[00:33] kgunn, no === ejat- is now known as ejat === jono is now known as Guest37503 === ubot5` is now known as ubot5 === jhodapp is now known as jhodapp|afk === jhodapp is now known as jhodapp|afk === ricmm_ is now known as ricmm === shiznix_ is now known as shiznix [07:11] Saviq: why did you approve the height/animation thing? [07:11] didn't we agree it didn't work good enough? [07:35] veebers: sorry... slept in. [07:42] tsdgeos, didn't we agree that it's good enough for now, since we don't have better ideas yet? [07:42] maybe [07:42] i did have an idea yesternight [07:42] it's not nice but it ought to work [07:43] i'll give it a try later [07:43] now busy trying to find out why pstolowski code makes stuff break [07:44] * pstolowski is keeping fingers crossed [07:45] pstolowski: it's not cool, if i add all the mirrormodels code, then it works [07:45] ouch [07:45] pstolowski: which means stuff works if i slow it down (because that's the only think mirrormodels do) [07:45] which is not good [07:45] :==/ [07:50] tsdgeos: this is weird, coz I'm running unity8 (and able to reproduce) in virtualbox with no 3d acceleration, so things run very slow there [07:50] it is [07:51] want to give the patch a try and see if it seems to fix stuff for you too? [08:10] Saviq, are we still scrambling to get that autopilot issue fixed? [08:10] dandrader, which one? [08:10] Saviq, the one that was getting in the way of having unity8-mir as the default [08:11] dandrader, AFAIK it's all better now [08:11] dandrader, thomi got it going last night [08:11] Saviq, great [08:12] MacSlow, two last comments on ext SDs https://code.launchpad.net/~macslow/unity8/extended-snap-decisions-part1/+merge/187312/comments/431919 https://code.launchpad.net/~macslow/unity8/extended-snap-decisions-part1/+merge/187312/comments/431924 [08:15] dandrader, just went through my email - and the output is https://bugs.launchpad.net/mir/+bug/1233944 [08:15] Ubuntu bug 1233944 in Mir "Unity8/Mir is unable to open autopilot uinput devices" [Critical,New] [08:18] Saviq, bummer [08:21] MacSlow, but can you confirm that's the case? that the menumodel goes away and that's why it contracts? [08:22] MacSlow, we can work around it with a ShaderEffect hack, but probably not at this point [08:26] Saviq, I'll try to do the unexporting of actions and menu-model after the quit/close-handler to see if that changes anything... [08:27] MacSlow, but you can see/reproduce? [08:29] Saviq, I think I see what you are refering too... the ext. snap-decision notification seems to "shrink upwards" before it actually disappears. [08:29] Saviq, that's what you mean? [08:29] MacSlow, yup [08:30] Saviq, I just tried to do the unexport later... but that didn't change anything in this visual behaviour [08:30] MacSlow, I wouldn't expect it to [08:30] MacSlow, anyway - that's a glitch, let's do the more important things now [08:31] Saviq, indeed... those kind of things should be addressed in one go with the push on "spit&polish" [08:31] MacSlow, the "simple" workaround I can think of is having a non-live ShaderEffect that we enable and update just before we unexport the actions [08:31] MacSlow, this will create a "snapshot" of the notification while it disappears [08:32] Saviq, I hope there's a more generic solution to this... such "special case" handling doesn't sound right... we do too much of this already [08:33] Saviq, besides... my current headache revolves around the checkmark in the wifi-selection not working, because of the Loader. [08:33] MacSlow, I don't think there is - not until we build the thing into unity-notifications, so that it will keep the structure while it's being removed [08:34] MacSlow, right [08:34] MacSlow, did you try/manage to reproduce in a small standalone QML file? [08:35] MacSlow, or otherwise pinpointed the issue? [08:36] Saviq, yeah... the Loader-item used for the submenu used for the wifi-accesspoints causes the issue... OptionSelectorDelegate fails to set its internal listView correctly, thus there are several runtime-failures with null-references... [08:37] Saviq, I don't know how to approach this in a different way [08:38] MacSlow, you mean that OptionSelectorDelegate isn't a direct delegate of OptionSelector's, but you have a Loader in the middle? [08:38] Saviq, fact is... wifi-selection does work... only the checkmark isn't showing... [08:38] MacSlow, i.e. OptionSelector { delegate: Loader { sourceComponent: OptionSelectorDelegate { } } } more or less? [08:39] Saviq, still this is a nasty issue... not giving the user a visual clue what's currently selected. [08:40] Saviq, yes... http://bazaar.launchpad.net/~macslow/unity8/extended-snap-decisions-2/view/head:/Notifications/NotificationSubMenuItemFactory.qml [08:40] Saviq, and http://bazaar.launchpad.net/~macslow/unity8/extended-snap-decisions-2/view/head:/Notifications/NotificationMenuItemFactory.qml [08:41] mzanetti, what? [08:41] mzanetti, https://code.launchpad.net/~cimi/unity8/unity8.hud-2_hint-reveal-commit/+merge/187488/comments/431739 [08:41] I think I want to drop a bomb [08:41] Saviq, are there any QML-tricks to avoid this? [08:42] Cimi: haha... yeah. autopilot [08:43] MacSlow, where is AccessPoint.qml? [08:43] mzanetti, I did what you asked [08:43] mzanetti, doesn't seem to work [08:44] Saviq, in NotificationSubMenuItemFactory.qml [08:44] Saviq, line 54 [08:44] MacSlow, no, that *uses* AccessPoint, where is that component defined? [08:45] Saviq, hold on... that's not the correct revision... [08:45] MacSlow, anyway, you just need to make sure that OptionSelectorDelegate is passed the needed properties [08:45] mzanetti, might be my fault, was damn working this morning [08:46] mzanetti, maybe I did not update the test_hud.py on the phone and I thought was working [08:46] mzanetti, anyway, you know how I can access __init.py__ class from TestHud? [08:46] Cimi: of course its your fault... it doesn't break on its own... that said, happens to me just as often. and I could start killing too when this sort of failures show up [08:46] MacSlow, you need to bind the delegate's listView in the Loader [08:48] Saviq, listView of the OptionSelector is readonly... will that work? [08:48] Cimi: __init__.py defines UnityTestCase [08:48] MacSlow, then it needs to be made non-readonly [08:48] mzanetti, otherwise I revert to the previous working revicion [08:48] MacSlow, in UITK [08:48] Cimi: test_hud.py inherits UnityTestCase [08:48] mzanetti, __init__.py inside emulators [08:48] Cimi: no, that won't work [08:48] MacSlow, but [08:48] mzanetti, I tad you yesterday [08:48] *told [08:48] maybe we misunderstood? [08:49] MacSlow, I don't think we should have that Loader at all [08:49] mzanetti, what's a common place for hud.py and test_hud.py? [08:49] MacSlow, we don't support any components other than OptionSelectorDelegate in an OptionSelector [08:49] MacSlow, not dividers, not standarditems [08:50] Saviq, yes... what other way can I use to pull the info from the subMenu-model? [08:50] Cimi: I think the main __init__.py would work [08:51] mzanetti, let me try [08:51] Cimi: in tests/autopilot/unity8/shell/__init__ [08:51] MacSlow, just include whatever's pulling that data in your OptionSelectorDelegate [08:51] MacSlow, you shouldn't use the SubMenuItemFactory at all [08:51] mzanetti, although doesn't have a class [08:52] Cimi: yeah... well, I'm really not a python expert either. you need to try or find someone who know python [08:52] mzanetti, I'm trying [08:52] mzanetti, nope [08:52] mzanetti, let me revert to 330.. [08:52] MacSlow, everything you need is passed through to the delegate [08:52] mzanetti, which btw was failing on jenkins [08:53] MacSlow, menuModel, menuData, menuIndex - those are just properties [08:53] Saviq: do you know a solution to this? [08:53] Saviq, regarding your most recent comments in the MP for the search indicator. I added a cross fade. [08:53] Maybe it's too quick though, I'll experiment a bit more, otherwise I guess we'll have to check the search time. [08:53] Saviq, how do I install a "devel-proposed" ubuntu-system image using phablet-flash? [08:53] dandrader, --channel [08:53] nic-doffay, k, will have a look, although when I tried a *real* cross fade it didn't look good, we rather need a sequential fade out + fade in [08:54] Cimi: oh... wait [08:54] Cimi: in test_hud.py we can access the emulators stuff [08:54] Saviq, I'll try that now. [08:54] Cimi: with self.main_window.xxx [08:54] nic-doffay, yeah, that's gonna be tricky without moving all that to states [08:54] nic-doffay, and I started doing that yesterday but fooked something up [08:54] Cimi: so if you put the drag() in the hud emulator you should be able to use it in test_hud with self.main_window._drag() [08:55] nic-doffay, let me get you a diff I started out wiht [08:55] let's try [08:55] ah, phablet-flash ubuntu-sytem --help mentions devel-proposed by I just didn't see it :/ [08:56] nic-doffay, something along the lines of http://paste.ubuntu.com/6183052/ [08:56] nic-doffay, but it didn't work right yet [08:57] mzanetti, File "/usr/lib/python2.7/dist-packages/unity8/shell/tests/test_hud.py", line 99, in test_show_hud_appears [08:57] self.main_window._drag(swipe_coords.start_x, swipe_coords.start_y, swipe_coords.start_x, swipe_coords.end_y) [08:57] AttributeError: 'MainWindow' object has no attribute '_drag' [08:57] mzanetti, come on let's revert and have this thing merged.. [08:57] ok [08:57] mzanetti, but was good to try [08:57] mzanetti, we're not python experts [08:59] Saviq, I fixed it. [08:59] MacSlow, see [08:59] Saviq, one more tiny bit of QML-wizardry learned today :) [09:00] MacSlow, so you just flattened the SubMenuFactory? [09:00] MacSlow, to be the OptionSelectorDelegate directly? [09:00] Saviq, yeah... pushing... one sec... [09:03] mzanetti, Cimi, what's the deal there? [09:03] Saviq, we saved jenkins from a bomb [09:03] Saviq: we need a custom drag() method in emulators and tests [09:03] Saviq: right now it's copied and I wanted Cimi to put it in some common place [09:04] Saviq: but we