=== RAOF_ is now known as RAOF [08:06] Saviq: so it was me that broke the previews right? [08:06] tsdgeos, not really, it only ever worked by sheer luck [08:07] i see [08:07] tsdgeos, when app installed → scope would signal result invalidation, as well as send a new preview for the, now installed, app [08:07] but probably my refactoring broke it? [08:07] i mean there's not much more that would touch previews and such [08:07] tsdgeos, yeah it surfaced it [08:08] tsdgeos, so before, somehow, the results did not get reset for the previewListView, and it would not reset [08:08] tsdgeos, or they did reset later [08:08] tsdgeos, so when bringing in the new preview, currentIndex was already -1 [08:09] tsdgeos, as opposed to the real, currently selected one, 'cause the model for it was reset in the mean time [08:14] tsdgeos, anyway: https://code.launchpad.net/~saviq/unity8/fix-preview-updates/+merge/204559 [08:14] tsdgeos, if you could get it to Approved (mterry waited for -ci run that was late) [08:14] will do [08:15] tsdgeos, thanks === Wellark__ is now known as Wellark [08:39] Saviq: can you explain what the "onPreviewDataChanged: showProcessingAction = false" does? [08:40] tsdgeos, DashPreview.qml, at the bottom [08:40] tsdgeos, there's a MouseArea that eats input after you have triggered an action [08:41] tsdgeos, but it needs to be disabled when new previewData comes, as that action was executed, then [08:41] ok, makes sense [08:41] tsdgeos, again, TBH not sure how this could have worked before... [08:41] tsdgeos, most probably the PreviewListView did get reset, but somehow managed to maintain the currentIndex === asac` is now known as asac [09:00] Saviq, why did we loose 7.85 from new-scopes? [09:01] mhr3, not sure what you mean? the version number? [09:01] mhr3, probably because I cleaned up the recipes [09:01] Saviq: is there any chance we get the netowkrmanager tests fix in? bug1271676 seems to be something MWC people want to see fixed asap and without the tests fixed mzanetti is abstaniing from reviewing it, so it's all a bit of urgent [09:01] Saviq, yep, the version in the ppa is now lower than distro [09:01] bug #1271676 [09:01] bug 1271676 in Unity 8 "Launching an app from the app scope causes app scope to jump to the top" [High,In progress] https://launchpad.net/bugs/1271676 [09:01] which sucks [09:01] mhr3, not because we lost 7.85 [09:02] mhr3, but because there was a revert of unity8 yesterday [09:02] mhr3, should be sorted today [09:02] tsdgeos, it's waiting in the queue, will be released today with the fix for the previes [09:02] previews [09:03] tsdgeos, they won't get a fix for it if we don't release at all, either, so it just needs to run its course [09:04] ok [09:04] tsdgeos, what I mean is even if we got it in trunk any other way, if we don't release, it won't be there for the image [09:04] sure [09:05] tsdgeos, so we just need to be able to release again [09:17] sil2100, didrocks, FWIW unity8 fix is in CITrain, waiting for a silo [09:17] Saviq: excellent, thanks! can I ask you a favor? [09:17] didrocks, one more!? ;P [09:17] hehe, indeed [09:17] we try to see if we can promote the current image [09:18] (well 161, it's the one we have whole test results) [09:18] there is on unity8 failure [09:18] one* [09:18] on both mako and maguro [09:18] * sil2100 is flashing 161 just now [09:18] + 162 fails the same on mako [09:18] *162 [09:18] ;) [09:18] didrocks, upgrade unity-notifications [09:18] Saviq: so, if we promote the current image [09:18] didrocks, it's because of the unity8 rollback [09:18] that will be broken? [09:19] didrocks, just the test [09:19] what will be broken exactly [09:19] hum [09:19] are you 101% sure? :) [09:19] if so, I'm happy to promote :p [09:19] :D [09:19] (well, considering for promotion) [09:19] didrocks, yes, I am [09:19] Saviq: trusting you then! thanks a lot [09:19] * didrocks hugs Saviq [09:19] * didrocks is a lot happier :p [09:19] Phew! [09:19] \o/ [09:20] didrocks, https://code.launchpad.net/~macslow/unity8/unity8.fix-test_sd_incoming_call/+merge/203567 [09:20] didrocks, that commit fixed the test, but was backed out yesterday [09:22] /bbiab [09:22] Saviq: perfect! I wasn't even asking you for the commit to prove it's only a test change without any effect on runtime, I trust you enough to take your words :) [10:09] Saviq: ok, sounding good for the image, we are doing dogfooding [10:09] as no QA around [10:09] and sounds we can promote [10:09] then, we'll just add your unity8 fix as a priority reland [10:09] (we prioritize the reverted components first) [10:15] stupid irc... [10:18] dednick, freenode is bitchy today [10:19] Saviq: just had to change from irc.freenode.net to chat.freenode.net and seems to work now. [10:19] didrocks, sure, will we land all that's in queue or would you want to land just the fix? [10:26] Saviq: as you prefer, if you land everything, you are taking more risk I would say [10:29] greyback: ping [10:30] tsdgeos: pong [10:30] greyback: i'm trying your "do not draw branch" [10:30] and i can't see the difference [10:31] am i supposed to see less " Breakdown of frame time;" calls [10:31] or shorter ones? [10:31] tsdgeos: much less calls [10:31] also on desktop? [10:32] tsdgeos: probably not as noticeable, since the lenses are already loaded on desktop when you start unity8.. But on device, it's very noticeable [10:33] i'm actually getting more calls on the desktop :D [10:33] 126 vs 117 [10:33] let me see the phone [10:34] didrocks, not with the set of changes we have there, most are not even in use currently (prep for new scopes) [10:40] greyback: how did you test it on the phone? i.e. which modification you did to run_on_device? [10:40] it seems i'm not passing the envvar correctly [10:40] did you run manually without using run_on_device? [10:40] * tsdgeos tries that [10:41] tsdgeos: manual run [10:44] greyback: manual run gives me 345 vs 316 [10:44] is that something like you were getting? [10:45] mzanetti, yay [10:46] Saviq: no... no yay [10:46] mzanetti, ? [10:46] Saviq: znc still won't connect [10:46] mzanetti, well, at least you connected [10:46] (connectes fine to canonical's server tho) [10:46] mzanetti, got a znc account ready for you here if you'd want one [10:46] well... that's not really the solution [10:46] but thanks for the offer [10:46] mzanetti, I know, that's why I said "if you'd want one" [10:47] interestingly I get Connection refused now instead of timeouts [10:50] tsdgeos: I wasn't counting, but the difference was noticeable [10:50] i see [10:50] tsdgeos: I'm not unlocking, I just start unity8 and watch [10:55] ok [11:03] greyback: nexus4? [11:03] tsdgeos: nexus10 I tested on [11:05] this is weeeeird [11:05] i added [11:05] onFullyCoveredChanged: console.log("MOOOOOOOOOOOOO", fullyCovered) [11:06] and it never shows anything ?¿ [11:09] greyback: my greeter does not have a fullyshown property ?¿ :S [11:10] tsdgeos: "fullyShown" ? [11:10] i added this [11:10] Component.onCompleted: console.log("BAAAAAAAAAAAAAA", fullyCovered, gree [11:10] ter, greeter.fullyShown) [11:10] and i get [11:10] BAAAAAAAAAAAAAA false Greeter_QMLTYPE_97_QML_209(0x47becd38, "greeter") undefined [11:11] and grepping i don't see how it can be defined for you [11:11] seems fullyShown is only there for stages [11:11] which the greeter is not one [11:15] tsdgeos: damn, I swore that fullyShown was part of Showable. I must be wrong [11:17] i can be wrong too, has happened before you know ;-) [11:18] tsdgeos: well reading the code, I think you're correct. I'm more confused about what I was finding last night... [12:13] mhr3_: pstolowski: Saviq: what am i supposed to do with the "id" field of a type: "actions"? [12:14] tsdgeos, it's just an id picked by the scope [12:15] tsdgeos, perhaps set objectName to it? [12:16] tsdgeos, activation api expected id of a button back; it will also expect that actions id (not yet in the api) [12:17] so i'll have to call something back [12:17] ok [12:17] but that doesn't exist yet [12:17] tsdgeos, yeah [12:17] tsdgeos, it exists, but