=== _salem is now known as salem_ === salem_ is now known as _salem === tvoss is now known as tvoss|exercise [07:05] good morning [07:06] hey mzanetti [07:06] Hi. Will Compiz be the standard WM for 13.10 ? [07:07] MCR: yeah, Unity + Compiz will still be the default until Unity Next is up to parity for a desktop experience [07:07] MCR: which realistically won't be the case for 13.10 [07:08] didrocks, thanks for the answer - TBH I do not think anything could replace this combination anytime soon... [07:08] at least not without a ton of regressions [07:08] MCR: we'll, we are only talking about Unity default experience [07:08] not all the compiz crazyness [07:08] and that's our goal to not regress for this [07:08] yeah, sure [07:08] anything that doesn't have ezoom doesn't deserve to be called a replacement! [07:09] hyperair: ezoom was never supported in Unity btw [07:09] so we can only do better there :) [07:09] didrocks: i know, which was why i brought it up. [07:09] didrocks: but compiz contains ezoom, and unity+compiz works with ezoom well enough. [07:10] hyperair: "well enough" != enough IMHO [07:10] didrocks, it works perfectly ;) [07:10] didrocks: where does ezoom fall short? [07:10] if we really want it to be effective, you need to be able to zoom in the UI [07:10] MCR: no it doesn't [07:10] you can't zoom the shell [07:10] ? [07:10] zoom in the UI? [07:10] you can't zoom the dash [07:10] oh that [07:10] yes [07:10] yes [07:10] so not "well" supported [07:11] come to think of it, why did unity need to be a window manager? couldn't it just have been a gnome-panel thing? [07:11] i mean a gnome-panel drop-in replacement [07:11] hyperair: because you need deeper integration to control windows states [07:11] didrocks: what window states does unity control? [07:11] that's why GS and Unity are in the wm process [07:12] yeah but what window states does unity control, exactly? [07:12] hyperair: if maximized or not, where the windows are, which workspace, how many of them [07:12] most of that is outside of the unity plugin [07:12] hyperair: no, it's not, look at the unity code [07:12] because unity needs to know about it [07:12] compiz already had logic for which workspace a window is in [07:12] oh [07:12] so it's more of querying [07:13] rather than controlling? [07:13] right [07:13] with some control to [07:13] "move to ws X" [07:13] or move this windows from ws X to Y [07:13] wasn't there some nice compiz dbus api? [07:13] didrocks: well, Compiz needed until 2013 to learn how to unsnap a horizontally maximized window - in fact Compiz in Raring still doesn't know about it... [07:13] hyperair: "nice" isn't the definitive word [07:13] hyperair: the API is never stable [07:13] didrocks: heh well, it could have been improved? [07:14] hyperair: and always poking dbus is not an option :) [07:14] and stabilized? [07:14] ah yes, true. [07:14] hyperair: people tried, in metacity, mutter, compiz [07:14] Try this: right click on maximize button -> now drag the titlebar [07:14] hyperair: everybody failed [07:14] there is maybe a reason :) [07:14] hmm [07:14] wmctrl? [07:14] maybe dbus is just slow. [07:14] MCR: I'm not interested in knowing where compiz fails, I know there is a lot of cases ;) [07:15] but for all the caching of pixbufs and what other fancy stuff unity does that causes its memory to swell like that... [07:15] i'm not seeing the speed improvements that should have come with it [07:15] gnome shells [07:15] GS's "dash" is much faster than unity's [07:15] hyperair: it does a lot less [07:15] like not showing as many icons, no previews… [07:15] you can't really compare the dash :) [07:15] there's this very annoying bug where i key in stuff, it types out my query *slowly*, and then it spins for a while, and stops spinning without refreshing. [07:16] then i have to press another key, wait for the spinner, press another key, and then it'll work. [07:16] hyperair: did you report a bug about it? [07:16] didrocks: yes, well, super+a should be as fast as GS. [07:16] didrocks: hmm, i don't