[08:40] mzanetti: ping [08:41] mzanetti: i'm getting strange results from the unitytestcase::findChild function. [08:41] I added a console log to print the array each loop. http://paste.ubuntu.com/8112553/ ever seen something like that? [08:42] mzanetti: find invisible seems to work though [08:42] dednick: maybe a loop in some item.children? [08:42] dednick: is this your recursive qml hack? [08:46] mzanetti: it is, but same if i remove it [08:47] weird... [08:47] dednick: well, no, I haven't seen that before... [08:47] dednick: let me know if you need me to have a closer look. right now I can't tell much [08:48] mzanetti: hm. it is the children var though. if i remove that it works [08:48] mzanetti: it's a model [08:49] dednick: ah, so you have a var named children? [08:49] mzanetti: yes [08:49] dednick: ok... findChild() does item.children [08:49] dednick: the qml parent mechanism works through a variable named children [08:49] mzanetti: ah. doh [08:49] so you might not want to use that name [08:49] stupid [08:50] heh. yeah :) [09:08] tsdgeos, can you merge scope settings? https://code.launchpad.net/~aacid/unity8/category_view_invisible_in_preview_mode/+merge/231844 [09:08] tsdgeos, as well as adding tests [09:09] Cimi: eh? [09:09] tsdgeos, https://code.launchpad.net/~cimi/unity8/scope-settings [09:09] which test do you want me to add? [09:09] tsdgeos, that it turns visible :) [09:10] tsdgeos, just add a visibility check in openPreview/closePreview and settings [09:10] tsdgeos, one liner [09:11] tsdgeos, looks like you forgot a prerequisite, too, since silo 16 didn't land yet [09:11] and won't, until Daniel shows up :| [09:11] damn [09:11] nah i just had a polluted "clean" unity8 [09:11] for some reason [09:11] Cimi: not going to merge it for now [09:11] *or* I strip the qtmir changes out from the silo... [09:11] maybe that's more productive actually [09:12] Cimi: i'm in my world of trying to make stuff a bit faster, will care of merging when it's merged :D [09:13] Cimi: i'll add a test, good idea [09:20] tsdgeos, if you don't merge scope settings I will have conflicts === Trevinho|holiday is now known as Trevinho [09:22] well, let's just get scope settings merged first hten [09:22] and i will have conflicts [09:28] lol [09:28] making the thing non visible [09:28] changes its content height [09:28] and stuff breaks [09:28] that is totally unpredicted [09:28] greyback, bad news in the morning... had to pull qtmir changes from the silo - most probably lifecycle broke trusted prompts [09:29] greyback, landing just unity8 changes now, will recreate the silo with qtmir back after that [09:29] Saviq: ok [09:58] Cimi, https://code.launchpad.net/~saviq/unity8/header-customizations/+merge/230719/comments/564353 [10:00] Cimi, had a thought... for the bottom part of the divider highlight [10:00] Cimi, we could have a shader for one pixel and stretch that [10:01] Saviq, what's the advantage? [10:02] Cimi, I've not a good idea how to come up with the right colour for it... [10:02] Cimi, without introducing nasty apis [10:02] Cimi, but let me try that first [10:02] Saviq, the current color is not right? [10:02] Cimi, it is [10:02] Cimi, but not when there's departments [10:02] Saviq, I stop you, shading one pixel is too risky, we need more samples [10:03] Saviq, what if we have one bg image with some details there (or a origami efect or whatever) [10:03] Cimi, well, that only depends whether we want to support the highlight to be non-uniform [10:03] Saviq, it might make the shader change [10:03] you see the highlight changing color [10:04] Cimi, well, with the *other* solution, I only go for a flat colour too [10:05] Saviq, but this does not change [10:05] Cimi, it doesn't look at the image at all, either [10:05] Cimi, so what colour do I come up with then? [10:06] Saviq, either you sample more than 1 pixel, like 1gu [10:06] at least [10:06] Cimi, you know well that that's just as prone to failure as anything else [10:06] Cimi, what if that one gu at the top left is black [10:07] Saviq, so again, this is broken [10:07] Cimi, but