takes just one id (button); it will be enhanced with one more id [12:18] (actions) [12:18] pstolowski, why? do we want to support non-unique action IDs for some reason? [12:18] mhr3_: pstolowski: Saviq: also can you confirm that one type: "actions" with two elements in the "actions" array is a combo with two entries and not two buttons? [12:19] i.e. if you want two buttons you need two type "actions" elements? [12:19] tsdgeos, confirmed [12:20] Saviq, widget_id + action_id [12:20] mhr3_, yeah, why? shouldn't action_id be unique already? [12:20] there's nothing forcing it to be unique [12:20] Saviq, to be on safe side? otherwise yes, they would need to be globally unique [12:21] mhr3_, there's nothing forcing widget_id + action_id to be unique, either [12:21] mhr3_, pstolowski, I'd go for requiring unique action_id, TBH... [12:21] Saviq, we won't allow pushing widgets with the same id twice [12:22] mhr3_, will you check action_id for uniqueness? [12:23] we could within a widget === Trevinho_ is now known as Trevinho [12:28] Saviq, mhr3_ no strong opinion, just a feeling that perhaps widget_id + action_id is more future-proof... [12:29] mhr3_, pstolowski, I don't really care, feels overkill to me, though === alan_g is now known as alan_g|lunch [12:29] Cimi: where's the comment you mention in https://code.launchpad.net/~unity-team/unity8/unity8.previews_audioPlayer/+merge/203387 ? [12:31] tsdgeos, it should be there, no? [12:31] don't see it [12:32] ah [12:32] see it now [12:33] Cimi: it still has the urls that seem real [12:33] Cimi: and the stop for the audioplayer [12:39] Saviq, i agree with pstolowski, more future-proof [12:43] Saviq: so on the actions thing, how do you think we should expose the "signal" of it being pressed or the combobox changing, add a signal to the base PreviewWidget? === cimi4 is now known as CImi3 [12:44] tsdgeos, I think so, yes, more widgets will actually trigger actions [12:44] tsdgeos, (like the progress widget, the review-input one) [12:45] mhr3_, pstolowski, do we have ideas on how will we pass data back with the action? think a single QVariantHash arg would be ok? === MacSlow is now known as MacSlow|lunch [12:46] Saviq, yep [12:46] i mean... widget_id + action_id + variantmap [12:46] tsdgeos, so the signal signature would be: (string widgetId, string actionId, var data) [12:47] Saviq: "activate"? "activated"? [12:47] Saviq yeah, as mhr3_ said [12:48] tsdgeos, the Qt way - activated, or triggered, maybe? [12:48] tsdgeos, please decide the name of the property :) [12:48] tsdgeos, for focused [12:48] triggered sounds ok [12:49] Cimi: as the comment says, i am not sure we even need a name [12:49] Cimi: why can't you use visible? [12:49] tsdgeos, because the preview might still be visible, but no longer focused [12:50] tsdgeos, when you're switching previews, for example [12:50] Saviq: but you said we're killing the switching of previews [12:50] tsdgeos, no no no [12:50] tsdgeos, we're killing the bottom "see-through" [12:50] ah [12:50] ok [12:50] tsdgeos, you will still be able to swipe between them [12:51] /food [12:51] Cimi: isCurrent then, focused is too close to focus to me [12:51] mzanetti, why we have on the music preview [12:51] or maybe Saviq has a better suggestion for it [12:51] mzanetti, to stop audio on destruction? [12:51] Cimi: why not? [12:52] mzanetti, isn't automatic? [12:52] hmm... probably not, otherwise it wouldn't be there I'd say [12:53] need to check the code. one minute [12:53] tsdgeos, fine by me, it should actually be previewFocused / isCurrentPreview or something [12:53] tsdgeos, as it's not about the widget being current / focused, but the whole preview [12:57] Saviq: right, which also shows we're having too many properties in every widget instead of a "global" one per preview, but that's the price you pay for decoupling :D [12:57] food! [12:57] tsdgeos, well, sure, if we put it all in one .qml, we wouldn't need it indeed [12:57] Cimi: right... might actually be a bug in the qtmultimedia backend [12:58] Cimi: 'cause the playback is handled by the media service [12:58] Cimi: so destructing it should implicitly send a stop to the media service, probably it doesn't [12:58] Cimi: I remember having talked to jhodapp about this when implementing it [13:10] Hi guys [13:11] Saviq: would you like me to add tests for title alignment, or is this "look and feel" like you considered something a while back? [13:12] I rebased it on the branch you mentioned [13:12] lp:~unity-team/unity8/new-scopes-title-alignment [13:14] tsdgeos, ^ [13:15] dang, I should have added a prerequisite branch [13:15] karni, test for CardTool, yeah [13:15] karni, should be a simple one, too [13:15] Saviq: ignore that MP for now [13:16] I think my bzr config went haywire, changed username [13:16] karni, no need to reject all the MPs, just make them WIP and resubmit then if needed === dandrader is now known as dandrader|afk [13:16] Saviq: ack [13:16] * karni moves to different building [13:19] tsdgeos, if we could do attached props in QML, that'd help some [13:20] mhr3_, pstolowski, how do we communicate the number of columns for a preview? [13:21] Saviq, working on that, but it'll be a prop on the PreviewStack object you get from scope.preview() [13:23] Q_PROPERTY(int widgetColumnCount READ widgetColumnCount WRITE setWidgetColumnCount NOTIFY widgetColumnCountChanged) [13:23] good enough? [13:23] mhr3_, how do we do the splitting in case the scope didn't define it? [13:23] Saviq, that is a good question [13:24] ideas? [13:24] anpok: hey, saw your comment on the bug. what exactly does it mean? i.e. where does the 16ms value come from? [13:25] Saviq, doesn't help that the scope can keep pushing more widgets any time it wants [13:25] mhr3_, indeed [13:25] Saviq: hm, you mentioned there's a landing plan for the Unity8 regression fix? [13:25] Saviq, i'm leaning to doing something stupid... 1 widget per column, everything else into last col [13:25] sil2100, yup, waiting for silo [13:25] sil2100, row 20 I think [13:26] mhr3_, I have two ideas: a) you expose all configured models at the same time as an array, and when there isn't one that matches the current setup, shell takes over and does its thing [13:26] Saviq: oh, hm, I think we cannot re-use already Landed silos [13:26] sil2100, meant row 40, sorry [13:26] Saviq: can you just slap the MR to a new row? Since I don't even know whic MR it is [13:27] sil2100, I'd rather land it all together, don't want to waste time on separate regression fix and actual code that's waiting for almost a week now :/ [13:28] mhr3_, b) you call into the shell when receiving a preview (and/or subsequent preview widgets) and the shell comes back with lists split into columns [13:29] mhr3_, b) is ~same idea as plugin calling into shell to potentially apply any rules / changes on template / components [13:29] Saviq: ok, just remember that if we find a regression again by accident (;p) we'll revert everything again ;p [13:29] sil2100, I'm fine with that [13:29] Saviq: but if you're ok with that risk then I prepare stuff for it [13:29] sil2100, I'm completely ok with you guys reverting us if there's a regression identified === popey_ is now known as popey [13:29] Saviq, i don't like either :) [13:30] mhr3_, other (better) ideas? === dandrader|afk is now known as dandrader [13:35] Saviq, i'm leaning to forcing scope devs to provide the layouts, so if a scope wouldn't specify a layout, shell would display everything in one col === alan_g|lunch is now known as alan_g [13:36] Saviq: silo assigned! It's getting prepared this instance [13:40] mhr3_, future-proof you said? [13:40] sil2100, thanks [13:40] Saviq, touche [13:41] Saviq, my primary worry is how would things look like if a scope pushed one widget, then another one 500ms later and then 10more another 500ms later... why just can't make informed decision if we want to display things as fast as possible [13:41] s/why/we/ [13:42] mhr3_, of course, I think there's two cases: preview creation and widget addition (removal not