both failed to find one [09:04] Saviq, I wanted it too but at this time I just want the tests to pass [09:04] Saviq, there you go http://bazaar.launchpad.net/~macslow/unity8/extended-snap-decisions-part2/view/head:/Notifications/NotificationMenuItemFactory.qml [09:04] Saviq, I failed here http://bazaar.launchpad.net/~cimi/unity8/unity8.hud-2_hint-reveal-commit/revision/331 [09:05] mzanetti, Cimi you just need a small class DragMixin somewhere in __init__.py, import that and make the emulators inherit from it where needed [09:05] mzanetti, Cimi one that will contain just the drag() method [09:05] Saviq, yeah but I'm still learning python [09:06] Saviq, and I have other tasks to do! :( [09:06] Saviq, before the freeze [09:06] Saviq, once lp:~macslow/unity8/extended-snap-decisions-part1 made it into trunk I'll update the "part2" for frontend and backend and MR them, so we get the wifi-selection use-case in place too [09:06] Cimi, sure [09:06] MacSlow, yup, cool beans [09:06] Saviq, was good to learn a bit but now I really want to move forward [09:07] mzanetti, just tested again all hud tests with rev 330 and they pass [09:07] ok. lets see what jenkins thinks of it [09:07] it's pushed as rev 332 [09:07] jenkins would better think he likes [09:07] lol [09:07] I might become very upset [09:08] mzanetti, Cimi something of the sort v [09:08] http://pastebin.ubuntu.com/6183074/ [09:09] Saviq: is the "Mixin" just some common thing to do or does it have a special meaning? [09:10] mzanetti, just a name [09:10] ok [09:10] mzanetti, you usually combine multiple mixins to get a union of multiple parent classes [09:10] mzanetti, that's simply multi-inheritance [09:11] dednick, minor thing here https://code.launchpad.net/~nick-dedekind/unity8/indicator-titles/+merge/186049/comments/431233 [09:11] Saviq: yeah, got that. Just noticed that I see the name "Mixin" quite often so I was wondering if it actually _has_ to be named like that [09:11] mzanetti, python is all about conventions :) [09:12] Saviq, let me try [09:12] Saviq, I'm not giving up [09:12] dednick, also, any input on https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1215644 ? [09:12] Ubuntu bug 1215644 in unity8 (Ubuntu Saucy) "Messaging menu does not change icon when a new message appears" [High,Confirmed] [09:13] Saviq: just upgrading phone now. I think it's fixed and released already. [09:13] dednick, ah ok [09:13] dednick, although I had a blue icon with no message this morning :D [09:14] Saviq: er. yeah. i dont seem to be seeing any messages at the moment :( [09:14] dednick, yeah, no messages in the messaging menu [09:14] dednick, I just got the notification, but nothing in the menu [09:14] larsu, ↑ [09:16] fginther, hey there... [09:16] MacSlow, he's asleep, can I help? [09:16] mzanetti, issue https://code.launchpad.net/~mzanetti/unity8/switching-previews/+merge/187327/comments/431952 [09:16] Saviq, the autopilot-issue on jenkins seems fixed... what was the issue? [09:16] Saviq, just curious [09:16] MacSlow, we knew the issue straight away [09:16] MacSlow, it was multi-fold [09:17] Saviq: looking [09:17] MacSlow, a) we started depending on libunity-core-6.0-8, which depended on libunity-protocol-private0, which required, but didn't depend on, a new libunity9 [09:18] MacSlow, so our two dependencies got upgraded but libunity9 didn't, and we were killed because of a missing gsettings key that was introduced recently [09:19] Saviq, ha... so basically one-line entry missing to cause all the trouble?! :) [09:19] MacSlow, b) the machines (hardware, not VM) running the tests were not upgraded recently, but they do have unity7 installed, so the above occurred only on the hardware, as VMs are bare metal, and the new libunity9 was installed as usual [09:20] MacSlow, c) when Francis switched us temporarily back to VMs, where the issue wasn't a problem - the jobs weren't compatible or something - and were failing to pick up the artifacts - so all the child jobs were SUCCESS, but the parent job still FAILED [09:21] MacSlow, d) there was a huge build queue on jenkins yesterday, that only cleared up late in the evening, which meant that some of the -ci / -autolanding jobs before that took up to 4 hrs [09:21] MacSlow, so yeah, *EVERYTHING* happened [09:21] Saviq, holy crap :) [09:21] Saviq: hmm... I do see an issue, but I'm not sure if its the same as you describe [09:22] Saviq: indicator-messages seems to be crashing [09:22] larsu: ^ [09:22] Saviq: this is the other height animation patch https://code.launchpad.net/~aacid/unity8/enable_height_animation_2/+merge/188785 [09:23] Saviq: it works better, but it's a bit nastier [09:23] tsdgeos, k thanks [09:23] i'll let you decide what we do :D [09:23] i think this one makes more sense at this stage since the other doesn't really work [09:23] but oh well [09:24] i'm not happy with this one either :D [09:24] Saviq, thanks, it works (cc mzanetti) [09:25] ~o/ [09:25] MacSlow, I think that's a valid failure: http://s-jenkins:8080/job/generic-mediumtests-runner-maguro/1760/? [09:25] MacSlow, due to moving the PinPad to components [09:26] mzanetti, http://bazaar.launchpad.net/~cimi/unity8/unity8.hud-2_hint-reveal-commit/revision/333 [09:26] Cimi: +1 [09:26] MacSlow, ah, why did you move PinLockscreen.qml at all? [09:26] MacSlow, that should remain in Greeter, just the PinPad.qml should be moved [09:27] MacSlow, that's probably my fault? [09:27] Saviq, you told me to [09:27] and PinPadButton.qml probably [09:27] Saviq, but the failing test pass here locally with that branch... [09:27] MacSlow, you need to take me with a grain of salt ;) [09:27] MacSlow, /me tries [09:27] Saviq, ok :) [09:28] Saviq, yup... both pass just fine [09:29] MacSlow, fail here [09:29] MacSlow, remove builddir/install [09:29] MacSlow, make -C builddir install [09:29] MacSlow, PYTHONPATH=tests/autopilot/ autopilot run unity8.shell.tests.test_lock_screen.TestLockscreen.test_can_unlock_pin_screen [09:30] MacSlow, and make sure you don't have unity8 installed on your system - it will test that instead [09:30] MacSlow, but anyway Greeter/PinLockscreen.qml: File not found [09:30] Saviq, so next in priority [09:30] MacSlow, that's a quite obvious error message ;) [09:31] Saviq, welcome wizard or rendered for weather, dash plugins, or carousel fix, video renderer? [09:33] Cimi, do the dash plugins renderer [09:33] Cimi, should be simples [09:33] mhr3, can you give Cimi a way to determine that a category is dash plugins? [09:33] mhr3, I'm coming to the office in an hour and a bit [09:34] Cimi, i'm not in today [09:34] mhr3, ok thne [09:34] mhr3, so my desk is free :P [09:34] tsdgeos, aaargh ;( [09:34] Saviq, do I still need this ppa on my desktop? http://ppa.launchpad.net/ubuntu-unity/next/ubuntu [09:34] Cimi, yea, let's say that i knew that you want to come :) [09:34] hah [09:34] dandrader, *no* [09:35] dandrader, that's dangerous :) [09:35] Saviq: yeah :D [09:35] Cimi, but yea, i need to expose one extra property for that to work [09:35] hehehe, makes me wonder how I managed to survive so far [09:36] tsdgeos, can we at least make the onDestruction less ugly? [09:37] Saviq: ideas on how to do it? [09:37] tsdgeos, like put some prop on RunningApplicationsGrid instead of looking at the source path? [09:37] Saviq: well, it's wath onLoaded does [09:37] tsdgeos, yeah, but I mean for the: [09:37] if (source.toString().indexOf("Apps/RunningApplicationsGrid.qml") != -1) { [09:37] tsdgeos, if RunningApplicationsGrid had a prop holding [09:37] (results.firstModel.count + results.secondModel.count == 0) [09:38] tsdgeos, we could use hasOwnProperty("blah") [09:38] tsdgeos, and use that prop; else true; [09:38] dednick: my fixes from last Friday haven't been released yet [09:38] larsu: ah ok, i havent taken a look yet [09:38] larsu, dednick, I'll build a package to test, then [09:38] Saviq, or the renderer could use regular scope models and we could get rid of that special case :) [09:39] Saviq: i'll see what i can do [09:39] mhr3, got the scope for me yet? [09:39] Saviq, it's been in the image for a few weeks now [09:39] dednick, Saviq: I'm pretty sure this fixes the issue. If not, let me know ;) [09:39] mhr3, and working? with surfaceflinger? ;p [09:39] question is: does it work? :) [09:39] (well, I'm sure you will) [09:40] larsu, Saviq: i need to test out some other odd behaviour i'm seeing as well. The icon still doesnt seem to be changing when recieving/removing items still. (looking at the data from the model here) [09:40] pstolowski, were you able to test runningapps scope on the device? [09:40] pstolowski, does it return anything now? [09:40] larsu: also, the sensitivity of the clear button is set to false. [09:40] mhr3: no, I plan to do today [09:40] dednick, ok, I'll get you the package from i-m trunk [09:40] Saviq: ta [09:41] guys, little guide to run mir on the phone? [09:41] I promised kevin I'd have tested the new hud on the phone with mir [09:42] Cimi, adb shell touch ~phablet/.display-mir; adb shell reboot [09:42] Saviq, brilliant, thanks [09:43] dednick: I'll have a look (but I think I remember that working last Friday) [09:45] larsu, if I have to install one more package on the host to just build a source package out of indicator-messages! you mark my words! [09:46] Saviq: huh? [09:46] larsu, had to install dh-autoreconf, dh-translations, gobject-introspection [09:46] larsu, just to build a source deb :P [09:47] Saviq: you need these for 80% of our desktop projects... also: learn about apt-get build-dep :P [09:47] larsu, I don't want that [09:47] larsu, am using sbuild to keep the host clean [09:47] well, you need to install all deps manually then [09:48] larsu, didn't have to install anything for indicator-network, unity8, unity-notifications, mir, platform-api, unity-mir... and then some [09:48] larsu: it's easier to see what's going on using the indicators-client app on the phone. It gives you a textual representation of the unitymenumodel row data. Although the current release is not working. need trunk or even better: