos> comment the line, 7 items
[14:31] <tsdgeos> uncomment it again 6 items
[14:33] <dednick> tedg: except class encapsulation
[14:34] <dednick> tedg: this is c++ after all :)
[14:35] <tedg> dednick, Yes, classes are definitely too cheap in C++.
[14:35] <tedg> Too bad tvoss is off today.  He needs to see how bad C++ is :-)
[14:36] <dednick> tedg: better to keep it readable (unlike C ;). anyway. the code looks to be in the right place.
[14:38] <tedg> dednick, readable_is_only_a_few_underscores_away  -- thanks for looking!
[14:39] <dednick> andHereIWasThinkingCamelCaseWasBest
[14:40] <mlankhorst> opinionHuNoBtr
[15:11] <Saviq> tsdgeos, do you guys still have a list of tasks for unity8-mir integration somewhere?
[15:12] <tsdgeos> Saviq: http://studio.sketchpad.cc/UFFAX8fxc2 ?
[15:14] <Saviq> tsdgeos, greyback|latelun I added implementing QDesktopServices:openUrl so that we don't misplace that task
[15:14] <tsdgeos> ooka
[15:14] <tsdgeos> mhr3: something seems to be wrong at the QSortFilterProxyModel level (or we are breaking it somehow)
[15:14] <tsdgeos> because i do stuff like
[15:14] <tsdgeos> give me your number of rows and the number of the rows of your source model
[15:15] <tsdgeos> and get
[15:15] <tsdgeos> 7, 6
[15:15] <tsdgeos> and why would the proxy have more rows than the source?
[15:15] <tsdgeos> mhr3: i'm going back to finish my other dash task, but this defenitely needs someone investigating
[15:16] <tsdgeos> Saviq: ↑↑↑
[15:16] <mhr3> tsdgeos, i tried adding a qDebug to the sortfilterproxymodel::data, but it's not really called much
[15:17] <mhr3> tsdgeos, where were you debugging this?
[15:17] <Saviq> tsdgeos, mhr3, interesting, did you manage to build a minimal reproducible test case?
[15:17] <tsdgeos> Saviq: nope
[15:17] <tsdgeos> mhr3: http://paste.kde.org/~tsdgeos/pe86c3eff/
[15:18] <tsdgeos> http://paste.kde.org/~tsdgeos/pe86c3eff/raw/
[15:18] <tsdgeos> includes your changes
[15:18] <tsdgeos> and a few qdebus
[15:18] <tsdgeos> +g
[15:19] <mhr3> tsdgeos, yea, so why doesn't it call data() on that object, it's pretty weird
[15:19] <tsdgeos> on the model?
[15:19] <tsdgeos> it does
[15:19] <tsdgeos> the data there is not the *right* one
[15:20] <tsdgeos> virtual QVariant
[15:20] <tsdgeos> data(const QModelIndex & index, int role = Qt::DisplayRole) const
[15:20] <tsdgeos> is the one you want to reimplement if you want to hijack them
[15:20] <tsdgeos> the one there is for "direct use from qml"
[15:22] <mhr3> ah, maybe we could use clearer names next time
[15:26] <MacSlow> nic-doffay, can the divider in an OptionSelector be disabled completely?
[15:27] <MacSlow> nic-doffay, just curious... not that important
[15:27] <nic-doffay> MacSlow, yeah
[15:27] <MacSlow> nic-doffay, didn't find the property yet for this
[15:27] <nic-doffay> MacSlow, sorry the components have support for this the OptionSelector doesn't.
[15:27] <MacSlow> nic-doffay, only the internal "showDivider" flag
[15:27] <nic-doffay> MacSlow, worth adding?
[15:28] <nic-doffay> If so I'll throw it in now.
[15:28] <MacSlow> nic-doffay, *shrugg*
[15:28] <nic-doffay> MacSlow, might as well :)
[15:28] <nic-doffay> showDividers bool it is then.
[15:29] <MacSlow> cool
[15:31] <Saviq> MacSlow, nic-doffay I think this should be theme-driven, no?
[15:31] <MacSlow> Saviq, yeah... maybe even that
[15:31] <Saviq> MacSlow, nic-doffay, I don't think we should be adding a property to disable them
[15:31] <Saviq> MacSlow, and in any case... I don't think there is real visual design for them...
[15:32] <Saviq> MacSlow, if you look at http://design.ubuntu.com/wp-content/uploads/option_selector_01.png for example
[15:32] <Saviq> MacSlow, there are dividers
[15:32] <MacSlow> Saviq, true... was just looking through the code of OptionSelector and was wondering as it is still work-in-progress
[15:33] <Saviq> MacSlow, when you're wondering such things, you need to have a red light light up in your brain - does that even make sense? ;)
[15:33] <Saviq> MacSlow, don't take visuals for granted - push back when you see something inconsistent
[15:35] <nic-doffay> Saviq, totally forgot about themes.
[15:38] <Saviq> MacSlow, even for https://lh4.googleusercontent.com/goPxF-p_6nxA8PM8vdawkie6HVW4ubTJ-hi_ajOHKUs2IQH34uvqIFLofNPP37GUj7enLg_f9kgaT8kegrKvWa3WyISjMmhGs4kRb8Uw_JWFt0v9Af0fC26f there are dividers, where did you see them not being there?
[15:41] <MacSlow> Saviq, one of mpt's scribbles... but again I was just curious and not really pushing for anything -> https://wiki.ubuntu.com/Networking?action=AttachFile&do=get&target=phone-wifi-prompted.png
[15:41] <Saviq> MacSlow, right, don't consider those anything close to visual design :D
[15:42] <Saviq> MacSlow, but in any case - changes like this should be theme-driven
[15:42] <MacSlow> Saviq, of course
[18:14] <Saviq> mterry, if you have a minute: https://code.launchpad.net/~saviq/unity8/prepare-cross-build/+merge/179250
[18:15] <mterry> Saviq, looking
[18:28] <Saviq> mterry, yeah, I thought about the FATAL_ERROR
[18:28] <Saviq> mterry, can make it non-fatal
[20:39] <ESphynx> Hey guys... is any of you going to be at FOSSCON this weekend by any chance? :)
[20:39] <ESphynx> We've got this bug that we'd really like to be fixed in Saucy Salamander and not quite sure whether it's our toolkit's fault or Unity's fault...
[20:39] <ESphynx> Said bug being @ http://ecere.com/mantis/view.php?id=850