just next to it something starts happening [10:07] Saviq, that's why is not a big improvement over the current solution [10:07] Cimi, yeah, so what then? full shader for just that stupid highlight? [10:07] Saviq, so let's leave as it is :) [10:08] Cimi, it can't be left as is [10:08] Saviq, or just sample 1gu in the middle [10:08] Cimi, I need to do *something* [10:08] ahah ok [10:08] :) [10:08] Cimi, and either I support an image or not [10:08] Cimi, if I don't, I might as well just sample the top 1 px (or find out the top color from the background definition) [10:09] Cimi, if I do, it needs to be a full shader [10:09] Cimi, truth be told, if it *is* an image, maybe we just require the highlight to be built into it? [10:09] and just not do anything [10:10] Cimi, I think that might really be the best thing to do [10:11] Saviq, we want a solution that we can port to the sdk [10:11] Cimi, not here we don't necessarily [10:11] Cimi, that navigation thing does not exist in the sdk [10:11] Cimi, and if it did, we don't know if it would support images anyway [10:12] Cimi, *and* if it did, if you wanted an image, just highlight it yourself [10:12] Cimi, it's not like we're that flexible in the SDK [10:14] Cimi, ah dammit, it's just freakin' 4 pixels high or something [10:14] ahahahah [10:14] Cimi, you are a bastard for forcing me to do this :P [10:14] Saviq, I like you being perfectionist :) [10:15] Cimi, *I'm* not, not really, I wanted a compromise :P [10:21] oh... only 19 approved branches [10:21] seems a review a day just moves the queue to another place :D [10:23] zbenjamin: hey, regarding this: https://code.launchpad.net/~zeller-benjamin/unity8/scope-url/+merge/231749 [10:24] mzanetti: yes [10:24] zbenjamin: is performQuery really what you want? [10:24] mzanetti: thats what Saviq told me to use [10:24] in other words, does this what you want? [10:25] mzanetti: well when it shows the scope , it does [10:25] mzanetti: i could not test it because it crashes unity, Saviq told me to upload the branch so he can have a look [10:26] ah ok [10:27] mzanetti, yeah, that should be what he wants, wasn't able to get on it yet [10:28] I would have expected to call goToScope() [10:28] but I realize that might not work for non-favorite scopes [10:29] soooo, i just discovered we should not make height depend on visible, otherwise when you hide a whole root because you don't want to show it [10:30] the inner parts get all the sizes changed and you get lots of updates for no reason [10:30] you were trying to save rendering and suddently all your nodes moved around [10:30] tsdgeos, ugh [10:30] and we do that a lot :D [10:46] Cimi, ok, task for you: find out a shader line I need to use to calculate the highlight color :P [10:51] Saviq, the line where the shader needs to be? :) [10:51] the pixels under the divider? [10:52] Cimi, no, the actual shader code [10:52] Saviq, same thing we have now [10:52] Cimi, there is no .lighter() in glsl [10:52] Saviq, lighter is an algorithm [10:52] Cimi, yes, find it [10:52] Cimi, find out how do I, in shader code, do .lighter(1.2) [10:54] ok [10:54] Saviq, convert to HSL [10:54] Saviq, multiply s and l for that argument [10:54] Saviq, IIRC [10:55] Cimi, there is no .convertToHsl in gsls [10:55] Saviq, which colorspace we have? [10:55] Saviq, frb? [10:55] Cimi, rgb [10:55] *rgb [10:55] ok we need to convert to hsl [10:55] Cimi, and .lighter in Qt does hsv, and v * argument [10:55] Cimi, nothing with s [10:55] you can do hsv then [10:56] Saviq, changing saturation as well is cool [10:56] better fake with a lightness vector in rgb [10:57] http://en.wikipedia.org/wiki/Luminance_(relative) [10:58] anpok, yeah, howto? :) [10:58] Saviq, or http://css-tricks.com/snippets/javascript/lighten-darken-color/ [10:59] 1.2 is a lightness factor? [11:00] or increas in luminosity [11:00] anpok, that's what .v gets multiplied by [11:00] anpok, in hsv that is [11:03] fragcolor * mat4(1,0,0,0.2126,0,1,0,0.7152,0,0,1,0.0722,0,0,0,1)*1.2; would be my guess. [11:04] + additional fiddling on the vector, gamma