recall [07:16] maybe i didn't. [07:16] hyperair: RGBA GLX is slow [07:16] that's why it had to be a shell plugin [07:16] smspillaz: GS also uses RGBA, does it not? [07:17] smspillaz, hi. [07:17] hyperair: the shell runs in process [07:17] smspillaz: yes? [07:17] smspillaz - great you have set up a 0.9.10 PPA - but it is empty :( [07:17] smspillaz: as does unity. [07:17] smspillaz: so why is the dash so goddamned slow? [07:17] i type super+a and it takes forever to search for whatever i type in [07:17] hyperair, it is already fast here [07:18] MCR: you running on an SSD? [07:18] yep [07:18] see, therein lies the problem. [07:18] but it got faster and faster each version [07:18] hyperair: dash blurs probably [07:18] everyone uses a damn SSD and says "oh my god, it's so blazing fast" [07:18] smspillaz: disabled. [07:18] the blur is heavy on the hardware [07:18] hyperair: you are talking about the dash search [07:18] not the UI [07:18] didrocks: yes. [07:18] and this is not in the process [07:18] and goes through dbus btw :p [07:18] didrocks: the blurs exacerbated the issue by having a horribly slow UI response [07:19] so what you are asking, putting the shell not in the wm is something totally separate [07:19] hyperair: *shrug* Trevhino and I were looking into something the other day as to why the dash was slowing down. We think its a set of draw calls somewhere that's quite heavy, but its going to require a bit more time to consider [07:19] hyperair: we've figured out that the slow dash isn't usually CPU bound [07:19] smspillaz: oh yay, finally. [07:19] like how MCR and the rest of the SSD users are seeing blazing fast dash searches === jono is now known as Guest50668 [07:21] smspillaz, I fixed all issues in my MPs btw (except screenshot) and am waiting for approvals... [07:21] smspillaz, also I would like to apply as Compiz maintainer... [07:21] MCR: I think we can arrange that [07:22] great, number 1 + 2 ? ;) [07:22] MCR: yeah [07:22] I'll get on to both ASAP, pretty busy this weekend [07:22] Cool, thx [07:23] hyperair: if you know of any good OpenGL profiling tools let me know. I haven't reallly been able to find any other than gDEbugger and its a pain to get that to work [07:23] smspillaz: i think there was a mention of something in phoronix that valve used.. [07:23] hyperair: it has a high price tag [07:24] hyperair: if you go to their website it says "contact us about pricing details" [07:24] that's code for "this is several thousand per seat and we only negotiate enterprise contracts" [07:24] smspillaz, do you have an ETA for the new PPA ? - I am eager to test it (altough most of 0.9.10 runs here already, ofc) [07:25] MCR: PPA of what ? [07:25] compiz trunk? [07:25] https://launchpad.net/~smspillaz/+archive/compiz-dev yes [07:25] 5 minutes [07:25] https://code.launchpad.net/~smspillaz/+recipe/compiz-development-head [07:25] smspillaz: oh dear, that sucks. [07:28] smspillaz: i suppose you've looked through all of these: https://www.opengl.org/wiki/Debugging_Tools ? [07:29] I don't recall coming across that [07:29] smspillaz - great, the availability of this PPA should be announced also to get more testers, but I am sure you've planned that already ;) [07:29] dunno the best place to announce it really [07:30] Planet Compiz ? [07:33] the ayatana mailing list? [07:33] unity-design now, iirc [07:33] It could also be added to all the bugs that are fixed in 0.9.10, so people could immediately test those and report any issues [07:34] I could do that [07:54] MCR: zoom only works on things because those things are X windows that Compiz can stretch. That's not true for Unity elements :( [07:55] hi duflu :) [07:55] Hi [07:56] smspillaz, compiz in the PPA is still named 0.9.9 - not sure if that is correct, should be 0.9.10, no ? [07:56] On the other hand, you could just figure out the tranform matrix for the zoom (including translation) and pass it through to Unity/Nux. It could work [07:56] *transform [07:57] duflu, I know that this UI zoom was requested and would make much sense for