=== jhodapp|afk is now known as jhodapp === jhodapp is now known as jhodapp|afk === OutOfControl is now known as benonsoftware [08:42] * Saviq needs to sleep more... had a "nightmare" about Ubuntu Phone going to become just a mobile browser (with all the UI served remotely)... [08:45] lol [08:45] yeah you need more sleep [09:01] * Saviq hears sizzling... ah, it's the cat found some sun... now waiting for the smell of something burning... [09:07] :D [09:10] tsdgeos, so, we got the "expand music" postEvent crash back, and seems to be easier to trigger [09:10] bug #1262711 [09:10] bug 1262711 in unity8 (Ubuntu) "Unity crashes with lots of music displayed in expanded music category in home scope" [Undecided,Confirmed] https://launchpad.net/bugs/1262711 [09:10] :/ [09:10] I got some symbols (last comment), but they're not helpful much I'd say [09:10] it's definitely related to the album art provider [09:11] trying to find out whether it happens on desktop, too [09:12] but seeing as it's a race, I doubt it... [09:19] anyone know why phablet-flash isn't working at the mo? [09:22] dednick, https://lists.launchpad.net/ubuntu-phone/msg06981.html [09:22] Saviq: ah. yeah i guess that would be it [09:22] :) [09:23] ta === jhodapp|afk is now known as jhodapp [11:59] mhr3, do you have apps in new click scope? [12:00] saviq, i did yesterday [12:00] and we didn't land click since afaik [12:01] hmm /me wonders if my locale affects it [12:01] saviq, anything in scope-registry.log? [12:02] hmm [12:02] ScopeBase::run(): unity::LogicException: Error checking for key existence (/home/phablet/.local/share/applications [12:02] hmm indeed [12:03] alecu, ^? [12:03] it was an empty .desktop file [12:03] saviq, hm? [12:03] mhr3, broken file [12:03] mhr3, empty [12:03] who tried to read it? [12:03] mhr3, scope did [12:04] how come it threw unity::logicexception? [12:04] nvm me, /me digs [12:08] mhr3, alecu yeah, broken .desktop files make the scope die [12:09] ah they actually use our iniparser [12:09] ok then [12:09] should have made it private :P [12:10] LOL === MacSlow is now known as MacSlow|lunch [12:38] tsdgeos, btw fix for the centered "Apps" header is sitll coming to new-scopes? [12:38] mhr3: didn't work on that [12:38] plan to? [12:38] Saviq: ↑ ? [12:39] mhr3, would rather get the new header in [12:39] mhr3, it's a dying component [12:40] saviq, it's just... very visible, always there after reboot [12:40] first thing designers noticed btw :) [12:41] mhr3, maybe they'll get the new header designed quicker ;) especially for in-dash interaction [12:41] your call [12:41] mhr3, I know, I'd like to fix it, too [12:41] mhr3, I just don't want to spend a day fighting with a dying component that's proved difficult over and over again === alan_g is now known as alan_g|lunch [13:06] Saviq: so, how did you get an empty .desktop, btw? [13:07] alecu, interrupted test fixture [13:07] or some such [13:08] ah, sounds right. Ok, I'll add some checks so it fails gracefully on parse errors there [13:17] /we need to do some optimizing for new scopes... ;? === _salem is now known as salem_ [13:41] ping tsdgeos. I have the new scopes in my phone now. Do you need some more help with the autopilot tests? [13:48] elopio: can you confirm unity8.shell.tests.test_emulators.GenericScopeViewEmulatorTestCase.test_open_preview fails? [13:48] i think it's a LVWPH but if you can have a quick look [13:48] it'd help [13:53] tsdgeos, it doesn't fail here [13:53] Saviq: on the phone? [13:53] tsdgeos, been running it the whole morning today [13:53] interesting [13:53] tsdgeos, on phone, yes [13:53] was failing yesterday for me :/ [13:54] tsdgeos, I get some random failures due a crash I'm trying to pin, but nothing more [13:54] ok [13:54] and apparently a lockup on exit sometimes [13:56] oh, crash on exit, too === alan_g|lunch is now known as alan_g === dandrader is now known as dandrader|afk [14:03] tsdgeos: I can, give me soem minutes. [14:06] Saviq, yes to optimizing, scrolling is pretty bad with new-scopes [14:07] Saviq, and once michi is back we'll hopefully stop hogging the ui thread :) [14:08] but you didn't even notice that, did you? :P [14:09] mhr3, just a little bit [14:09] :D [14:09] mhr3, but yeah, the Card approach is nice and all... but results in unused objects all over the place [14:09] Saviq: any