[07:09] * tsdgeos waves [07:46] Saviq: Cimi: r1151 is confusing [07:47] "Fix anchor in PreviewListView.qml. " [07:47] and then all it does is remvoe two properties? [07:52] Saviq: r1170 looks confusing too, i was told explicitely we didn't want previews in overview, did that change? [07:53] also i'm not sure i understadn the change in the tests for r116 [07:53] r1169 i mean === anpok_ is now known as anpok [08:04] tsdgeos, lemme look [08:04] tsdgeos, r1151 just got fixed in a prerequisite branch [08:05] tsdgeos, or well, a previous merge [08:06] tsdgeos, in r1140 [08:06] ah [08:06] tsdgeos, it does add a test to verify that fix though ;) [08:07] oki [08:08] tsdgeos, we don't want previews in overview *except* for scopes [08:08] ok [08:09] tsdgeos, we might actually enable it completely [08:10] Mirv: Saviq: we may want to get https://codereview.qt-project.org/#/c/91828/ [08:10] tsdgeos, r1169 test changes → they don't hardcode an expected height, but just rely on the real height of the card as laid out [08:10] Saviq: but why not use expected height? [08:10] tsdgeos, because it might change [08:11] tsdgeos, as we change designs of the card [08:11] right [08:11] but has the benefit of you having to double check why it changed [08:11] if the height is not expected to change with this particular design chang [08:11] it's a bit like the card creator test [08:11] it's a pita to maintain [08:12] but makes you think if that change was exppected or not [08:12] tsdgeos, sure [08:13] but not complaining [08:14] tsdgeos, yeah, that's an approach that other tests in that same suite use, if we decide to change it - fine [08:23] @unity, so, we actually *can* get green ci now, please start looking closely at any failures and if you don't know where they're coming from, let me know so we can find out [08:35] yay [08:38] tsdgeos, so, because I probably won't be able to focus on this here [08:39] tsdgeos, there's a bunch of segfault failures in our CI (like 3 in alt_nav_support) [08:39] tsdgeos, could you please have a look at those [08:39] sure, let me finish deleting email and i'll do that [08:39] tsdgeos, might that even be the thing you linked? [08:40] may be [08:40] tsdgeos, one thing I realized is that waitForRendering on null is SIGSEGV [08:42] yep [08:42] that's been like that for a while [08:43] so we should always verify() stuff that might not be there [08:45] Saviq, can you give me more insights on the in line diff? [08:48] Cimi, where? [08:49] starting from the first readonly property bool settingsShown: item ? item.settingsShown : false - Maybe move up to group? [08:49] in dashContent [08:49] that property needs to be in the loader I think [08:53] Cimi, you should've started with "Saviq, about settings review..." [08:53] Saviq, what's needed to run ap-test for unity8 on the desktop currently... I get lots of "CalledProcessError: Command '['/sbin/initctl', 'stop', 'unity8-dash']' returned non-zero exit status 1" [08:54] Saviq, ECIMIFORGETSCONTEXT [08:55] MacSlow, install unity8 [08:55] MacSlow, or copy data/unity8-dash.conf to ~/.config/upstart [08:56] MacSlow, but it's safer to just install it to not have outdated jobs [08:56] Saviq, certainly did that [08:56] MacSlow, `initctl start unity8-dash` and `initctl stop unity8-dash` after? [08:59] run.sh still brken? [09:00] mzanetti: can you still reproduce this ? https://bugs.launchpad.net/unity8/+bug/1348108 [09:00] Ubuntu bug 1348108 in Unity 8 "Departments combobox can be scrolled when open" [Undecided,New] [09:01] Saviq: is anything for alt_nav missing besides those crashes you mention? [09:04] tsdgeos, yeah, Cimi's 3D effect on the divider that needs to switch between background and nav background color... [09:04] oh fancy :D [09:06] Saviq: hmmmm, testShell is also segfaulting in https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-utopic/977/? but