possible, right?) [13:42] right [13:42] mhr3_, so when preview is "lazy" with adding widgets, and does not provide a layout [13:43] mhr3_, we'd only append widgets at the end of columns [13:43] columnS? [13:43] like randomly...? [13:44] mhr3_, not randomly, but based on some rules [13:44] mhr3_, say, if there's art in the first column, we wouldn't add anything more there [13:45] mhr3_, otherwise we might try some more fancy determination [13:45] so you'd try to do something like the horizontal journal is doing? [13:45] mhr3_, something of the sort, yeah [13:46] mhr3_, but not based on actual rendered widgets, just their type and props [13:46] mhr3_, but I'd rather the plugin not be part of that [13:46] which means we'd need to notify the shell with about-to-be-added widget [13:46] mhr3_, yes [13:47] did i mention the data for those widget might not be present immediately? [13:47] mhr3_, that's fine, I think [13:47] mhr3_, again, I don't want to base the decision on actual data [13:47] Saviq, that means the props will be pretty much empty [13:47] mhr3_, if something's not there, it's not like we can do anything about it [13:48] mhr3_, what I'm saying is I'd like to make a semi-informed decision as soon as a widget becomes "known" to the system [13:48] mhr3_, it will be a corner case I think, that we'd get a widget with no data [13:48] oh well... you can surely make better decision than the plugin [13:48] mhr3_, but anyway its type would be enough information [13:48] mhr3_, exactly [13:49] but not thrilled about having to call into js [13:49] how does that even work [13:49] ? [13:49] mhr3_, good question ;), but it can be as simple as a property that the shell sets [13:50] mhr3_, after you trigger a signal [13:50] mhr3_, problem with that would be that it'd be required to do it [13:50] aren't signals going to js asynchronously? [13:50] mhr3_, they are [13:50] well.. that's a problem [13:50] mhr3_, so yeah, I think we'd rather do sync, wouldn't we [13:51] i basically need return value from js [13:51] as in "int giveMeColumnNumForWidget(widget)" [13:52] mhr3_, QMetaObject::invokeMethod [13:53] mhr3_, i.e. you'd have a property (default null) on Scopes [13:53] mhr3_, say, previewProcessor [13:53] mhr3_, if it's non-null, you'd call a method on it with some data [13:53] i can already see the lost frames :P [13:53] mhr3_, and use the output [13:53] mhr3_, it will be maybe 20 lines of JS at the mos === _salem is now known as salem_ [13:54] mhr3_, we're running much more js on every frame than that ;) [13:54] it's just awful :P [13:54] anyway... /me lunch [13:54] mhr3_, we can always do vala, if you'd rather ;P [13:55] mhr3_, anyway, we can implement the processor in C++ [13:55] mhr3_, and use the same mechanisms [13:55] slightly better [13:55] now get rid of the QMetaObject and i'm happy [13:55] :) === MacSlow|lunch is now known as MacSlow [13:55] mhr3_, wanna rewrite unity8 in C/vala/C++? be my guest [13:56] let's see how that goes :P [13:56] Saviq, i'll think about it :P [13:56] mhr3_, enjoy lunch [13:59] mhr3_, http://qt-project.org/doc/qt-5.0/qtqml/qtqml-cppintegration-interactqmlfromcpp.html#invoking-qml-methods BTW [14:05] sil2100, "In silo landing-005. Can't build: Some projects are missing revisions from distribution." [14:05] didrocks, ↑ [14:05] what do I do now? [14:06] Saviq: ah! Yes, so you need to check a checkbox, let me see the naming [14:06] sil2100, will do [14:06] sil2100, FORCE_REBUILD it seems [14:07] Saviq: yes, since we don't want to have the changelog entry for the revert [14:07] karni, drop the TODO, it's a general thing we'll need to tackle together [14:07] ok [14:08] karni, I think letting JS "booleanize" undefined / null / "" would be good enough [14:08] karni, in isOnlyItem [14:09] karni, what is components["art"]["image"]? [14:09] sil2100, hi, would you mind reviewing that? https://code.launchpad.net/~dandrader/ubuntu-keyboard/outOfSourceBuilds/+merge/204292 [14:09] Saviq: I wasn't sure how to implement that correctly, I'm still not confident