plan on caching app icons yet? I think that would add a lot to perceived performance [14:10] mzanetti, cache: true, or something more elaborate like bug #1224998 [14:10] bug 1224998 in Unity 8 "Cache preview and thumbnail images" [Medium,Triaged] https://launchpad.net/bugs/1224998 [14:10] the latter [14:11] mzanetti, SDK topic [14:12] mzanetti, so yeah, there is a plan (that I wrote in the bug), but that's about it [14:12] mzanetti, and I really don't think *we* should be doing anything about it [14:12] mzanetti, it needs to be part of the platform - replacing the default image provider [14:13] mzanetti, but anyway we have cache: true in the Card [14:13] probably should have added it to the CardHeader. too [14:14] mzanetti, but it's worse than that, we need to simplify the delegates as much as possible [14:14] because now, for each icon + title, we instantiate like a dozen other invisible objects [14:15] yeah... [14:21] man... I think I need to reinstall my machine. its broken all over the place since the last big upgrade [14:21] now apps are frozen for 5 seconds when getting focus === MacSlow|lunch is now known as MacSlow [14:26] tsdgeos: what branch are you using to run the tests? [14:26] elopio: new-scopes-clean-to-trunk === dandrader|afk is now known as dandrader [14:27] * MacSlow can't connect to the mumble-server [14:28] MacSlow: huh? but you do appear in our channel [14:29] mzanetti, mumble window is greyed out and it's eating 100% of one cpu-core... no clue what's going on [14:29] ah [14:31] * MacSlow tries to reboot :/ [14:43] saviq, any way to make it so that the card objects would be owned by cpp code? [14:45] tsdgeos: it fails with the unity8 installed from the landing013 PPA. I'm trying now running it from the same branch. [14:46] mhr3, not really [14:46] mhr3, the view needs to create/destroy them as they come into / out of view [14:46] mhr3, well, in that sense they are, but by all the *Views [14:48] saviq, well if gc is the problem, then we should just try to hide all the small objects into a one big :) [14:48] mhr3, no, there's no gc [14:48] mhr3, they're destroyed [14:48] mhr3, it's just that when they're recreated again when coming on screen [14:48] mhr3, all the invisible ones get created (in QML world), too [14:49] mhr3, even if not passed to the GPU [14:49] first thing to do would be to wrap all the bigger pieces in Loaders [14:49] and only load them when we know they'd be used [14:50] second would be to create a library (remember? a library?) of Cards that would be selected based on what's mapped and such [14:50] dandrader, hey, you said in your video you did something to Mir (patch?) that prevented it from stopping the rendering for unfocused surfaces [14:51] dandrader, icanhasthat for testing a theory? [14:51] Saviq, I did in in the "unity-mir"/QPA code [14:51] dandrader, works for me [14:52] Saviq, so it just doesn't call application->setState(Suspend) [14:53] dandrader, ok, so that's different :? [14:53] * Saviq needs mir to not stop rendering when screen off [14:59] tsdgeos: Ran 1 test in 49.855s [14:59] OK [14:59] elopio: ok, must be me, Saviq was saying it also works for him [14:59] but it does some weird things. I don't see the preview being opened. Maybe it's just too fast, I'll debug. [15:00] ok === alan_g is now known as alan_g|tea === alan_g|tea is now known as alan_g === gatox is now known as gatox_lunch [16:02] Saviq: can you do a quick swing through [16:02] https://docs.google.com/a/canonical.com/document/d/1A1wmF92jlwupi1rPspEKUuuJYNWvN3Io4vjqxoofxy8/edit [16:03] kgunn, ah! [16:07] kgunn, done [16:07] ta === gatox_lunch is now known as gatox [16:30] unbelievable... placing the N4 on my keyboard makes my notebook fall asleep [16:31] :D [16:31] reminds me of the N900 which injected keypresses when placed on a keyboard cable [16:32] Saviq, what's your feeling about this? https://code.launchpad.net/~paulliu/unity8/zoomImage/+merge/207941 [16:39] Cimi, other than it could potentially go into SDK, from a quick glance looks tested and all [16:39] Cimi, didn't look at it more, why? [16:39] Saviq, I don't know about this algorithm, those 0.98 scaling values [16:40] Cimi, did Paul answer your question on that? [16:40] Saviq, paul didn't write this code, he took it somewhere from meego [16:40] uh oh [16:40] that sounds like