correction,... [11:04] oops [11:04] the other way around ofc [11:04] * anpok hides in shame [11:07] anpok, yeah, that became green/yellow from black :) [11:07] anpok, but you have to bear with me here, I've no idea what I'm doing :) [11:11] bregma, hey, you around? [11:12] greyback, did you manage to get a unity8 desktop session on your machine in the end? [11:12] Saviq: yes, I've had it working some time now [11:13] whenver it lets me log in ofc [11:13] greyback, could you test silo 16 (`citrain host-upgrade 16`) for bugs #1350878 and #1353041 ? [11:13] bug 1353041 in unity8 (Ubuntu) "Logging out of the Unity8 desktop preview session hangs" [High,In progress] https://launchpad.net/bugs/1353041 [11:13] bug 1350878 in unity8 (Ubuntu) "lockscreen doesn't properly unlock on desktop when a password is set" [High,In progress] https://launchpad.net/bugs/1350878 [11:13] Saviq: on it [11:13] greyback, thank youse [11:15] are special qt libs necessary to run unity8 on desktop? (GL vs GLESv2) [11:16] anpok: nothing special no [11:19] Holas [11:19] o/ [11:21] hola Saviq :) [11:22] Saviq: who wrote the "just run with "QSG_VISUALIZE=batches" and scroll in the dash. You see lots of colours and they change on scroll - which means Qt has to re-batch them each frame, which is very slow. https://bugs.launchpad.net/ubuntu-ui-toolkit/+bug/1350863" part ? [11:22] Ubuntu bug 1350863 in Ubuntu UI Toolkit "UbuntuShape incompatible with Qt Scenegraph renderer batching strategy" [High,Fix released] [11:22] anpok, so, pointers on how to make the lighting happen? You saying "gamma correction of course" didn't really register ;) [11:22] tsdgeos, greyback [11:22] I think [11:23] tsdgeos: yes it was I *big reveal* [11:23] greyback: i don't see it happening [11:23] tsdgeos: it was fixed [11:23] ah [11:24] right i should read what ubot5 says [11:25] tsdgeos: you can still see a bit of it in the dash today, but for other reasons I suspect [11:25] greyback: it's because the "dash overview thing in the bottom" changes opacity [11:25] causing the scene graph to rebuild itself [11:26] tsdgeos: yes that would do it, but you see it even just scrolling around. (this was me trying 3 weeks ago though, it may have improved) === MacSlow is now known as MacSlow|lunch [11:27] greyback: the "dash overview thing in the bottom" changes opacity while scrolling [11:27] i.e. it deppends if your oon top or not to show itself or not [11:27] tsdgeos: ah I see. I would've expected that to be in a separate batch though, it shouldn't cause the whole SG to be rebatched [11:28] it does [11:28] ouch [11:28] greyback: http://paste.ubuntu.com/8113535/ [11:30] so it appears. [11:32] Saviq: fragcolor.r += fragcolor.r*0.21*lightness_factor; fragcolor.g += fragcolor.g*0.72*lightness_factor; fragcolor.b += fragcolor.b*0.07*lightness_factor; [11:33] anpok, thanks! [11:33] Saviq: with gamma I meant removing gamma correction before applying the ligthness and reappling it afterwards [11:33] not sure how to do that efficient in ges [11:33] *glsl [11:34] factor is maybe the wrong word .. btw it is rather a lightness translation .. means 0 is no change.. [11:35] anpok, ok thanks [11:36] Saviq: do you remember the things that broke here? http://paste.ubuntu.com/8113598/ [11:38] tsdgeos, the sizes were not calculated right [11:38] tsdgeos, because if invisible, text wasn't laid out [11:38] which may be fixed in my "don't use visible to calculate size" thing? [11:38] i've made it non visible now and can''t find any issue [11:38] was it obvious everywhere? [11:40] let me try it in an unpatched one [11:40] :D [11:40] * tsdgeos slot sometimes [11:41] slot -> slow [11:41] oh yes it breaks [11:47] Saviq, tsdgeos I have an issue with test Card [11:47] the variables title, art, etc etc [11:47] inside the testcase [11:47] they don't get updated quick enough when I change index [11:47] tsdgeos, well, the calculations were wrong is all [11:48] Cimi, all the other tests seem to manage? [11:48] so at the beginning of a new iteration of a test, I might have the old title item [11:48] Saviq, all other tests seem