in those constructs [14:09] Saviq: seems the naming was clear enough? \o/ [14:10] didrocks, the message said "force rebuild", so yeah ;) [14:10] Saviq: the image of the art. I assume it's sufficient to test ["art"] only? [14:10] karni, there is no "image of the art" [14:10] karni, and remember you're looking at the configuration, not at any particular result [14:10] karni, and why would we always left-align when art is there? [14:11] sil2100, I've picked you as you name pops up in "bzr log" and tmoenicke is no longer around... [14:11] Saviq: I'm sorry if I didn't make this clear - I'll make it work in progress [14:11] dandrader: sure, let me take a look [14:11] karni, I don't think art has any bearing on center vs. left alignment... [14:11] karni, also, because components might be: [14:11] "art": "foo" [14:11] as well as: [14:12] "art": { "field": "foo" } [14:12] Saviq: It does. if there's art and it's horizontal, it should be left aligned, if there's no art, it should be center aligned [14:12] let me find the spec [14:12] karni, right, you need to differentiate between horizontal and vertical, then [14:12] Cimi: there's still one "../../tests/qmltests/Dash/Music/data/testsound1.ogg" left [14:13] karni, the condition should be: [14:13] subtitle && (typeof subtitle === "string" || subtitle["field"]) [14:13] ack, thank you :) [14:13] karni, so, you check that subtitle is defined at all in the first place (and non-empty) [14:13] * karni nods === mterry_ is now known as mterry [14:14] karni, and then it needs to be a string, or subtitle["field"] needs to be non-empty [14:14] Saviq: Cimi: besides that one ogg path still looking like it should exist, it looks ok to me so i've approved https://code.launchpad.net/~unity-team/unity8/unity8.previews_audioPlayer/+merge/203387 want me to top approve or want to have a second approve? [14:14] tsdgeos, I updated the path [14:14] tsdgeos, top-approve away [14:14] tsdgeos, maybe I can do [14:15] /path/not/existing [14:15] Cimi, "../../tests/qmltests/Dash/Music/data/testsound1.ogg" [14:15] let me do that [14:15] Saviq, I pushed :| [14:15] Cimi: you updated 3 of them, forgot 1 [14:15] Cimi, tracksModel1 [14:15] or launchpad is behind [14:15] Cimi, or that ↑ [14:15] Cimi, just drop the extensions, it really doesn't matter what you put there [14:16] Saviq, ok [14:16] sorry guys, was killed by headache till 10 mins ago [14:16] Cimi, tsdgeos, I wonder, should we update the duration once the stream is loaded? in case the scope lied? [14:16] since 7am [14:16] :( [14:16] Cimi, you need to fix your head! ETOOMANYHEADACHES [14:17] Saviq: i was thinking that the other day too [14:17] don't know tbh, i'd trust the scope [14:17] but i'm open to both [14:17] tsdgeos, we don't display the current time anyway, do we [14:17] there's also the question if we should autoload or not [14:17] Saviq: no, just progress [14:18] tsdgeos, ok so unless people use stopwatches or somethin' we'r [14:18] e good [14:18] yep [14:18] let's leave it be [14:18] Saviq, don't ask me why [14:18] Cimi, please put */ on separate line [14:18] Saviq, I did work this morning, but I wish I was without headache [14:18] Saviq, ok [14:19] Cimi, London doesn't agree with you ;) [14:19] Cimi, you Italians need more sun ;) [14:20] Saviq, yes, glad I'm going to mwc [14:20] Saviq, I'll be working there also the week after [14:21] Saviq, rented flat with alex chiang [14:21] Cimi, cool [14:22] Cimi, docs are wrong [14:22] widgetData["source"] widgetData["title"] widgetData["subtitle"] widgetData["length"]. [14:22] those don't exist [14:22] dammit [14:22] you're right [14:23] Cimi, it's enough to say: it expects each element in the array to be: { "source": "url://", "title": "Title", "subtitle": "Subtitle" // optional, "length": 120 // in seconds } [14:25] Saviq, or you prefer http://paste.ubuntu.com/6873262/ [14:25] ? [14:25] Cimi, that's still wrong ;) [14:25] Cimi, it's missing the index in there [14:25] Cimi, let me check one thing === rachelliu_ is now known as rachelliu [14:30]