not shown in there [09:06] fginther`: ↑ [09:07] tsdgeos, yeah, but isn't it segfaulting *after* the test there? [09:08] tsdgeos, yeah, the test completed [09:08] Saviq: yep, that's still a bug [09:08] something is crashing on shutdown [09:08] tsdgeos, is that current? [09:08] it's the latest run of alt_nav [09:09] tsdgeos, http://bazaar.launchpad.net/~unity-team/unity8/trunk/revision/1175 should've fixed that [09:09] ok, let me remerge then [09:09] tsdgeos, waait [09:09] tsdgeos, it's, header, preview, table, seeLess, alt... [09:10] tsdgeos, let me remerge in that order [09:10] ok [09:10] and we should make sure that the qmluitest job actually works off of trunk [09:11] yep [09:16] FAIL! : qmltestrunner::ShellWithPin::test_wrongEntries() property text [09:16] Actual (): Enter your password [09:16] Expected (): Enter your passcode [09:16] Loc: [/home/tsdgeos_work/phablet/unity8/unity8/tests/qmltests/tst_ShellWithPin.qml(205)] [09:16] didn't we fix that? [09:16] or am i the wrong branch? [09:23] tsdgeos, LC_ALL=C [09:23] Saviq: LC_ALLC=C is fixing that?¿ [09:23] tsdgeos, yeah, I know [09:24] Saviq: testShell crashed in unity8 trunk [09:24] after some looping [09:24] same shutting down problem [09:24] tsdgeos, ok, looks like a different crash then [09:24] let me attach gdb [09:24] or maybe I didn't fix it right :P [09:24] dednick, can you please sanity-check the ApplicationInfo change in http://bazaar.launchpad.net/~unity-team/unity8/trunk/revision/1175 [09:25] Saviq: sure [09:30] dednick, there's also something confusing in setSurface() that it registers nullptr with SurfaceManager (maybe that's expected?) but then it tries to connect() to it, which is fortunately just a warning, but confusing nevertheless [09:30] Saviq: i've fixed this all in my model branch. [09:31] dednick, oh ok [09:31] Saviq: as for your code, doign setFurcace(nullptr) will call surfaceManager::unregisterSurface which will emit the destroySurface signal [09:31] so you can remove that from the MirSurface destructor [09:31] Saviq: but yeah. it's all a bit twisted at the moment. [09:32] my branch makes it a bit more straight forward. [09:32] dednick, ok, I'll leave it to you :) [09:35] dednick, it did fix the immediate crash, so unless it's crazy stupid, let's leave it in [09:36] Saviq: should be ok. [09:36] Saviq: crash on exit? [09:36] dednick, yeah [09:37] Saviq: yup. seen that. fixed in my branch as well, but could take a little time before MP approved. [09:37] dednick, it was deleting ApplicationInfo, which resulted in a deferred deletion of the surface, which called setSurface(nullptr) on the in-process-of-being-deleted ApplicationInfo [09:37] that crashed because it went appId() on itself [09:37] for the logging [09:37] Saviq: yup [09:38] i removed the deferred delete shizzle. [09:42] Cimi, ok, not entirely sure any more what I meant by the settingsShown comment... [09:42] Cimi, aah [09:42] Cimi, just 2 lines up :) [09:42] Cimi, hit me for more [09:45] Saviq: i'd totally disagree with that "you on't reboot" tbh [09:45] it's the only thing i do with my blackberry [09:45] tsdgeos, yeah, why? [09:45] it's the only way to get the email program to behave again when it goes into dumb mode [09:46] tsdgeos, yeah, fix the email program ;) [09:46] tsdgeos, the only reason to reboot a *working* phone is to upgrade, which happens automagically [09:46] tsdgeos, for other cases, just power off and back on [09:46] Saviq: ok, so we'll have no bugs on release, awesome :) [09:47] tsdgeos, having a reboot option in the dialog is not a way for fighting bugs [09:47] world is not black and white, if you can alleviate people's live a bit by having a reboot option i think it's worth it [09:47] would make me much happier [09:48] tsdgeos, just to back my stance up, both the main mobile OSs