licensing issues [16:40] indeed [16:40] don't know what he modified or if it was only 'inspiration' [16:42] Cimi, please add these concerns as comments to the MP [16:42] Cimi, and ask Paul to reply there [16:43] Cimi: good catch Cimi [16:43] * Saviq just noticed the flights next week in his calendar... London, here I come! :P [16:43] Saviq, are you here next week? [16:44] Cimi, 31st [16:44] Cimi, and then 6th [16:44] Saviq: i'm getting like seriously worried about you, malta and a non-fish food selection [16:44] what? [16:44] not fish? [16:44] kgunn, === dandrader is now known as dandrader|lunch [16:46] Saviq, kgunn: can you please take a look at this document? [16:46] https://docs.google.com/a/canonical.com/document/d/1yB6myUu4JzkI315ZNE3hBgWGIQVpcdpJWX-bb3q7sr0/edit#heading=h.xbdfutgfq4lh [16:47] those are the features that we think should be tested on every new unity release. If you have comments or doubts, you can ping me or rhuddie. [16:49] elopio, whoa, nice doc, will definitely read through [16:51] Saviq: rhuddie does magic with test plans. I'm a mess, so it's nice thanks to him :) [16:52] elopio, saviq hopefully it's a good starting point... please feel free to add any comments :) [16:53] for defs [16:58] mterry, kgunn, so you say there is ABI breakage in the branch / mir devel? [16:59] Saviq, https://code.launchpad.net/~paulliu/unity8/zoomImage/+merge/207941/comments/500681 [17:00] whoa, 6pm crept up on me [17:07] Cimi, k === om26er_ is now known as om26er [17:13] ok, /me kicks mir-alone build, then [17:38] Can I tag this one bitesize? https://bugs.launchpad.net/ubuntu/+source/unity/+bug/1103833 [17:38] Ubuntu bug 1103833 in unity (Ubuntu) "Wine's icon and description appears for all Windows applications" [Medium,Triaged] [17:44] Saviq, btw design acked https://code.launchpad.net/~unity-team/unity8/new-scopes-clean-to-trunk/+merge/209459 [17:44] time to update the desc ;) [17:44] and f..ing merge it! :P [17:46] mhr3, tell that to Qt 5.2 ;P [17:48] kgunn, y u aborted? [17:48] Saviq: cause it wasn't gonna work [17:49] Saviq: reconfig and rekick [17:49] kgunn, ah, good reason [17:49] ....it won't let me [17:49] is it tied to your name to reconfig ? [17:49] kgunn, might be [17:49] Saviq: sorry...about that. i should've told you [17:49] kgunn, no worries [17:50] kgunn, why wouldn't it work? [17:50] but i added the mp's that should make mir happy [17:50] ERROR:root:session-manager-touch was not in the initial list of components for that silo. You can't reconfigure the silo yourself. Please ask the landing team to reconfigure it for you. [17:50] mir version 0.1.8 already on dev branch [17:50] kgunn, right [17:50] Saviq: huh....i thot we could do that [17:50] kgunn, not if the components change [17:50] kgunn, or well, not if you add some [17:51] kgunn, so yeah, my name shouldn't matter I think [17:51] kgunn, AlbertA, btw, confirmed [17:51] next song plays with AlbertA's patch [17:51] not good [17:51] * Saviq updates bug [17:52] Saviq still needs a reason to rebuild [17:52] kgunn, hm? [17:53] Saviq: so i just added the 0.1.8 mp's like i would normally...its all good...don't worry [17:53] famous last words [17:53] rebuild for unity-mir/papi [17:53] gotta run, be back in an hour === dandrader|lunch is now known as dandrader === alan_g is now known as alan_g|EOD [18:32] finally we have the vkb at our compositing mercy :P http://ubuntuone.com/5lUS2QMOKej1rNPDf5n9hy [18:34] \o/ [18:50] dandrader: yay! I think we should leave it like tat :P [18:51] greyback, it's a dream coming true. I can just remove the tons of hacks we have around the OSK subject and simply position it like any other Item in the unity8 scene! [18:51] dandrader: yep, sooo much nicer [18:52] dandrader: I want it to slide up/down! [18:52] not just appear, which I always hated [18:52] greyback, yeah, sure it will [18:58] * greyback away === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [20:45] greyback, dandrader, I'm not sure we want to be too smart in the shell about how the OSK behaves [20:45] greyback, dandrader, sliding up should be handled by the OSK itself [20:45] we need to think about 3rd party input methods [20:46] IMO the vkb should be a rather freeform window (not necessarily even bottom-attached) [20:48] dandrader, one thing we do need to do, though, is treat the vkb as part of the