https://code.launchpad.net/~nick-dedekind/unity8/indicators-client.actionStateParser/+merge/188790 [09:49] Saviq: follow up on yesterday night: this works: http://paste.kde.org/p8ce24ba6 [09:49] notice line 8 [09:49] mzanetti, HUH! [09:49] removing that breaks it [09:49] Saviq: because you already had the stack for those installed... [09:49] larsu, no, because everything happens *inside* sbuild there [09:50] larsu, building the source package for those is just a case of putting stuff in the tarball and .dsc file [09:51] greyback: you around yet? [09:51] larsu, for -messages, there's autoreconf, translations and stuff that happens during building of the source package, which happens outside of the chroot [09:51] mzanetti: yes [09:51] Saviq, wow the hud button doesn't work on mir [09:51] Cimi: heh, yep, known issue. I'm on it [09:51] Cimi, racarr was working on that yesterday [09:51] ah ok [09:51] in Mir that is [09:52] Cimi: yeah... racarr provides me an API to re-inject the mouse event's we're filtering [09:52] mzanetti, although I can't say I get what the issue is - the fact that we want the input in both app and shell? isn't that the case everywhere? [09:52] mzanetti, I don't get why that would be the case (re LazyImage) [09:52] mzanetti, I'm not touching "height" anywhere there [09:53] mzanetti, and anyway the tests *do* test the height [09:53] mzanetti, or width, if scaleTo: "height" [09:53] greyback: ping [09:53] pstolowski: pong [09:53] greyback: morning! [09:53] Saviq: let me push, you can try with that use case if you want [09:53] pstolowski: yes :) [09:53] Saviq, so now it should be ok [09:54] greyback: is this https://lists.launchpad.net/ubuntu-phone/msg04088.html the only thing needed to get mir on the phone? how do I know if it's running (process name?) [09:54] MacSlow, k thanks, and sorry for pushing you in the wrong direction ;) [09:54] mzanetti, it's not about trying ;) I believe you that's your issue - just asking whether you have an idea why that happens? [09:54] Saviq: pushed. [09:55] Saviq: no, not really [09:55] mzanetti, height: implicitHeight should be.... implicit, shouldn't it? [09:55] pstolowski: yes, that's all you do. To know it worked, does unity appear? Then check surfaceflinger is not running - then it must be Mir [09:55] mzanetti, to where? :D [09:55] Saviq: I guess it should, yeah [09:55] Saviq: lp:~unity-team/unity8/fix-genericpreview [09:55] mzanetti, k [09:56] mzanetti, re: switching-previews, you really think it's related to the header? why? [09:57] Saviq: the only issue I can see is if you're opening a preview on an item which is partly covered by a header [09:57] mzanetti, no, that's not it [09:57] Saviq: couldn't find any issue with the bottom [09:57] mzanetti, expand "Installed" in Apps [09:57] Saviq: ah I see what you mean. That's the age-old decision that you don't need anything but a shell, make, and a compiler to build autotools-based packages [09:57] btw everytime i try to build unity8 pkg i get a failure in launchermodeltest, known? [09:57] mzanetti, open a preview on the last visible row, and go back to the beginning [09:58] mhr3, no [09:58] oh... [09:58] mzanetti, same with the other way [09:58] greyback: yeah, unity still works, I was just expecting performance drop on maguro, but can't see real diff... no surfaceflinger process running [09:58] mzanetti, open a preview in the first row of the *last* category [09:58] mzanetti, and go towards the end [09:58] mzanetti, it will get confused [09:58] mzanetti, but it's no surprise [09:59] Saviq: right... I got it [09:59] mzanetti, if the ListView is open near the top - you can't scroll it enough to reach the bottom [09:59] mzanetti, you need to move the gab [09:59] gap [09:59] pstolowski: that's interesting, I must flash my maguro too to see. There was a perf drop on it, possibly fixed [09:59] Saviq: aye.have to go to a meeting now. will see if I can fix it afterwards [09:59] mzanetti, o/ [10:00] larsu, huh... http://pastebin.ubuntu.com/6183192/ [10:01] Saviq: oh wow, why didn't CI catch that? [10:01] larsu, indeed [10:02] MacSlow, hmm hmm hmm, why does notifications need to import Greeter? [10:02] MacSlow, does it use PinLockscreen? [10:02] MacSlow, and not just the PinPad? [10:02] Saviq, yes [10:04] MacSlow, ok, I need a little bit more looks at that then... [10:04] larsu: i think the remove-all action is coming across with the incorrect action. it's showing up in unitymenumodel as "indicator.indicator.remove-all" [10:05] MacSlow, feels like we should rename PinLockscreen... and yes, move it into Components, then... [10:05] dednick: give me a sec to fix that compile issue (and find out why CI didn't catch it) [10:05] Saviq, are you sure :) [10:05] MacSlow, no - not yet ;d [10:05] larsu: which is why it's not sensitive i think. I'm getting the dednick log messages saying Enable remove-all. [10:05] mzanetti, should the SIM PIN snap decision be using PinLockscreen directly? [10:05] larsu: ok [10:05] MacSlow, will let you know when I am ;) [10:05] Saviq: yes [10:05] Saviq, I switched up the search indicator to use states and transitions and sequential, however I'm not sure how effective it was. === mzanetti is now known as mzanetti|otp [10:06] "dednick log messages" [10:06] the best kind of log mesages. [10:06] *messages [10:06] lol [10:06] nic-doffay, k, will have a look (or mzanetti will, while I'm looking at his stuff) [10:06] *debug [10:07] MacSlow, ok, let's leave as-is for now, if CI goes through, I'll approve, we can move around later [10:08] Saviq, ok [10:09] larsu, seems it's built fine out of sbuild... *weird* [10:10] Saviq: did you override -Wall or something? [10:10] larsu, sbuild might [10:11] Saviq: it definitely should be on though. I don't know why it isn't and this is a very legitimate bug [10:11] larsu, yup [10:12] mzanetti|otp: ping [10:14] Saviq: https://code.launchpad.net/~aacid/unity8/enable_height_animation_2/+merge/188785 less ugly? [10:18] tsdgeos, why do we need to compare the source? [10:18] tsdgeos, can't we have a property *in* RunningApplicationsGrid [10:18] tsdgeos, that GenericScopeView looks for/at? [10:19] Saviq: isn't that the same that comparing the source? [10:19] larsu: fyi, indicator-messages r380 did in the remove-all action. [10:19] tsdgeos, it's not, as we can rename the grid, we can add the same behavior to other renderers etc. [10:19] tsdgeos, and it will work - now it's filename-specific [10:19] totally yep [10:20] dednick: you're missing a verb there [10:20] tsdgeos, and also, do the "firstModel.count + secondModel.count" *in* RunningApplicationsGrid === mpt_ is now known as mpt [10:20] larsu: na, speakin' street yo. [10:21] dednick: hah :) [10:21] Saviq: you're aware that this also needs the enableHeightBehavior in DashApps that is what makes it persistent over the destructions of RunningApplicationsGrid and that making it generic is not going to get us much really' [10:21] i can do it [10:21] just don't see how is going to help [10:22] tsdgeos, it's going to help me sleep at night ;) [10:22] ok [10:22] that's a good reason enough [10:23] greyback: I've just tried to running-apps scope on the phone with mir, and it doesn't find com.canonical.Unity.WindowStack. is it available? [10:24] pstolowski: org.canonical.Unity.WindowStack [10:24] greyback: ah, thanks [10:25] greyback, we're canonical.com, not canonical.org :) [10:25] :) [10:25] mhr3: patches welcome :P [10:26] Saviq: -Wall is set but I still don't get that error. Anyway, fix is here: https://code.launchpad.net/~larsu/indicator-messages/set-status-return-value/+merge/188803 [10:26] larsu, tricky :) [10:26] Saviq: ya... I get the warning if I run the preprocessor manually :-( [10:26] I'll try to figure this one out - but not now [10:27] mzanetti|otp, Approve! [10:28] Saviq, we need faster reviews, need to create 4th unity8 dir cause the other 3 are still not merged :P [10:28] mhr3, use colocated branches [10:29] * mhr3 heads off to read bzr docs === mzanetti|otp is now known as mzanetti [10:34] tsdgeos: pong [10:35] Cimi: \o/ [10:37] mzanetti: http://paste.ubuntu.com/6183307/ otherwise we assert on debug builds [10:37] mzanetti: me or you? [10:37] dednick, http://people.canonical.com/~msawicz/indicator-messages/ [10:37] Saviq: thanks [10:37] tsdgeos, that mhr3's issue? [10:37] tsdgeos: oh... thanks for pointing it out [10:38] Saviq: no clue what's mhr3 issue :D [10:38] Saviq: where do i read about it? [10:38] tsdgeos, mhr3 had some test failure in that test [10:38] tsdgeos, or, a test for that [10:38] I *think* [10:38] but well, that was not for debug builds [10:38] not sure if that has any real world impact [10:38] as it was during package build [10:38] debug build of qt i mean [10:39] that has all the asserts enabled [10:39] dednick, larsu, the icon did *something* on incoming text, but it did not get blue... [10:40] Saviq: yeah, it's still not working properly for me either. [10:40] dednick, ok, you're on it so it'll be fixed in 5 I'm sure [10:40] putting some trace logging in now to figure it out. [10:41] lol. um, probably with another bug built in. but no, i dont understand much in libmessaging unfortunately. [10:41] the icon choosing is fairly complicated. [10:42] dednick: leave those bugs to me. No need to waste your time on them [10:42] dednick: as for the indicator.indicator.remove-all. Are you sure you're not adding one yourself along the way? [10:42] larsu: ok :) [10:42] I changed it in r380 to be consistent with all the other actions [10:42] i.e., in the menu, actions are referred to as indicator. [10:42] larsu: you added an "indicator." onto the menu action in r380 [10:43] larsu, http://pastebin.ubuntu.com/6183316/ :D [10:43] dednick: yes. that's what I just said... [10:43] :) was busy typing :) [10:44] Saviq, my issue is http://paste.ubuntu.com/6183319/ btw [10:44] larsu: erm, i'm just looking at the action role on the unitymenumodel. [10:44] mzanetti, ^^ [10:44] tsdgeos: I'm actually wondering if it shouldn't be this: http://paste.ubuntu.com/6183320/ [10:44] mhr3, mzanetti I *think* we're interrogating the real AccountService there are we not? [10:44] Saviq: cool. As I said, I haven't yet managed to get these errors :( [10:45] Saviq: I'll fix them as soon as I've figured out why [10:45] Saviq, ehm... don't do that? [10:45] mzanetti: actually, yes [10:45] mhr3, orly? [10:45] yep :P [10:45] mzanetti, ↑ [10:45] yes... [10:45] mzanetti: fixit! :D [10:45] mhr3, gdbus call --system --dest org.freedesktop.Accounts --object-path /org/freedesktop/Accounts/User1000 --method org.freedesktop.DBus.Properties.Set com.canonical.unity.AccountsService launcher-items "<[{'defaults' : }]>" [10:46] mhr3, assuming you're UID1000 [10:47] larsu: i still don't gel well with the namespace thing, so I dont know. the unitymenumodel gtk code seems to do some interesting things with namespaces when it comes to sections [10:47] Saviq, but, but, i don't want to loose my laucher items [10:47] mhr3, you won't *lose* them, you'll just reset them ;d [10:48] ah, ok [10:48] that makes me feel so much better [10:52] dednick: ah, it's indeed my fault: there's a namespace set in addition to the "indicator.remove-all" name. The question is: why does this work on the desktop... [10:52] bug 1234054 [10:52] bug 1234054 in unity8 (Ubuntu) "App preview does not work in the home scope" [Low,New] https://launchpad.net/bugs/1234054 [10:52] dednick: I'll remove it. [10:52] any takers/responders ? [10:53] om26er, right [10:53] om26er, it's the same reason - there's no scope behind it [10:53] Saviq, ok, understood. [10:53] Saviq, how about bug 1227595 ? [10:53] bug 1227595 in unity8 (Ubuntu) "[touch] The screen brightness change is not realtime" [Low,Confirmed] https://launchpad.net/bugs/1227595 [10:54] dednick, ↑should be an easy fix [10:54] I just changed one line live: false to live:true on my phone and things are working fine [10:54] om26er, yeah, exactly [10:54] om26er, so, you proposing an MP? [10:54] that also fixed the problem for the sound slider [10:54] Saviq, sure. [10:54] dednick, unping [10:55] mhr3, do we have a plan for apps in home scope? will they come from a scope somewhere? [10:56] dednick, reping https://code.launchpad.net/~nick-dedekind/unity8/indicator-titles/+merge/186049/comments/432038 [10:57] Saviq, step 1) we're going to get rid of the apps scope [10:57] step 2) figure out if click scope is going to provide those results [10:58] dednick: https://code.launchpad.net/~larsu/indicator-messages/remove-stray-namespace/+merge/188809 [10:58] larsu: btw, I had already put the trace logging in and built, so i just ran it up quick. Here's the output to help you on your way with the icon bug. http://pastebin.ubuntu.com/6183346/ [10:59] dednick: ah cool thanks. So the icon doesn't change to "new" when it should? Or it doesn't change back to "normal"? [10:59] larsu: it changes to new, then straight back to normal [11:00] get a message_added, then a source_changed [11:01] okay, I'll look into it right after lunch [11:02] tsdgeos: https://code.launchpad.net/~mzanetti/unity8/fix-quicklist-inserting/+merge/188811 [11:07] Saviq: https://code.launchpad.net/~aacid/unity8/enable_height_animation_2/+merge/188785 better? [11:07] tsdgeos, http://pastebin.ubuntu.com/6183391/ [11:08] Saviq: any more backtrace? [11:08] https://code.launchpad.net/~om26er/unity8/slider_live/+merge/188810 [11:08] tsdgeos, not atm, trying again [11:08] Saviq: that doesn't help much, there's lots of calls to ListViewWithPageHeader::ListItem::y [11:08] tsdgeos, k, trying [11:09] need to know where to see why the listitem was null [11:09] since that should never happen [11:10] mhr3: https://code.launchpad.net/~mzanetti/unity8/dont-load-as-for-testing/+merge/188812 [11:11] tsdgeos, that's positionViewAtBeginning again [11:11] tsdgeos, http://pastebin.ubuntu.com/6183404/ [11:12] tsdgeos, so the un-revert from yesterday... [11:12] and why nobody could make it crash yesterday? :D [11:12] tsdgeos, steps to repro: start shell, start app, swipe from left to get back to dash apps straight away [11:12] tsdgeos, i.e. when app is launching === jhodapp|afk is now known as jhodapp [11:13] i did that :/ [11:13] but yeah [11:13] let's try again [11:13] tsdgeos, 100% reproducible here [11:13] mzanetti, sorry already did a reset, so test build wouldn't fail now... but the code looks ok === iahmad_ is now known as iahmad|afk [11:14] mhr3: yeah... this functionality has been in there already for the launcherbackendtest. but I don't manually instantiate the backend when doing the "frontend" tests. [11:15] my Phone is currently in this state. i.e. I try to change to switch to dash Home the app comes back automatically: http://videobin.org/+6xp/8lp.html [11:16] that video opens in firefox only [11:16] om26er: I've seen this too. can you report a bug please? [11:16] Saviq: grrr, ok [11:16] mzanetti, I was wondering if there any logs that I could provide ? [11:16] tsdgeos, got it? [11:16] no, but i can see why the code may be wrong [11:16] let me try to run it [11:17] om26er, steps to reproduce would be best - logs are rarely useful for us [11:17] om26er: hmm. not sure if ~/.cache/upstart/unit8.log provides something useful in this case. you can still attach it you think there are some hints in there [11:17] Saviq, I am afraid there are no obvious steps but I think music-app was running in the background before it. [11:17] will try to come up with steps to reproduce the issue [11:17] om26er, I know, as usual [11:17] mzanetti, ok [11:18] Saviq: you start app from home or from app scope? [11:18] tsdgeos, home === MacSlow is now known as MacSlow|lunch [11:19] tsdgeos, start from home, and while the white rect comes in, I drag from left [11:19] nah [11:19] all is fine here [11:19] :( [11:20] let me try with SF [11:24] nothing :/ [11:26] bug 1234108 [11:27] bug 1234108 in Unity 8 "Swiping away to dash brings back the app automatically" [Medium,New] https://launchpad.net/bugs/1234108 [11:28] om26er, thanks [11:29] greyback: having an issue with winstack api; didn't we agree it would provide app names that map directly to .desktop files? it gives me "Browser" (app name), but the actual desktop file is webbrowser-app.desktop [11:30] mzanetti, trying your previews branch on the phone [11:30] pstolowski: we did. That's a mistake [11:30] * mzanetti wonders if this is the beginning of a bugreport or just a FYI [11:30] thought I fixed that [11:31] mzanetti, usually the lazyimage shows the "no-image" state as semi-transparent box, but sometimes it doesn't show anything [11:31] Saviq: do you happen to have a loner backtrace of that crash? [11:31] mzanetti, fixable? [11:31] tsdgeos, longer? [11:31] tsdgeos, will try and get more symbols [11:31] mhr3: can you define the "sometimes" a bit better? [11:31] Saviq: want to know if it's directly coming from qml or reentring from the lvwph code [11:32] not sure how it can happen yet [11:32] mzanetti, i see it if i preview twitter [11:32] greyback: can you fix that soonish? [11:32] pstolowski: in progress... [11:33] mzanetti, or rather... don't see it :) [11:33] greyback: thanks! [11:33] mhr3: hmm... I just saw that with trunk too [11:33] mhr3: will check [11:34] bug 1232381 [11:34] bug 1232381 in unity8 (Ubuntu) "Home Scope: Apps should be shown first" [Medium,Confirmed] https://launchpad.net/bugs/1232381 === iahmad|afk is now known as iahmad|afk|reall [11:35] currently looks like this: http://i.imgur.com/gWCvF7h.png [11:35] mzanetti, also, movie previews don't have the lazy image? [11:36] Saviq: if i can't make it crash do you think we can make some peer-debugging (i.e. i add some debugs and you tell me what you get)? [11:36] mhr3, is that ^ something to be fixed in the backend ? or unity8 needs to take care of that ? [11:36] tsdgeos, sure [11:36] ok, let me try to prepare a branch [11:36] om26er, it's being worked on [11:37] mhr3, ack [11:44] Hey guys I've been testing a n-m fix for 3g. This has meant I've had 3g everytime I have booted up the phone. Out of 10 reboot 2 have shown click packages for install and online scopes options on the apps scope is there any info I can get that would be useful to help with this? [11:46] Saviq: lp:~aacid/unity8/positionAtBeginning_crash_debug [11:47] pstolowski: I'll change org.canonical to com.canonical too, ok? [11:48] tsdgeos, sec, just getting more symbols [11:53] greyback_, yes pls [11:55] greyback_: +1 === hikiko is now known as hikiko|lunch [11:58] pstolowski: https://code.launchpad.net/~gerboland/unity-mir/dbus-names-wrong/+merge/188820 [11:59] tsdgeos, http://pastebin.ubuntu.com/6183551/ - dunno where to get more [12:00] tsdgeos, running your debug mode now [12:03] Saviq, could you look at https://code.launchpad.net/~unity-team/unity8/scope-isactive/+merge/188731 once you have a sec? [12:03] mhr3, yeah, I know I still need to fix the test there [12:03] should be familiar :) [12:03] and not yet sure how [12:03] right === iahmad_ is now known as iahmad| === iahmad| is now known as iahmad [12:05] tsdgeos, http://pastebin.ubuntu.com/6183566/ [12:10] mzanetti, I added the steps to reproduce bug 1234108 [12:10] bug 1234108 in Unity 8 "Swiping away to dash brings back the app automatically" [Medium,New] https://launchpad.net/bugs/1234108 [12:10] please give it a try if you can [12:10] om26er: awesome, thanks, I'll check it out [12:11] we should probably have an autopilot test for that when the bug is fixed. [12:11] dandrader: if you're not already, would you mind reading up & trying to help debug https://bugs.launchpad.net/mir/+bug/1233944 [12:11] Ubuntu bug 1233944 in Mir "Unity8/Mir is unable to open autopilot uinput devices" [Critical,New] [12:12] kgunn, have been on that since I start work this morning. check my comments there [12:12] awesome! dandrader ...love it when you guys are ahead of me === alan_g is now known as alan_g|lunch [12:23] Saviq: yeah... I know about that thing in the carousel [12:29] mzanetti: From comments I gather you are working on that fix-genericpreview branch? Looking forward to it. [12:38] it's such an awesome branch :) [12:46] karni: yes, I am [12:46] great === dandrader is now known as dandrader|lunch === hikiko|lunch is now known as hikiko [12:57] Saviq: VisualDataModel::item: index out range 0 0 is really bad [12:57] hmmm [12:57] ah wai [12:57] Saviq: do you have empty scopes? [12:57] mzanetti: are you fixing the testing on fix-genericpreview branch? or I'll fix it? [12:57] tsdgeos, dunno [12:57] i mean scopes without items? [12:57] tsdgeos, can't reach them [12:58] that may very well be the case [12:58] tsdgeos, ah well, yeah [12:58] tsdgeos, music and video [12:58] stupid me :D [12:58] paulliu: what's wrong with it? [12:58] * tsdgeos hits himself [12:58] Saviq: fighting with that carousel... it looks really bad when not forcing the currentHighlight to be centered [12:58] mzanetti: the actions test should moved to test of GenericPreview. And apps/movie tests other stuff. [12:59] paulliu: I can fix that === alan_g|lunch is now known as alan_g [12:59] mzanetti: ok [13:00] mzanetti, I thought we can get away with "select an item in $renderer; scroll vertically to match $gap (or move $gap if needed); move the pointer to $highlightedItem.center" === _salem is now known as salem_ [13:03] Cimi, you'll want lp:~mhr3/unity8/expose-renderer-hint the dash plugins category will have rendererHint set to "toggled" (rendererId: "grid", contentType: "apps") [13:05] Saviq: ping [13:05] pstolowski, pong [13:07] Saviq: hey! I'm testing and fixing some outstanding issues in running-apps scope (with mir's winstack api on the phone), so will probably be ready to land it soon; will you be able to implement rederer/ fix the Dash to actually use it? [13:07] pstolowski, yeah, we should be good [13:08] Saviq: ok, great [13:14] greyback_: thanks [13:16] greyback_, Saviq : one more thing re running apps scope - the scope currently returns "image://screenshot/" for thumbanils; is this still correct? it doesn't produce any thumbnail at the moment, I'm guessing it's a renderer TODO? [13:18] pstolowski, yeah, there's no image provider for it yet, I think [13:19] pstolowski, it needs to use ApplicationImage [13:19] Saviq: is this on anybody's plate? [13:19] Saviq: pstolowski there is an image provider in unity-mir. ApplicationImage is just a wrapper around it [13:20] greyback_, ah, then is image://screenshot/appid good? [13:20] Saviq: yes [13:20] ok, then we'll need to debug [13:20] pstolowski, can we enable the scope ~easily for testing? [13:21] Saviq: yes. just disable current application.scope mapping in ScopeDelegateMapper. And then you need the fix in Mir which greyback_ just implemented [13:22] pstolowski: also please keep eye on the unity8 output. You might see messages from ApplicationScreenshotProvider that are relevant [13:22] Saviq: https://code.launchpad.net/~aacid/unity8/empty_lvwph_crash_positionAtBeginning/+merge/188830 [13:22] Saviq: the scope has already been present on the image for quite some time, just to really active [13:22] s/to/not/ [13:23] pstolowski, yeah, cool [13:27] tsdgeos, happroved [13:27] greyback_: yeah, I think something is misbehaving http://pastebin.ubuntu.com/6183829/ [13:29] pstolowski: weird [13:29] greyback_: plus I copied webbrowser-app.desktop to Browser.desktop to circumvent the problem with winstack api giving me appnames, and this won't work with screenshots provider I guess [13:30] greyback_: the existing implementation of running apps in the shell uses some other way of obtaining screenshots? [13:30] pstolowski: yeah, screenshot provider wants app id. You were accidentally getting the app name. [13:30] pstolowski: same way [13:31] greyback_: does it, by any chance, expect .desktop extension? [13:31] via this ApplicationImage component; http://bazaar.launchpad.net/~mir-team/unity-mir/trunk/view/head:/src/modules/Unity/Application/OSKController.qml [13:31] pstolowski: nope, appId only [13:32] greyback_: hmm, so at least those apps from my pastebin should work [13:32] pstolowski: sorry, bad link: http://bazaar.launchpad.net/~mir-team/unity-mir/trunk/view/head:/src/modules/Unity/Application/ApplicationImage.qml [13:34] greyback_: this is matches scope's code http://bazaar.launchpad.net/~unity-team/unity-lens-applications/trunk/view/head:/src/running-apps-search.vala#L163 [13:36] pstolowski: yeah, that log even gives a correct URI. [13:37] pstolowski: I've one thing to do, then I'll be able to investigate. [13:39] pstolowski: if you want, you could recompile unity-mir, adding debug to src/modules/Unity/Application/applicationscreenshotprovider.cpp - requestImage [13:40] Mirv: ping [13:41] nic-doffay, can you hear us? [13:41] nic-doffay, your turn === MacSlow|lunch is now known as MacSlow [13:43] Cimi: you in office tomorrow? [13:43] greyback_: trying [13:43] dednick, nope [13:43] dednick, friday? [13:43] Cimi: I've got to come in tomorrow. [13:43] Cimi: I could maybe on friday [13:44] dednick, I have my ocado delivery tomorrow :) [13:44] lol [13:53] can anyone here point me to the unity window switching API? I believe we will need to use it to keep HUD working when we switch to Mir [13:53] I'm told this is a DBus API (and I'm guessing that Unity exports it, but I could be wrong) [13:57] mhr3, can you drag the XDA screenshots with fix-screenshots all the way to the right? [13:57] pete-woods, try unity-mir? [13:57] mzanetti, ↑↑? [13:57] mterry: thanks, will have a look [13:58] Saviq, sorry, reflashed the phone 10minutes ago [13:58] mhr3, nw [13:58] Saviq: what? [13:58] mzanetti, if you open the preview for XDA (the first on in recommended) [13:58] mzanetti, with fix-genericpreview branch, that is [13:58] Saviq: yep [13:59] mzanetti, the list of screenshots gets confused [13:59] does it? looks good here [13:59] mzanetti, I can't reach the far right edge [13:59] hmm [13:59] mterry: looks like that's it :D (http://bazaar.launchpad.net/~mir-team/unity-mir/trunk/view/head:/src/modules/Unity/Application/dbuswindowstack.h) [13:59] mzanetti, I can overshoot and then it goes back some 5 GU or so [13:59] mhr3, Is ZG seeded for touch? [13:59] Saviq, fwiw i did test that and it worked fine for me [13:59] tedg, no [13:59] Saviq: not happening here [13:59] mzanetti, I think it gets confused with the other images getting unloaded [14:00] yep, seems so [14:00] Saviq: it has some weirdness here too [14:00] Saviq: but not the one you describe [14:00] mzanetti, if you overshoot more [14:00] Saviq: anyways, should I bump the cacheBuffer? [14:00] mzanetti, it will destroy the first item [14:00] tedg, chicken and egg problem, zg not there cause nothing uses it, nothing uses it cause it's not there... :/ [14:00] Saviq: yep, seeing it know [14:00] mzanetti, yeah, I'm thinking... probably [14:01] mhr3, upstart-app-launch uses it. [14:01] mzanetti, as it's really a corner-case scenario (just over ListView.height items that are changing dimensions) [14:01] Saviq: it'll be destroyed anyways on any other interaction... so not a memory problem I'd say [14:01] mzanetti, yeah [14:01] mhr3, We gen an error if it isn't :-) [14:01] Saviq: ack, will fix [14:01] mzanetti, also, the initialWidth on them seems to low? [14:01] mhr3, And I thought the recent apps scope used it? [14:01] mzanetti, our devices generally have 16x10 or so? [14:01] Saviq: ok, will check that too [14:01] mhr3, How else do you know recent apps? [14:01] mzanetti, thanks [14:02] tedg, i was told things trying to push data to it isn't a reason to seed it [14:02] Saviq: mind noting it down in the MR? [14:02] tedg, shell does, we don't [14:02] mzanetti, will do [14:02] Saviq: kinda in the middle of fixing other stuff in this branch [14:02] mzanetti, yeah [14:02] *cough*bullshit*cough* [14:03] … [14:05] mhr3, no MP for this? https://code.launchpad.net/~mhr3/unity8/expose-renderer-hint [14:06] Cimi, whoops [14:06] mhr3, remove the console.log [14:06] mhr3, when you propose [14:06] Saviq, strange thing happening there. [14:06] whoops #2 [14:06] mhr3, but thanks for putting it ;) [14:06] It doesn't look like the SequentialAnimation happens. [14:06] But I threw in a ScriptAction to print at the end of it. [14:06] And both do? [14:07] nic-doffay, tried onRunningChanged in the Transition or the SequentialAnimation itself? [14:08] nic-doffay, onOpacityChanged on primaryImage / searchIndicator, too? [14:08] Cimi, https://code.launchpad.net/~mhr3/unity8/expose-renderer-hint/+merge/188848 [14:10] Saviq: not getting the warning is actually a gcc regression: http://paste.debian.net/48653/ [14:11] larsu, oh, interesting [14:12] greyback_: re the problem with screenshot provider in mir - i'm getting this error for every app: https://pastebin.canonical.com/98374/ [14:12] Saviq: ^ [14:13] Saviq, sorry I'm not following what you mean? [14:13] Saviq: is https://bugs.launchpad.net/mir/+bug/1233870 a new bug? is that something under control from your point of view? [14:13] Ubuntu bug 1233870 in Mir "unity8 cpu spike after blank/unblank (nexus4)" [High,In progress] [14:13] The state changes fine. [14:13] nic-doffay, did you try printing stuff onRunningChanged and onOpacityChanged [14:13] nic-doffay, to see if the transitions are triggered / opacity animated at all [14:13] nic-doffay, also, verify the states are changing as you expect them to [14:13] Saviq, not yet, I just tried the states to see if they were hit. [14:13] greyback_: perhaps timing is different when results come from real scope [14:14] pstolowski: ok, that's unfortunately due to you taking an app screenshot before the app has even drawn anything to screen. We're working around it by grabbing screenshots more (too) regularly [14:14] didrocks, well, it's in progress? [14:14] Saviq, it appears the transitions aren't affecting the values. [14:14] Saviq: yeah, I think the management is asking for an ETA ;) [14:15] Saviq: the spike shown in https://docs.google.com/a/canonical.com/document/d/1b-X9tN2Q9c_5r39XzA-Ppebbjuin5zVf9SkAGMcdp9Q/edit# is the same, right? [14:15] didrocks, kinda not my place to determine ETA for Mir... [14:15] kgunn, ↑↑? [14:15] greyback_: the scope just gives icon uri to the shell, so it's still the shell requesting it, so I suppose