on the market don't have a reboot option [09:48] yeah [09:48] and android has flight mode on the shutdown button [09:48] worst place ever [09:48] yeah [09:49] well we don't ;) [09:49] so don't use them as an example of good UX ;) [09:56] why airplane in shutdown is bad? [09:56] in power button [09:56] I find it quite handy [09:59] tsdgeos, can I use iOS an example of good UX? they don't have a reboot either :P: [10:00] Saviq: i have not used ios at all, can't comment [10:01] Cimi: because i don't want to shut down the freaking thing, i just want to toggle the radios [10:01] so it should be in network [10:01] at least, i don't mind it being in shutdown in addition [10:01] it not being in networks is weird [10:01] tsdgeos, well, plane mode is there in the indicators on android too [10:01] ok, maybe i've never used a newer android D: [10:03] ;) [10:04] tsdgeos, it's just a shortcut under the power button [10:08] Saviq, there is no color property for listitems, right? [10:08] Cimi, no idea [10:08] Saviq, I think those settings might not work with different themes [10:09] dednick, you know about theming of listitems? [10:09] Cimi: nope. don't think they are themed. [10:10] dednick, I was wondering if I can set colors and such [10:10] Saviq, I think it can be complicated [10:10] Saviq, shall we keep scope settings with the normal plain look? [10:11] Cimi: hm. imo the colours are a bit wacked in sdk. we seem to have gone the route of. explicit colour setting. [10:12] Cimi, yeah, there's barely a scope that changes those TBH [10:12] Cimi, or not even "barely a" but rather "there is none" [10:12] Cimi: i think theming will change text color and such. [10:12] Saviq, I'll remove scopestyle and such from settings [10:12] Cimi, [10:12] don't, file bugs with SDK instead [10:12] dednick, yeah but that's theming as in Theme, global, we need it per component in here actually [10:13] Saviq: i c [10:13] unless we get theming that we can override for a certain hierarchy [10:13] Saviq, but if they read scopestyle, colors will be broken [10:13] Saviq, there is no way also to theme switch buttons [10:13] Saviq, or the text fields [10:13] qss shizzle. [10:13] checkbutton [10:13] etc [10:14] Saviq, we canot use scopestyle... [10:14] if we have a black theme, for example, all will look broken [10:14] can't we use qss with qml? [10:16] dednick, qss? [10:16] qt style sheets [10:16] dednick, not with plain QtQuick, maybe with Controls [10:16] dednick, but since we have our own Theming engine... [10:18] yeah... don't get me started. [10:19] dednick, yeah, but even with Controls there's a special thing http://qt-project.org/doc/qt-5/qtquick-controls-styles-qmlmodule.html [10:24] sorry had connection dropped [10:24] Saviq: crash still happening on 5.3 tip [10:25] going to try to figure it out [10:25] backtrace is insanely long [10:25] sorry had connection dropped [10:25] Saviq, what was last message you received? [10:25] your last message: if we have a black theme, for example, all will look broken [10:26] ok [10:26] if we want theming using scopestyle, I have to not use listitems and implement mines, and even then I will have issues with the button textures not adapting to colors [10:27] if we have a black theme, for example, all will look broken [10:28] Cimi, don't, file bugs with SDK instead [10:31] tsdgeos: are we seeing the bug #1358169 in the wild, ie would it be a needed item for RTM? [10:31] bug 1358169 in qtdeclarative-opensource-src (Ubuntu) "Avoid double deletion when deleting an incubating component." [Undecided,New] https://launchpad.net/bugs/1358169 [10:31] Mirv: not that i know [10:32] Mirv: but the fact that you don't see it doesn't mean it's not there [10:32] Saviq, so I should keep reading scopestyle and have things look broken? [10:32] in a month or so we could have 5.3.2 too (depending a bit on how it can be scheduled to fit in...) [10:32] it does not make sense to me [10:32] Saviq, part of the scope setting will look themed, part will look of a different color [10:32] Cimi, for now things won't look broken because there isn't a scope that sets background or foreground [10:32] tsdgeos: true. ok, I'll try to see to its landing to 5.3.0. [10:32] Saviq, but there might be in months time [10:33] Cimi, file critical bugs with the SDK [10:33] Saviq, and RTM code will contain bugs [10:33] Cimi, and *if* they don't get fixed in time [10:33] Cimi, we'll work around it then [10:33] Saviq, we can't add bugs hoping they will get fixed in time IMHO [10:33] we don't add bugs period [10:33] Cimi, we have to ;) [10:34] Cimi, that's not "adding bugs" [10:34] that's "add code that now is broken" [10:34] Cimi, that's an incomplete feature, dependant on SDK [10:34] I don't agree with you on that [10:34] Cimi, please just file bugs against unity8 and sdk and let's talk to them [10:34] Cimi, what we don't have the time [10:34] for [10:35] it works for cards are headers because we are responsible for their style [10:35] Cimi, untrue [10:35] but if we use listitems in scope settings, we know that they are broken [10:35] Cimi, header got fixed with icons [10:35] yes, by you [10:35] no [10:35] weeks ago [10:35] by SDK team [10:35] ah ok [10:35] Cimi, and it was broken initially, which just got it more attention [10:39] Saviq, so you are asking me to add something that is broken? [10:40] as soon as someone will use foreground color it will break [10:41] Cimi, let's just make sure the bug gets fixed first [10:42] Saviq, I don't care of sdk now, I want to fix the branch and get it approved [10:42] we can submit a bugreport but I believe what the answer will be [10:44] Cimi, I'm annoyed enough already by the chat in -touch about assumptions, presumptions and all that, please don't get me even more annoyed by it here [10:44] Saviq, eh?? [10:45] Cimi, "I believe what the answer will be" [10:45] Saviq, the all SDK theming is "if you don't like default, redo the style component" [10:45] Saviq, even buttons don't have text color property [10:46] Saviq, there is no themeability of fb/bg in sdk... [10:46] Cimi, which is a bug that I filed with the SDK [10:46] sometimes there is fg (label), sometimes only bg (button) [10:47] Cimi, but going "eh, they're not gonna do it anyway" is not getting us anywhere [10:47] Saviq, I don't think they are going to fix this by RTM, one week time [10:47] Cimi, we need bugs and escalate them [10:47] Saviq, I am on lp with the bug page opened [10:48] Saviq, just sharing my concern with you... [10:48] Saviq, we are both on the same line of thoughts about SDK stuff... [10:48] Saviq, https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1358241 [10:48] Ubuntu bug 1358241 in Ubuntu UI Toolkit "Listitems should allow color properties for themeability" [Undecided,New] [10:53] Saviq: hmm... do we actually even want the AP lockscreen tests? [10:53] meh [10:53] http://paste.ubuntu.com/8079042/ [10:54] tsdgeos, a beauty [10:54] what's Moth :D [10:55] no clue tbh [11:24] how do I write this? validator: root.mode == "number" ? DoubleValidator {} : undefined [11:27] Saviq, ? [11:31] Saviq: ping [11:31] ok had to declare it anyway outside === alan_g is now known as alan_g|lunch [12:02] Holas === alan_g|lunch is now known as alan_g [12:25] mzanetti, pong [12:26] Saviq: https://code.launchpad.net/~dandrader/unity8/lifecycle/+merge/230090/comments/562134 [12:30] mzanetti, https://code.launchpad.net/~dandrader/unity8/lifecycle/+merge/230090/comments/562138 [12:32] thanks [12:34] dandrader: good morning [12:34] mzanetti, morning [12:34] I see you started working on the lifecycle again :) [12:34] dandrader: I'll push the stuff into the right-edge ppa so we can