not to use those [11:48] tsdgeos, not sure the test actually covered this issue [11:49] Saviq, I started having this issue using GRID_UNIT_PX different than 8 [11:49] mzanetti, is waitForRendering enough? [11:50] Cimi: hmm... isn't that already there? [11:50] mzanetti, it was [11:50] mzanetti, not helping [11:50] a wait(50) helps [11:50] Cimi: not happy with wait(50) [11:50] I tried deactivating the loader at the end of a test [11:50] no joy [11:51] Cimi: right... you could set the vars to null in cleanup() [11:51] Cimi: then set them again in init() [11:51] zbenjamin, did the url dispatcher work for you on desktop? [11:52] Saviq: i tried only on the phone [11:52] zbenjamin, k [11:52] mzanetti, not helps [11:52] mzanetti, there is also cardTool in between [11:53] mzanetti, giving the sourceComponent to the loader [11:53] in few words, this testCase is asking for races [11:56] zbenjamin, hmm I wonder how it even got to your dash... unity8-dash isn't an u-a-l-launched application... so not sure how url dispatcher could get to it :/ [11:58] * Saviq thinks we need to make it an u-a-l launched one... [12:00] Saviq: ok, did it not happen for you at all? [12:00] zbenjamin, yeah, url dispatcher just throws stuff around and fails, it never reaches unity8-dash [12:02] zbenjamin, from dbus-monitor: http://paste.ubuntu.com/8113738/ [12:02] zbenjamin, it tries to launch unity8-dash as an app under u-a-l and that fails [12:03] Saviq: ok :/ [12:03] zbenjamin, I'll have a chat with tedg today how to clear this up [12:03] Saviq: so its not so trivial after all.. [12:03] Saviq: ok thx very much for helping with this [12:03] zbenjamin, it should be possible to make unity8-dash being wrapped with u-a-l, which would also give us lifecycle for it [12:04] Saviq: yes, sounds perfect to me [12:04] wonder why it does fail actually [12:09] +1 on landing just the unity8 stuffs [12:09] ...and i was really suspicious of that pin lock [12:12] greyback, any feedback on desktop silo 16 yet? [12:13] Saviq: my test machine is a bit on the slow side, it's almost finished updating [12:13] greyback, ah [12:15] kgunn, yeah, but I was almost sure we did the same for ap tests that we do for ./run.sh, obviously not, but should be an easy fix [12:17] sure === MacSlow|lunch is now known as MacSlow [12:39] Cimi, merge prerequisite into scope settings please, there's conflicts or at least --weave is required [12:39] hmm [12:39] *or* there's simply a conflict [12:43] * tedg is confused, why is URL dispatcher touching the dash? [12:43] Saviq: yeah there's still something wrong [12:43] Saviq, ^ [12:44] tedg, because we want it to support scope:// urls [12:44] Oh [12:44] Hmm, yeah, we can't use the standard mechanisms for that. [12:45] tedg, well, we could, if unity8-dash would be ual-launched [12:45] Saviq, Yeah, but that's a bad idea :-) [12:45] tedg, why? [12:45] tedg, it's just an app these days, only special thing about it is that it starts automagically and respawns [12:46] Because UAL does a bunch of stuff that you don't need and doesn't do a bunch of stuff you want. Setting up environments vs. respawn [12:46] tedg, right, so that is what I wanted to talk to you about ;) [12:47] I guess if respawn is the only feature you care about, you could do that manually. [12:48] tedg, or have a task on application APP_ID=unity8-dash stopped ;) [12:48] tedg, that would start it up again [12:48] Exactly [12:48] and on unity8 started [12:48] tedg, so, problem is that we have NoDisplay=true [12:48] tedg, that seems to make application-legacy unhappy :/ [12:49] Yes, because that means it's not an application :-) [12:50] I guess what I don't like is that I like the idea that "application" means something. It's not just "process". I worry about blurring that definition. [12:50] Application means it has an icon in the apps-scope, means it shows up on the launcher, etc. [12:50] tedg, could application be a wrapper around process then? ;) [12:51] It is, and the process manager is Upstart. [12:51] So that's why it feels more right for me to have the dash be an Upstart