you need same workaround? [14:15] Saviq: ok, full Mir-side, nothing on unity8? [14:16] didrocks, if I read the bug correctly - yes [14:16] didrocks, and it's the first time I've seen it, too [14:16] seb128: who do I talk to about gcc bugs? Upstream? [14:16] greyback_: btw, I just tested your mir branch on the phone, looks good now, except screenshot problem [14:16] ok, was mostly checking that, thanks! [14:16] let's wait for kgunn :) [14:16] seb128: gcc-the-compiler, not gnome-control-center ;) [14:16] Saviq, got it figured minor typo. [14:17] nic-doffay, it rarely is anything else - you need to dig more ;) [14:17] larsu, try doko on #ubuntu-devel [14:17] so unity8 on desktop will be pushed to 14.10? since it was dropped for 3.10 [14:17] pstolowski: yeah, the view will need to refresh that screenshot request at later stages. We've an updateScreenshots method right now that gets called for that very reason - the scope view will need to connect to that somenow [14:17] nic-doffay, in a WTF moment, step back and look again [14:17] didrocks: what's up...? breif update on making AP tests work...daniel on my team is going to do a quick hack on retries for getting input devices [14:17] hoping it works...if so, will need to mp mir... [14:17] larsu, but desrt will probably advice you to directly go to upstream as well, our guys tend to be busy/not always responsive to pings [14:18] didrocks: and just our luck...server api break as well :) [14:18] greyback_, pstolowski yeah, but the scope won't be doing anything in that case, right? [14:18] Saviq, it def looks better too. Ready for a re-review now. [14:18] Saviq: sure, it won't [14:18] greyback_, pstolowski it's the shell that will trigger updates on the image [14:18] nic-doffay, k [14:18] Saviq: yes [14:19] kgunn: the unity8 spike on blank: https://bugs.launchpad.net/mir/+bug/1233870 [14:19] Ubuntu bug 1233870 in Mir "unity8 cpu spike after blank/unblank (nexus4)" [High,In progress] [14:19] seb128: ah okay. Thanks [14:19] Saviq: pstolowski: yes the scope has done it's job [14:19] kgunn: I guess after the current discussion, this one is a blocker to switch Mir on by default [14:20] didrocks: why ? [14:20] didrocks: who discussed ? and how does it prevent development [14:21] didrocks: on a lighter note...kdub is at 7:20am atm [14:21] larsu, why issue did you find? [14:21] kgunn: rick, olli ;) [14:21] kgunn: so, if we can get an ETA on that one, that would be helpful I guess [14:22] mzanetti, give me a machine gun [14:22] Cimi: sounds like i need the machine gun first [14:22] seb128: -Wreturn-type doesn't work with g_return_if_fail macors anymore (or maybe with all macros?) [14:22] seb128: so when there's a return with a wrong type inside a macro, you don't get the warning [14:22] kgunn, let's see who has better reasons :) [14:22] seb128: you do get it when the return is in normal code [14:22] larsu, ok, seems like a good bug to ping doko about [14:22] seb128: will do [14:23] kgunn: can you update me once kdub is around? [14:23] didrocks: acl [14:23] ack even [14:23] thanks :) [14:24] Saviq, yeah just used to relying on a compiler to pick out my £!$%£"$%£ ups haha [14:24] didrocks, re power spike: don't agree [14:24] olli: that wasn't what was discussed? maybe I missed some words [14:24] olli: I understood this one was blocking, was I wrong? [14:25] didrocks, it blocked rick from using mir as default [14:25] nic-doffay, I don't see a fade-out of the primaryIcon, though [14:25] kgunn, having a look, you have reasons too… I'd sponson a couple of machine gun for our moments of anger [14:25] nic-doffay, just a one-frame visible: false, and then a fade-in of activityIndicator [14:25] *sponsor [14:25] nic-doffay, the other way 'round works fine [14:25] olli: maybe I extrapolated that it was a blocker for switching by default :) [14:26] kgunn: nice to see the screen turning off now with Mir btw :) [14:26] nic-doffay, actually no - the searchIndicator disappears in one frame, too [14:26] nic-doffay, to test just use a longer duration on the animations [14:26] nic-doffay, to clearly see that they fade out / in [14:26] Saviq, taking a look. [14:28] Saviq, should someone from #sdk review the thumbnailer branch, or are you going to top-approve it? [14:28] Saviq, looks like the search indicator vanishes in one frame. [14:28] mhr3, they should [14:28] Saviq, primary icon looks fine to me though? [14:28] nic-doffay, /me increases the durations to look clearly [14:29] Saviq, kgunn, in the EU midnight HO from yday, we discovered that AP testcases/test suites might be using the wrong desktop file [14:29] Mirv: ping again? [14:29] and thomi wanted to fix these [14:29] I assume he got pinned by the input bug [14:29] is there anything we can give jfunk to get started on fixing these? [14:29] Saviq, I reckon it's vanishing in one frame because running is set to false at the same time. [14:29] nic-doffay, it definitelyis [14:30] nic-doffay, just add PropertyAction at the end of the SequentialAnimation [14:30] nic-doffay, or just base running off of opacity > 0 [14:30] Saviq, yeah that was it. [14:30] nic-doffay, also, there's no need to provide both "to" and "from" on Transition [14:31] nic-doffay, one is enough [14:31] Saviq, yeah just wanted to be explicit though. [14:31] nic-doffay, or well, in your case, just one Transition { reversible: true } would probably be enough [14:32] Saviq, wasn't aware of that, will change it. [14:32] nic-doffay, btw, good call with the "internal" states - I was thinking about that late yesterday [14:33] Saviq, I think SnapDuration might be a bit short. [14:33] olli: i'd have to defer to Saviq on that..but iirc it seemed the "moved desktop file" wasn't an issue? [14:34] Although I'm not sure what would be best suited for what mhr3 brought up in the review. [14:34] nic-doffay, 2*SnapDuration, for fade-out, fade-in, should be fine I think [14:34] I personally think SlowDuration looks really cool. [14:34] olli: basically thomi ended up working through setup issues (which ended up in the preamble of the current input bug) [14:34] nic-doffay, you know where the designers are, don't you ;) [14:35] kgunn, that's my understanding, and as such I don't think he was able to work on fixing the desktop file issue in the suites [14:35] olli, yeah, looking through test suites [14:35] Saviq, they're hiding under their desks everytime any of us raises eyes from the laptops :) [14:35] olli, and making sure the commands they use to launch the app works [14:35] Saviq, jfunk & team can do that if they know what to look for [14:35] Saviq, the same ones who wanted to put the search indicator on the right? ;P [14:36] nic-doffay, yes, them! [14:36] olli, send them my way [14:36] olli, but TBH there shouldn't be much in terms of "needs fixing" [14:36] Saviq, so if you can give me some instructions of what's supposed to work from CLI I can have them go through the tests === greyback_ is now known as greyback|shops [14:37] olli, thomi was just unlucky with the suite he tried yesterday - it wasn't working on surfaceflinger either [14:38] olli, effectively running a suite for an app - with unity8 running - should at least launch the app and bring it to front [14:38] confirming: regardless of other outstanding AP/Mir issues [14:38] olli, yes === alan_g is now known as alan_g|tea [14:39] breakfast time! [14:39] thx for the reminder Saviq [14:39] ;) [14:39] * olli is having breakfast [14:39] :P [14:40] Saviq, ok well I'm done with those changes now. [14:47] Saviq, MacSlow, hope things work better today [14:48] fginther, yeah, more stable, but sloooow - mir took over two of the VMs again ;P === alan_g|tea is now known as alan_g [14:49] Saviq, the issue of the long build queue should be better, retoaded was able to grab us another build server [14:50] Saviq, and I'll see what I can do about the VMs [14:55] mzanetti, any idea why that needs fixing? [14:55] mzanetti, apart from testing our patience? [14:58] fginther, thanks [14:58] tsdgeos: pong. I'm working in weird hours today, ie. back now. about the Qt patch? [14:59] Cimi: no, I haven't looked at the log yet [14:59] Cimi: now I have [14:59] Mirv: yep [14:59] Cimi: did you read the logs? [14:59] Mirv: can you have a look at it? [14:59] Cimi: seems pretty clear to me what happened [15:00] Saviq: mhr3: fixed all your comments in fix-genericpreview [15:00] * mzanetti moves back to switching-previews [15:00] mzanetti, actually now there is a log [15:00] actually... /me takes a 10 minutes break [15:00] mzanetti, there wasn't a log when I asked [15:00] mzanetti, like, a big segfault page [15:01] Cimi: mhm... could be that publishing results to public jenkins takes a bit [15:02] mzanetti, awesomeness, no more comments from me, it's great [15:05] Saviq, any reason why if and switch were separate in GenericScopeView? [15:05] Saviq, I just did this http://paste.ubuntu.com/6184223/ [15:07] tsdgeos: yep, looking it now and then trying to get sponsored. earlier today I didn't get to it yet [15:07] Saviq, ignore that [15:07] hah [15:07] Mirv: awesome, tx [15:12] how do i take screenshot of the phone's screen? [15:14] bregma: ping [15:14] bregma: we have real AP results for unity now, but it doesn't look too good [15:15] sil2100, any results are better than no results [15:15] bregma: around ~31 failures per-platform [15:15] I'm seeing a lot of failure to get scope results, somcething that happens form time to time [15:16] and the non-Latin input methods are failing worse than usual [15:16] looks like someone may be playing with gnome-settings-daemon again, trying to break everything at the last minute [15:16] bregma: right... IBus doesn't seem to 'switch on' - I think the tests need to be re-written [15:17] bregma, we didn't have any g-s-d changes for weeks [15:17] bregma, well, ibus changes === greyback|shops is now known as greyback [15:19] sil2100, the problem with our tests is the Gnome paradigm changed so it's not possible to successfully switch to using a non-Latin input method i the middle of a session: we used to be able to