app surface, so that it appears together with the app in the spread and all the animations [20:49] Saviq, are the designs like that? [20:49] problem ↑ here is that we need a vkb screenshot (buffer?) per-app, and preserve state between apps (that's rather outside of our jurisdiction, though) [20:50] dandrader, yeah, otherwise apps would need to reflow or be blank at the bottom or whatnot [20:50] dandrader, app in the spread needs to be exactly as you left it, and how it will come back, part of that is the vkb [20:51] dandrader, so we might needs to consider vkb actually supplying a surface per app [20:52] this is not sounding good [20:56] dandrader, well, it doesn't have to be bad, we just need to think of a good way to do this [20:57] dandrader, we need something similar for trusted sessions, like with the content hub, for example [20:57] Saviq, we have to find the right balance between leaving things loose and abstract to let 3rd parties go crazy on input method plugins and a pleasing, smooth, user experience [20:58] you need to be able to swipe away an app that is in the middle of a content sharing session, and it needs to be rendered into the same app, at some point - maybe even all three - app, content hub and the sharing app, as those will be three distinct surfaces composited together [20:59] Saviq, sacrificing the end user experience in favor of hypothetical use cases (a 3rd party input method that might do this and that) is a bad thing to do IMO [20:59] dandrader, sure, but I'm afraid input methods are the things that go rather crazy [20:59] dandrader, we don't need to sacrifice, we might just need to delegate [20:59] Saviq, have you seem them going crazy on other platforms? [21:00] dandrader, depends on what is "going crazy", but there's plenty of weird input methods that are out there, and we need to strive not to lock any one of them out [21:00] dandrader, good thing is, though, it's just like an app - if it'll be bad, people will rate it down and uninstall it [21:00] Saviq, I need concrete examples [21:00] dandrader, so I don't think this is a really big issue UX wise [21:01] dandrader, I don't have them, but I'm simple, I use a US layout keyboard [21:01] Saviq, ok, so we are in the hypothetical ground here [21:02] dandrader, http://hubpages.com/hub/Android-Softkeyboard-Shootout-which-input-method-should-you-use [21:04] dandrader, so yeah, ultimately they all seem to be a bottom-anchored rectangle [21:05] jeez, the narrator here is a cheap text-to-speech software: http://www.youtube.com/watch?v=gHj2ePxxMo0 [21:05] dandrader, but why would we impose a slide-up transition for it? why not leave that to the vkb itself? [21:06] dandrader, but then there's stuff like dasher http://en.wikipedia.org/wiki/Dasher [21:06] erm http://en.wikipedia.org/wiki/Dasher_%28software%29 [21:06] and any number of things like that, that we should try and not preclude for no good reason [21:07] dandrader, and well, ultimately it's not our decision anyway, but UX design's, on what we want to allow [21:07] Saviq, I would ask the other way round: why limit our animations and interactions for no good reason [21:07] ? [21:07] dandrader, limit in what way? [21:09] Saviq, sliding up and down for instance. and, more importantly, how such animations are orchestrated to fit well with other UI components such as the launcher [21:14] dandrader, how would they be orchestrated? how does the launcher impact the vkb? [21:15] IMO the launcher should just go over the app (and hence the vkb, which is part of the app) [21:17] I don't see the reason to force a slide up/down animation, what if people want it to fade in/out, or plainly just show up? I'd see it as a user preference, and not something we want to implement, but the vkb should [21:17] but again, that's me [21:17] I don't have to be right, nor do I have the binding word :) === seb128_ is now known as seb128 [21:19] Saviq, this a good video on the current crop of vkbs: http://youtu.be/Ig7icgTHtDY [21:21] Saviq, I see the way you dismiss a OSK as a shell thing. like the way you dismiss a window [21:21] Saviq, from the shell point of view an input method is a "black" rectangle [21:21] Saviq, although it can draw outside its boundaries. [21:21] anyway, EOD === kenvandine_ is now known as kenvandine === salem_ is now known as _salem