test on devices [12:35] mzanetti, just fixing as per https://code.launchpad.net/~dandrader/qtmir/lifecycle/+merge/230092/comments/561458 [12:35] ah, I see [12:36] dandrader: so seems you broke the MockApplicationManager for tablet mode :/ [12:36] that's probably a bit more efforts to get fixed again [12:36] I hope to be finished with a proper test run on the phone/tablet by the time you have fixed that [12:37] mzanetti, will check you unity-api/lifecycle comment now [12:37] your === pete-woods is now known as pete-woods|lunch [12:42] Saviq: some good news [12:42] i know how to workaround the crash [12:42] the luminance: is making it crash [12:42] http://paste.ubuntu.com/8079669/ makes it go away [12:43] now i need to find out why [12:43] or at least find a better workaround :D [12:50] mzanetti, fixed all your latest finding I think [12:50] findings [12:50] dandrader: in unity8 too? [12:50] I find that hard to believe :) let me test [12:50] mzanetti, the version thing? [12:50] dandrader: the make tryShell in tablet mode? === MacSlow is now known as MacSlow|lunch [12:51] mzanetti, oh, there's a bunch of new stuff... I said that too soon :) [12:51] :) [12:55] mzanetti, so a qml Tansition is also a QtObject? [12:56] dandrader: yes [12:56] mzanetti, ah, good. so it's doable indeed [12:56] dandrader: but it's not contained in "item.children" but only in "item.data" [12:56] right [12:56] dandrader: that's the difference between findChild() and findInvisibleChild() [13:13] * greyback_ stumbled across that last week [13:13] ? [13:13] dandrader: that's the difference between findChild() and findInvisibleChild() [13:13] dandrader: your qtmir branch doesn't build in the ppa: https://launchpadlibrarian.net/182617174/buildlog.txt.gz [13:13] greyback_: ah :) [13:14] mzanetti: I was writing a test to call a function inside q private QtObject, and took me ages to figure out why findChild couldn't find the object [13:14] greyback_: heh, yeah, walking through data always takes ages when doing that on the shell object [13:15] greyback_: that's why findChild normally only walks through children [13:15] mzanetti: gotcha [13:27] Saviq: fyi: https://bugs.launchpad.net/unity8/+bug/1357321/comments/2 [13:27] Ubuntu bug 1357321 in Unity 8 "scope images do not load in HSDPA or 3G" [Undecided,Confirmed] === _salem is now known as salem_ [13:28] mzanetti, yeah, note Victor could still browse the web and all [13:29] Saviq: yeah, but he most likely restarted the browser after the roaming step [13:29] (actually I'm not sure if the browser suffers from this) [13:29] I noticed it with other apps [13:37] mzanetti, ah [13:41] tsdgeos, mzanetti, help me out here... remove transition not triggered if the last item in a model is removed... ideas? [13:42] hmm... [13:42] random stuff: model reset vs model propery removing stuff? [13:42] tsdgeos, it works on all the other removals [13:42] tsdgeos, just the last item is affected, and ListView.onRemove is triggered on it [13:42] Saviq: model may be doing different code for last? [13:43] tsdgeos, doesn't look like it [13:43] ok [13:46] no other random ideas other than a bug === pete-woods|lunch is now known as pete-woods [13:50] Saviq: so there is the thing that delegate0 is never destroyed (for example by cachebuffer) [13:51] don't know why... but maybe that affects this too [13:51] mzanetti, no, I didn't mean last as in the last remaining one, just the last one from the end [13:51] ah ok [13:56] mzanetti, have you ever looked for a Transition in one of your qmltests? [13:56] mzanetti, tsdgeos, confirmed, no remove on removing the last item: http://paste.ubuntu.com/8080113/ [13:57] dandrader: I did that with Animations, not sure about Transition itself [13:57] a behavior on contentX doesn't help [13:57] meh [13:58] Saviq: seems that http://paste.ubuntu.com/8080133/ made the crash go away :/ [13:59] tsdgeos, ugh [13:59] mzanetti, findInvisibleChildren doesn't seem to find them :( Maybe I should try to get the transitions array of the StateGroup instead [13:59] Saviq: let the loop run for a bit more, but it seems that doesn't crash anymore now [13:59] tsdgeos, yeah ok [13:59] tsdgeos, we probably should just move the luminance method into C++ anyway [14:00] dandrader: hmm... maybe worth a try... but if this is giving you a major headache let me know... I thought it should be rather simple to move it over [14:00] Saviq: yeah but that change fixing the crash makes me very unhapy [14:00] tsdgeos, yeah, it does suggest a nasty bug [14:06] mterry: updated the lockscreen branch [14:07] mzanetti, OK finishing up something but will re-look today [14:07] mterry: all tests passing now and the autoconfirm mode is back === zbenjamin_ is now known as zbenjamin [14:07] mzanetti, I asked Olga/Esti about device unlock autoconfirm UI. we'll see [14:08] It was part of another thread, forgot to cc you [14:08] ah ok... [14:08] well... I still think its wrong to limit the pin to 4 digits [14:08] if you *really* don't want to press more than 4 buttons you should set a 3 digits pin :D [14:10] Saviq: 10 minutes of looping and working, so do i commit that workaround with a todo? do we have time to properly investigate why this happens? given that i have no clue of that code it may take ages [14:11] tsdgeos, yeah, just commit the workaround === Estilanda_ is now known as Estilanda [14:14] Saviq: confused, i'm in tryCard, have selected "Art, header - grey background", now i go and add "card-layout": "horizontal", to the template [14:14] why is the grey background gone?¿ === MacSlow|lunch is now known as MacSlow [14:23] Saviq: commit to alt_nav or some other branch? === syst3mw0rm_ is now known as syst3mw0rm === dandrader is now known as dandrader|lunch [14:42] dandrader|lunch: please let me know when you fixed the build issues with your branches (or kick the ppa yoursef) [14:43] tsdgeos, I think one of the previous branches actually introduces this [14:43] tsdgeos, header-customizations [14:43] tsdgeos, so put it there [14:43] ok [14:44] Saviq: can't it's yours [14:44] Saviq: and i've already put it in alt_nav fwiw [14:45] tsdgeos, yeah, please undo and I'll put it in mine and remerge up [14:45] Saviq: ok [14:46] Saviq: i also have https://code.launchpad.net/~aacid/unity8/more_stable_overview_tests/+merge/231173 that we should integrate somewhen somehow [14:47] Saviq: want me to chain it after some branch [14:47] ? [14:48] tsdgeos, I think it doesn't conflict with anything else no [14:48] ok [14:48] tsdgeos, but truth be told.. [14:48] tsdgeos, maybe instead we should make tryCompareFunctions catch? [14:49] Saviq: ? [14:49] tsdgeos, right, wouldn't help in that case anyway [14:49] 'cause it'd catch and fail() anyway [14:49] tsdgeos, yeah, is good like it is now [14:49] gaaaah [14:50] and ListView.onRemove causes removeDisplaced to not happen [14:50] REALLY?! [14:52] WAAAAT [14:52] mzanetti, tsdgeos please tell me I'm blind or something [14:52] http://paste.ubuntu.com/8080481/ [14:53] remove 5, 4, 3 [14:53] Saviq: you're blind or something [14:53] fooking 3 gets a remove transition !?!?!?!?! [14:53] * mzanetti tries [14:54] :D [14:55] but if you overshoot it [14:55] it doesn't [14:55] like remove 5, 4 [14:55] yeah [14:55] drag to overshoot [14:55] just figured the same [14:55] WTF [14:56] Saviq: well, what's weird is this: [14:56] scroll to bottom, click on 5, 4 [14:56] or, well, one step back [14:56] click on 5, you'll see 4 [14:56] then click on 4, you'll see 3 [14:56] then click on 3, you'll see 0 [14:56] (unless you overshoot) [14:57] mzanetti, yeah, related to the lack of transition [14:58] looks weird yeah [14:58] mzanetti, HAH [14:58] originY [14:59]