do that so we could test [15:20] now we'd have to run separate test runs for each local to test effectively, and we're looking for a simple workaround [15:20] mzanetti: Do you think fix-genericpreview will make it to daily-proposed today? [15:20] I just saw it got +1'ed [15:20] I mean, approved [15:22] karni: yeah. I think Saviq want's to have a last look but I think we can make it today, yes [15:22] mzanetti: yes. and, good! :) [15:23] bregma: I didn't know that the changes were so radical... [15:24] * tsdgeos finds a bug in QSortFilterProxyModel [15:24] * tsdgeos goes to the corner and starts crying [15:27] mzanetti: want to give a look at my test to see if there's not something horribly wrong before i start complaining? [15:28] http://paste.ubuntu.com/6184295/ [15:28] tsdgeos: heh... sure, can do [15:28] * mzanetti pities tsdgeos [15:28] Saviq: greyback: ↑↑ can you also guys give it a quick look? [15:28] it's pretty trivial tbh [15:29] tsdgeos: I know a trap when I see one :) [15:29] :P [15:30] pstolowski: ah, and you too, since this is afaics what is making your code not work [15:30] pstolowski: http://paste.ubuntu.com/6184295/ [15:30] yay, the loader fix has been merged \o/ [15:31] now we just need the distro patch [15:32] I forgot how to enable write access... [15:32] on the device [15:32] touching which file? [15:33] Cimi: adb shell touch /userdata/.writable_image [15:33] karni, thx! [15:34] If you mean to make the system writable. then adb shell reboot [15:34] karni, yep that one [15:35] tsdgeos: didn't try to run it yet, but your model seems to lack Q_OBJECT [15:35] mzanetti: true, doesn't fix anything though [15:35] tsdgeos: do you have a qmake file for it? [15:36] since the object doesn't really have it's own signals/properties/anything [15:36] mzanetti: sure [15:36] http://paste.ubuntu.com/6184338/ [15:39] tsdgeos: wth [15:39] what? [15:39] tsdgeos: so we get rowinserted instead of datachanged in the proxy? [15:40] pstolowski: no, we get nothing, that's the problem :D [15:40] we should get rowInserted [15:40] we don't [15:41] i.e. QCOMPARE(proxyRowsInsertedSpy.count(), 1); from line 118 is failing [15:41] mzanetti: anything you find obviously wrong in the code? [15:41] tsdgeos: not sure... gimme a few more minutes [15:43] tsdgeos, so what you're seeing is that the call to proxy.rowCount() updates something in the proxy internally and that fixes the issue? [15:43] Saviq: yep [15:44] and that's why when i add billions of debugging to pstolowski code [15:44] it suddenly works :D [15:44] :/ [15:44] tsdgeos, right - yeah, nothing wrong I can see with your code [15:46] sil2100, regarding thumbnailer, if new branches land to trunk, will things to go universe? [15:46] sil2100, or rather, when will they? [15:46] Saviq: and that rowInserted missing makes the qml list not "see" there's new stuff [15:46] tsdgeos, yeah, of course [15:46] and we end up with an empty thing [15:46] tsdgeos: at least you know the workaround... but this is scary [15:47] * tsdgeos wonders if this works in qt4 [15:47] nah [15:47] let's try 5.2 [15:48] nah [15:48] lol [15:48] pstolowski: agreed, it is scary [15:49] mhr3, on https://code.launchpad.net/~unity-team/unity8/fix-genericpreview/+merge/188355 - you only did functional review, right? code review is still on my plate? [15:49] Saviq, yep, too much qml for my taste :) [15:49] ;) [15:50] tsdgeos: so what exactly is the weird thing? [15:50] * mzanetti sees lots of weirdness [15:50] mzanetti: what's weird for you? [15:50] mzanetti, hum, the image in video preview still starts up some 10 GU high, not 16x10 aspect ratio? [15:51] tsdgeos: well, calling rowCount() shouldn't have any affect on the tests afterwards [15:51] mzanetti: of course [15:51] but besides that? [15:51] the other turned out to be my bad... missed the filter [15:53] Saviq: initialHeight: width * 10 / 16 [15:53] Saviq: I think those previews are 4:3 aspect ratio [15:53] if you're having somewhat similar results on your video dash [15:54] mzanetti, yeah, even so - they start up really narrow (low height) [15:54] Saviq: but I assume that can change depending on the backend [15:54] oh really? lemme check again [15:55] mzanetti, karni and sorry, but it's not gonna happen today - I didn't even start the code review there - first thing tomorrow [15:56] fine with me [15:56] dednick, you saw https://code.launchpad.net/~nick-dedekind/unity8/indicator-titles/+merge/186049/comments/432038 ? [15:56] Saviq: nope. looking [15:57] ack, thanks for letting me know Saviq [15:59] nic-doffay, why didn't you go for a single, reversible Transition? didn't work? [16:00] Saviq: I'm afraid the height: implicitHeight breaks the initialHeight [16:01] mzanetti, huh, shouldn't matter at all - if it does - bug in LazyImage [16:01] mzanetti, as implicitHeight is just animated between initialHeight and image.height [16:02] Saviq: yeah, there is something fishy with implicitHeight in there [16:02] Saviq: the "height: implecitHeight" shouldn't be needed, yet it changes things a lot [16:04] mzanetti, damn, we need to get to the bottom of this [16:04] nic-doffay, http://pastebin.ubuntu.com/6184445/ apply that to search indicator [16:05] nic-doffay, if you agree, that is, and I'll happrove [16:05] nic-doffay, you shouldn't have any values in a Transition btw, that's what States are for [16:07] Saviq: isn't there a binding loop in there? [16:07] mzanetti, shouldn't? [16:07] Saviq: height: root.scaleTo == "width" ? implicitHeight * root.width / implicitWidth : implicitHeight [16:08] wouldn't implicitHeight change when changing height? [16:08] mzanetti, that's image's implicitHeight [16:08] Saviq, I wanted to add the PropertyAction at the end of the one. [16:08] yeah, and the image's height [16:08] Which is why I didn't opt for the reversible. [16:09] mzanetti, and no, implicitHeight never changes [16:09] Saviq: is that the initial sourceSize of the image? [16:09] mzanetti, it's internal to Image - is set to what the image's actual size is after applying sourceSize [16:09] Saviq, but it makes little difference to me. I'll change it to that. [16:09] after applying sourceSize... ok [16:10] nic-doffay, it could be a PropertyAction, yeah, but it's just as well - we really want it running whenever it's visible, so... [16:11] mzanetti: actually the move was wrong [16:11] mzanetti: needs to be http://paste.ubuntu.com/6184443/ [16:11] but that doesn't help at all [16:11] Saviq: fixed the conflicts. [16:11] dednick, thanks! [16:11] Saviq, changed. [16:11] nic-doffay, thanks! [16:12] Cimi: any movement on the ubuntu-settings-components branch in MP? [16:12] dednick, nope [16:12] Saviq, np [16:13] dednick, the rest is fine, I didn't test the calendar [16:14] nic-doffay, happroved [16:15] Saviq, any idea why jenkins is unhappy? === boiko_ is now known as boiko [16:15] nic-doffay, flaky test, ignore [16:17] Saviq, need your opinion on this too: https://code.launchpad.net/~nicolas-doffay/unity8/dismiss-keyboard/+merge/188599 it's my opinion that the InverseMouseArea is a bit flaky with the keyboard visible. [16:17] If so I need to bring it up with the SDK guys. [16:17] nic-doffay, the "z" there shouldn't be needed [16:18] nic-doffay, and it took popover.contentHeight into consideration for a reason - otherwise you tapping on the popover to select a previous search or close it would unfocus the input field [16:20] Saviq, ah I see. I'll change that. Still would like your thoughts on what I mentioned before though. [16:20] nic-doffay, it's possible, IMA proved troublesome in the past [16:21] dednick, is it designed that we now have "Incoming" instead of "Messages" and "Wednesday" instead of "Time & Date"? [16:22] larsu: ^ re Incoming ? i believe you changed that recently [16:22] Saviq, most notably I noticed there were loads of issues with typing and then trying the inverse mouse area. [16:22] Saviq: re Time & Date, unfortunately yes, it is as designed. [16:23] Either way I guess the QML is sound so we can land this? Pushed the modifications too Saviq. [16:23] dednick, that I'm ~fine with, but "Incoming" sounds weird [16:23] nic-doffay, I'll look at it closely tomorrow [16:23] Saviq, That is correct to design [16:24] tedg, :S [16:24] Saviq: oh boy... it works *sometimes* [16:24] mzanetti, uh oh [16:24] Saviq: yeah, the name isn't great, but i guess it's more "technically correct" seeing as it handles missed calls as well. [16:24] Saviq, Better discussed over several beers :-) [16:24] tedg, we're planning those [16:24] +1 [16:26] i would go for "Notifications" or something like that. [16:27] dednick, approved [16:27] 10 MPs awaiting... [16:28] 2 hrs per MP... [16:28] mterry, un-approving, then [16:29] Saviq, yeah, I'm a bit confused on why it's happening, will dig [16:29] mterry, thanks [17:02] kgunn, I think it worth mentioning for the wiki that PPAs will be easily available for those that want trunk Mir on their 13.10 installations [17:02] dednick: yes, design asked me to. === alan_g is now known as alan_g|EOD [17:27] Saviq, how do I run an AP test (say the one thomi used yday) from cli? [17:28] cli/shell [17:33] olli, autopilot run blah [17:33] olli, where blah is the test suite name [17:34] olli, on a readonly device - thomi's description on bug #1233944 has some details [17:34] bug 1233944 in Mir "Unity8/Mir is unable to open autopilot uinput devices" [Critical,In progress] https://launchpad.net/bugs/1233944 [19:03] morning [19:05] hey thomi [19:05] hey man [19:05] gotta go help move some furniture, brb [19:05] thomi, long story short: Mir tried to open the input devices before udev gave it the permissions [19:09] Saviq: yeah, I saw that [19:09] just building packages to test out the fix now === jhodapp is now known as jhodapp|afk [20:18] sergiusens: ping? [20:29] veebers, pong [20:39] sergiusens: hey, how you doing? I'm not sure I understand your response to my email (re: unity8 AP test dependencies etc.) [20:40] veebers, we had a very similar conversation that is about to happen with your emails ;-) [20:41] veebers, the best I can think