job. [12:51] tedg, ok then, how do we hook up url-dispatcher to the non-application dash then? [12:51] mzanetti: I trying to get to the pin unlock dialog still later today [12:51] * Saviq wanted lifecycle, too :/ [12:51] I think that we special case the scope URL. [12:52] although I've been up for 29h straight, so let's see how it goes [12:55] ChrisTownsend, oh, just the man I wanted to see [12:55] Saviq, I think the idea that an "application" is wrapper around a process just gave me a slide for my presentation in two weeks on confinement :-) [12:55] ChrisTownsend, care to test out silo 16 for unity8 on desktop (log in and log out mostly)? [12:56] Saviq: Hey, yeah, sure I can do that. [12:56] Saviq, So I guess I'm undecided, I definitely see your point, but I like the idea of application being smaller. [12:57] tedg, so the only thing that I see as not possible right now [12:57] tedg, is to launch an app that has NoDisplay=true, which I'm not sure I agree with [12:57] Saviq: both bugs are fixed [12:57] tedg, we have a launcher for the media player in dash apps [12:57] Saviq: the desktop session ones [12:57] greyback, awesome, thanks [12:58] tedg, which is useless, 'cause just says "you didn't pass a file, I'm done" [12:58] Saviq, Yes, media player is an interesting case, because it, for instance shows up on the launcher, what if you pin it there? [12:58] tedg, sure, you pinned it there, doesn't mean it should be *listed* as an app [12:58] dunno [12:58] Saviq, It seems to me "if it's an app, it should be an app" so it needs to fix that experience or become a trusted prompt session over the app playing the video. [12:59] tedg, right, or part of the app simply [12:59] tedg, as I said, "I'm not sure" [13:00] The problem, of course, with me talking about this is then you can easily ask "what is an application then?" and that's harder to answer :-) [13:00] Saviq: greyback: Regarding desktop logout, there is still the issue where it kicks you back to the Unity8 Greeter, but that is a separate issue. [13:00] ChrisTownsend: yep I reproduced that [13:00] ChrisTownsend, so long as you *can* log in and out, we're in a much better place than we were ;) [13:01] ChrisTownsend: I guessed it was some upstart job mis-behaviour [13:01] Saviq: greyback: If I add some pre/post-stop upstart directives that were originally in the unity8-desktop-session upstart stuff, then logout works. But I'm not sure if those break the phone. [13:01] Saviq: Exactly! [13:01] ChrisTownsend, well, there is not a "log out" option on the phone [13:01] ChrisTownsend, so unity8 basically never gets stopped there [13:02] ChrisTownsend, so from a first glance it looks like it could just be there [13:02] Saviq: Oh, well then, I'll propose a fix and you guys can ack/nack it. [13:02] ChrisTownsend, *but*, I must say I dislike the idea of unity8's post-stop job stopping the session [13:02] Saviq: Yeah, it seems kind of hack-y. [13:02] ChrisTownsend, I'd think this is something that should happen by unity8 asking logind or lightdm or something [13:03] Saviq: Right [13:03] ChrisTownsend, but I've not a good idea still what's the real relationship between all those [13:07] Saviq: Regarding the logout issue, once this new Unity8 lands, I'll enter a new bug, then we can go from there. [13:07] ChrisTownsend, yup [13:08] tedg, ok then, special handling, can we just use UriHandler as everything else does? [13:08] tedg, or does it not work without APP_ID or so? [13:08] (we can get it an APP_ID, too) [13:08] Saviq, Yeah, no the appid processing is only in URL Dispatcher. [13:08] Saviq: Ok, login/logout work now with package from the silo. [13:08] Saviq, Do you want it send via the FD.o interface on DBus? [13:09] Saviq, It'd be easier for me if it could register for a well known name. [13:09] tedg, could do, although that means *we* need to do more work ;) [13:09] tedg, and well, UriHandler registers itself anyway, since it's a singleton? [13:10] tedg, maybe we could make UriHandler accept a path/name as property? [13:10] Saviq, there's conflicts where? [13:10] Saviq, I believe so, on /$(appid) [13:10] Cimi, nowhere [13:10] Saviq, dash overview is in trunk [13:10] Cimi, ignore [13:10] Saviq, I merged trunk [13:10] Saviq, Uhm, probably not good since we don't want most apps registering names, confinement doesn't allow it. [13:11] Saviq, If anywhere in the app you register for a name, everything gets it. [13:11] tedg, can be hidden api ;) [13:11] Cimi, yeah, wrong ordering of branches in the train [13:11] * tedg prefers non-existant over hidden :-) [13:11] Cimi, everything's good [13:11] tedg, same as with NoDisplay=true eh? [13:12] Saviq, Yeah, that's such a BS thing to put in the file. [13:14] tedg, ok, we'll have to come up with a name for you [13:15] tedg, or you can come up with one for us (well, we have com.canonical.UnityDash already, so that probably won't change [13:16] Saviq, Oh, if you've got that already I can probably just use that. [13:16] Saviq, See, no work for you ;-) [13:16] tedg, and UriHandler will just work still? [13:16] Saviq, I believe so, you only get a name per-connection. So as long as you don't have multiple dbus connections it's fine. [13:16] (or that they're using the same one) [13:17] tedg, ok coolz, /me has no idea about dbus paths, names, interfaces, whatnots [13:17] tedg, we just need to make sure to keep it, as when we do move to urls, we could've just dropped that name, but won't, in that case [13:20] Saviq, So then what is the appid you're giving to UriHandler ? [13:20] tedg, you tell me :) [13:20] Saviq, "saviq-rocks" [13:20] tedg, where does it take it from? [13:20] Saviq, I'm not sure, that's the "Qt Magic" part for me :-) [13:21] tedg, in any case, it should be unity8-dash [13:21] I think that loicm did that work. [13:22] * tedg tries to remember [13:22] tedg, that's what we get under ~/.cache/ for example [13:22] as our writable cache dir [13:22] so that must be it [13:22] Probably the same [13:22] Saviq, So then for the URL format you're looking for "scope:///foo" or should there be more/less restrictions? [13:23] scope://* [13:23] tedg, or rather scope://.+ probably [13:23] as an empty scope://... [13:23] well, it can just focus the dash is all [13:23] tedg, so I'm fine with either scope://.+ or .* [13:23] Saviq, The problem with "//" vs "///" is that then the text has to match domain name rules, is that okay? [13:24] tedg, yes [13:24] tedg, the first part is fqdn-like always [13:24] K [13:25] We should call it "unity8-" it seems like the second dash is just being repetitive. [13:27] * tedg thinks this might have to go into kenvandine_'s "Great Naming Strategies by Ted" book. [13:36] :) [13:37] ;) === pete-woods is now known as pete-woods|lunch [13:49] mzanetti, can I add a wait? :) [13:50] greyback, I'm adding lp:~vanvugt/qtmir/support-non-usr-includes to the silo, too? [13:51] Saviq: go for it [14:08] pstolowski: https://code.launchpad.net/~aacid/unity-scopes-shell/resetMeansCountChanged/+merge/231898 [14:09] i spent a while trying to figure out what was wrong in my code to realize for once it wasn't me :D [14:09] mzanetti, http://paste.ubuntu.com/8114532 [14:09] I know is ugly [14:09] works though [14:09] duuude [14:09] no way :D [14:09] mzanetti, listen I have no other freaking idea! [14:09] :D [14:10] I am accepting suggestions if you want :D [14:10] I even tried putting a tryCompareFunction waiting for title to change [14:10] it's a freaking test, we cannot spend hours on it... [14:11] lol [14:11] looks uncool though [14:11] hm i get a runtime_error exception when I start unity8-dash manually [14:11] it says: [14:12] org.freedesktop.DBus.Error.NotSupported: Unable to autolaunch a dbus-daemon without a $DISPLAY for X11 [14:12] anpok: manually in pc or phone? [14:12] tsdgeos, you mean that fix? [14:12] dandrader, unity8 from silo 16 is almost released now, should I build the new silo with lifecycle already or wait for something you've in store? [14:13] Cimi: yep [14:13] Saviq, kgunn, fixed the prompt surfaces in the lifecycle branch. it was a dead-simple one-liner http://bazaar.launchpad.net/~dandrader/unity8/lifecycle/revision/1159 \o/ [14:13]