of for those deps you say are missing is to add them to autopilot-touch [20:42] sergiusens: well, I guess you could make an argument for python-mock being there (req. by ubuntu ui toolkit) but I"m not sure unity8-fake-env belongs there) it really sounds like a work around [20:44] veebers, the problem with the latter is that it's using packaging for something it shouldn't [20:44] veebers, but anyways, if that does not depend on anything else I can take care of it [20:45] veebers, although it wasn't in my plans to work on getting non click stuff before click [20:47] veebers, just to reinforce, both of these are workarounds, the proper solution is to have autopilot work in daemon mode [20:48] what's "daemon mode"? [20:48] thomi: I also ask thomis question [20:48] err, sorry for the ping thomi :-\ [20:49] thomi, the super nimble autopilot that we talked about during the sprint [20:50] sergiusens: oh, right. I think we're not going to have that any time soon, so we'd better stop building in workarounds/hacks and waiting for it to arrive [20:50] better to fix it properly for what we have today, IMO [20:50] thomi, well I don't know what the proper way is for unity [20:50] thomi, I'll leave that to you, the proper way for me is that said daemon mode [20:51] I don't really understand the problem you guys are talking about, as I lack context [20:51] thomi, don't go into read write mode and be able to test unity8 [20:51] sergiusens: why is that a problem today? [20:51] thomi, which is a req we are aiming for for click apps and have it solved, but not for unity8 [20:52] thomi, many packages [20:52] you're saying that the tests require additional stuff to be installed before they can run? [20:52] thomi, yes [20:52] ok [20:53] thomi, like the unity8-fake-env package [20:53] sergiusens: remember waaaaay back when you proposed the phablet-click-test-setup script, and I kept saying "yeah, but it'll need to handle dependencies", and you were all "nah, it'll be cool"... :P [20:53] seems that script really needs to handle dependencies :) [20:53] thomi, yeah, for click apps [20:53] thomi, never said anything about non click [20:54] thomi, want to play the remember game ;-) [20:54] sergiusens: how the application is packaged is irrelevant here - the test suite needs dependencies, and we're no longer installing the test suite using apt-get, so we need a replacement [20:54] whether the application is packaged as a click app or not is beside the point [20:54] or am I missing some nuance? [20:54] thomi, who came up with that req for ackaged apps? [20:55] sergiusens: you mean for the test suite? I don't understand your question, sorry [20:55] *ackaged == deb [20:55] the requirements on the test suite .deb files (like unity8-autopilot, for example) are set by the test authors [20:56] is that what you meant? [20:56] thomi, well let them figure out how to test without rw, from my pint of view, there is no reason they shouldn't keep doing what they are doing as they are image builders, not app devels [20:57] O.0 <- confusedface [20:57] the problem is that dev teams *are* nominating dependencies, but they're not getting installed [20:58] but naywya [20:58] *anyway [20:58] thomi, well that's easy, just install them [20:58] sergiusens: it'd be nice if that happened automatically when you grabbed the test suite, like it used to do when we could use apt-get. That's my complaint, and, I suspect veebers' as wel;l [20:58] it used to happen automatically, now it requires a manual step [20:59] thomi, phablet-test-run -n unity8-autopilot unity8 [20:59] and what's worse, a manual step in several locations [20:59] sergiusens: but that doesn't work on a RO FS, for that, we need to use phablet-click-test-setup [20:59] which is the bit that's broken [20:59] sergiusens: for that to work with ubuntu-system/ro you need to make a change right? [20:59] thomi, as the name says phablet-CLICK-test-setup [20:59] ah, what thomi said [20:59] as in CLICK [21:00] sergiusens: right, but is unity8 even packaged as a click package? [21:00] sergiusens: yesterday you told me that the 'click' part of the name should be removed. Want me to find that in the scrollback? [21:00] I thought it was the QA team's problem to solve that [21:00] thomi, yeah, if we eevntually solve that [21:00] thomi, sroll back if you want [21:00] thomi, I also said I wanted click to work first [21:01] sergiusens: I'm not saying that it's necessarily your problem, just that we need a solution, and it's not going to happen if everyone keeps saying "oh yeah, the problem can be solved by writing this massive autopilot feature" [21:01] so instead, let's figure out how to solve the problem properly, and then get it done [21:01] thomi, well we had that meeting during the sprint to talk exactly about this; I'm just seeing a rerun [21:02] maybe my memory is broken (quite probable), but I don't remember a meeting about provisioning test suite dependencies? [21:03] thomi, nope, not about that [21:03] well, that's what I'm talking about ;) [21:04] thomi, it was more general and on a pad which I never wrote down the address for [21:04] sergiusens: would you like me to take the lead on this, and talk to the CI & QA teams and figure out what the correct solution is? [21:04] sergiusens: we keps all the links in a SS, one sec, let me find the URL [21:05] sergiusens: https://docs.google.com/a/canonical.com/spreadsheet/ccc?key=0AoH1kjVoZRmBdEJ6eWZXNnRoNnFnUDBBY0xvVWIwdWc&usp=drive_web#gid=5 [21:08] thomi, seems https://docs.google.com/a/canonical.com/document/d/1sULsSfzHdTt_8ueJoD7U_vPsD9slC0qgjxBtLOOZaDc/edit wasn't fully passed in [21:08] thomi, I do recall one thing being said about test author being able to use whatever they wanted [21:08] as a pro for doing this [21:09] sergiusens: again though, that's a different topic. I remember *that* meeting - that was scoping out the remote test runner, to try and figure out whether we should build it or not. Yes, one of the advantages would be that it makes dependency installation easier, but it's a huge chunk of work, and I'm saying that we probably ought to solve this problem separately [21:10] since it's looking pretty unlikely that the remote runner will get done any time soon [21:10] thomi, tbh, it is the proper solution no matter when it happens [21:11] sergiusens: you may be correct, but a proper solution in 5 years time (or whevever) doesn't help us today, or for the next 5 years [21:11] I think we need a solution that's cheaper to implement in the short term [21:11] thomi, the side option as we discussed with you and doanac two months ago was to add all the deps to autopilot-touch that we needed to test the image [21:13] that's certainly a possible solution. It still doesn't give us the ability to change dependencies easily across multiple projects, but I suppose it's better than nothing [21:13] sergiusens: could your test setup script not read the dependency list from the debian/control file, and install them along with the test suite? [21:13] thomi, well I don't actually _install_ anything [21:14] well, ok, s/install/write them to disk somewhere so they can be loaded by the test suite/ [21:14] thomi, some of those dependencies could have weird requirements [21:14] thomi, some may require import paths for qt, other may have hard coded deps [21:14] s/deps/paths/ [21:15] yeah, it's a tricky problem - are we any closer to having an overlayfs that works on the phones? Cos that would make this trivial [21:16] thomi, you can bug the kernel team about that, but there's a different kernel per device and some devices have too much of an old kernel [21:16] hmm, ok [21:16] sergiusens: you agree though that it would solve the problem, thought? [21:16] *though? [21:17] thomi, yes, it would solve the problem of not being able to install directly, but then I'm not sure why you don't just break into rw for these two apps under test [21:18] well, from a QA perspective, i think it's better to never have to break into RW. Secondly, it's unlikely that it'll be just two apps. It might be 2 apps today... [21:18] thirdly, it's better to have a single approach that will work across *all* apps, forever :) [21:19] thomi, I don't think you want this in production/released phones [21:19] an overlayfs? why not? [21:20] thomi, with overlay it would be easy to break confinement, override drm and stuff like that [21:20] thomi, nothing prevents it now, but I do see that it won't go long term [21:21] OK. I'll go talk to kernel-type people, and see if we can't get some traction on this. In the mean time, please stop telling people that autopilot daemon mode / remote runner is the solution to their problems :) [21:22] ...unless, of course, you want to write & maintain it ;) [21:22] thomi, ok, I'l just forward them to you then [21:23] can somebody help me uninstall mir? it's seriously screwed on my system and following the uninstallation instructions just results in this: http://paste.ubuntu.com/6185686/ === salem_ is now known as _salem [21:32] robru, I 'think' the ppa info is outdated. try: "sudo apt-get remove unity-system-compositor;sudo restart lightdm " [21:33] fginther, thanks [21:37] fginther, you are a lifesaver! Now I just need to figure out why I can't access launchpad and i might actually be able to get some work done today... [21:38] robru, glad it worked. [21:40] fginther, you're not experiencing any issues accessing lp are you? multi-minute load times? frequent 'connection reset by peer' ? [21:40] robru, no, I don't see any issues today [21:41] fginther, hummm, there's no vanguard in #is... can you point me at 'RT'? I don't know what that is [21:42] robru, hang on, it just isn't your day is it? [21:42] fginther, nope :-/ [22:03] greyback, go away [22:03] Saviq: no you go away [22:03] greyback, no YOU [22:04] Saviq: you times 100 [22:04] greyback, no, you whatever times you say times 100 [22:07] oh, it worked :) [22:11] Saviq: I simply couldn't deal with that numbers that high [22:12] yay for day off! [22:12] * greyback gone [22:12] greyback, have a good one