visually impaired people, but it is not my main priority at the moment [07:57] first all window managing stuff needs to be rock solid imho [07:57] no more windows jumping around workspaces or maximizing on wrong workspaces for example [08:01] we are (finally) almost there... [08:04] tsdgeos, the plot thickens [08:04] tsdgeos, https://bugreports.qt-project.org/browse/QTBUG-30730 [08:05] so there's probably an internal object somewhere that is sometimes getting magically converted to a qobject and sometimes not [08:06] it seems [08:08] tsdgeos, what's more, the type of the singleton gets registered automagically, and in the global scope [08:08] tsdgeos, that's why I had to rename the singleton [08:08] hey sil2100! do you feel better? [08:10] Saviq: brrr [08:10] Saviq: i have http://paste.kde.org/~tsdgeos/728060/ working now :-) [08:10] tsdgeos, yay [08:10] just need to get it merged :D [08:10] it's two lines of code [08:10] but ... [08:10] didrocks: hi! I overslept by accident ;p [08:11] sil2100: seems you needed it then! :-) [08:12] sil2100, hey, happy friday ! do you feel better than the other day ? [08:12] didrocks: I got an e-mail about some apparent lockscreen errors from veebers in the 100scopes branch, might be that the fix I made is not enough - will look into that now [08:13] mzanetti, have you seen this before? https://jenkins.qa.ubuntu.com/job/unity-phablet-ci/637/testReport/junit/%28root%29/qmltestrunner/FilterGrids__test_clicked_signal/ [08:13] seb128: hi! Happy Friday indeed - I feel mostly much better now [08:13] sil2100: hum, I wonder if we shouldn't finish the hud stuff first [08:13] sil2100: so, what would I tell, regarding your email is to disable the failing test for armfh right now [08:13] didrocks: that as well - you got me e-mail? [08:13] yeah ;) [08:13] Oh, ok [08:13] sil2100: and powerpc is FTBFS, so maybe the same [08:13] sil2100: just write it in the spreadsheet as well :) [08:14] sil2100: then, I think we should try to rerun stack per stack [08:14] in head [08:14] and see if autopilot behaves good :) [08:14] dandrader: yes. let me check [08:14] making sense? [08:14] Ok, will do that - I just didn't want to disable the failing test because it was just failing because of the loading error [08:14] But let's do that not to waste time [08:15] dandrader: or better: let me Deebug it :D [08:16] morning nic [08:16] sil2100: yeah, we have tracks of what's wrong, let's try to get all the stacks running first :) [08:20] dandrader: really weird. can't reproduce this locally [08:34] mzanetti, there's a bunch of qmluitests that failed yesterday with that for every ui test [08:34] mzanetti, looks like GL accel broke on the VMs or something [08:35] Saviq: that's already fixed [08:35] Saviq: the current red ones are merge conflicts [08:35] mzanetti, I meant the shader issue dandrader mentioned [08:35] Saviq: and I'm currently Deebugging the yellow ones [08:36] mzanetti, great, thanks [08:36] I'm not touching anything then [08:37] Saviq: I'm not really sure how my commit passed CI tho and then broke all following ci runs... [08:38] mzanetti, I'm not entirely sure it's related [08:38] mzanetti, as it's just your filtergrid that fails [08:38] mzanetti, but the rest report the shader errors just as well [08:38] mzanetti, but what about the missing Dee module and "tst_FilterGrids.qml:91: TypeError: Cannot set property 'model' of null" [08:38] dandrader: I hope it'll be fixed by this https://code.launchpad.net/~mzanetti/unity/phablet-deebug/+merge/159773 [08:38] Saviq: https://codereview.qt-project.org/#change,54244 [08:38] Saviq: but that filtergrids test must have been passing at some point [08:39] Saviq: meanwhile yesterday i also fixed a bug dpm reported to me (no clue why) that is stalled because i haven't fixed it for windows and mac :-/ https://codereview.qt-project.org/#change,54214 [08:40] tsdgeos: I think for QtQuick its RGBA, not ARGB [08:40] tsdgeos, nice one, thanks [08:40] mzanetti, it's ARGB [08:41] didrocks: https://code.launchpad.net/~sil2100/hud/disable_source_test_and_tweak_timeouts/+merge/159776 [08:41] mzanetti: nope, it's not [08:41] ok [08:41] Saviq, I didn't get the InputFilterArea issue. So the removal of the fakes from the shell code made that thing stop working (when dealing with real apps)? [08:41] sil2100: waiting for the diff :) [08:42] mzanetti: http://paste.kde.org/~tsdgeos/728096/ that's the code they use [08:42] sil2100: what about powerpc FTBFS? [08:42] dandrader, yes, because it relied on the "can I import Ubuntu.Application" bool [08:42] tsdgeos: its Qt.rgba()... thats why I thought so [08:42] sil2100: it's the timeout being bigger? [08:42] dandrader, which you removed (and rightfully so, we just need to get rid of that wrapper) [08:42] mzanetti: yep [08:42] didrocks: it was a ekhm, timing issue... there's that one problem we have when tearing down the HUD connection [08:42] ok [08:42] Saviq, ah, got it [08:43] dandrader, and implement a noop filter in the fake plugin [08:43] dandrader, and again, DON'T YOU SLEEP!? [08:43] Saviq, yes [08:43] It's really irritating, since there's not much we can do than wait, we might want to redesign somehow in the future maybe... [08:43] Saviq, climbing this aftenoon :) [08:44] sil2100: hum, is hudtest(test-source test-source.c test-source.xml) only with platform api? [08:44] Saviq, I would rather say the opposite. If I log on at 5am. you're online. At 5pm, you're online as well ;) [08:44] dandrader, details [08:46] sil2100: commented [08:49] didrocks: good catch, no need to disable it on non-platform-api systems [08:49] sil2100: ping me back :) [08:52] didrocks: pushed and build-tested... but the first push went to some different branch, grrr [08:53] * sil2100 hates when the default push branch is some old branch [08:53] tsdgeos, and you guys complain about whitespace checks, "possible spelling error"? lol! [08:53] yeah [08:54] sil2100: approved :) [08:54] sil2100: ok, so now some stacks in head are running [08:55] sil2100: so, if you look, some have issues to fix, even before autopilot [08:55] sil2100: see he yellow indicator stack? it seems something is out of sync [08:56] sil2100: there is also qtvideo-node in the head/media stack which needs fixing [08:56] (don't look at the mediaplayer-app, I fixed it) [08:56] Looking looking [08:56] sil2100: phone and apps stacks are running autopilot, we'll see the result soon :) [08:57] Oh, since I didn't see those blinking, so thought it's over already ;p But I'll check the indicator stack [08:57] is anyone else's PA hogging the CPU constantly since yesterday? [08:57] sil2100: yeah, it's a "view", not a realtime status [08:57] sil2100: it's showing "latest finished state" :/ [08:57] the worst thing about javascript is that it fails silently [08:58] dandrader, yeah, undefined == false is the worst [08:59] mzanetti, http://s-jenkins:8080/job/unity-phablet-qmluitests/448/console seems to hang [08:59] Saviq: at least that spelling stuff is immediate, doesn't take 30 min like our whitespace check [08:59] tsdgeos, if you had tested locally ;) [08:59] yes, yes, i know, execute them locally :D [09:00] Saviq: *grrr* [09:00] Saviq: well, i saw your whitespace error too ;-) [09:00] Saviq: I just requested 2 additional VM's [09:00] tsdgeos, indeed [09:00] Saviq: before they go online I'll use one of them to debug this more [09:00] tsdgeos, but I'm not complaining, am I :P [09:00] mzanetti, k [09:06] Saviq: not only that, now i have to wrap the commit message at 72 cols [09:06] ... [09:06] sil2100: ok, it's merged, I'm relaunching the hud stack with only rebuilding hud FYI [09:07] haha [09:07] let's pretend we live in 1975 [09:08] Saviq: and they're not going to accept it for 5.1 so that sucks [09:09] didrocks: ok [09:09] tsdgeos, I gathered as much [09:09] tsdgeos, but it's fine, we have good processes for distropatching, I think [09:10] oka [09:10] didrocks: btw. the indicator stack... it seems all the jenkins stuff is month-old [09:11] tsdgeos, hmm, shouldn't we also support #ARGB? [09:11] sil2100: did you look at the branches in the config? [09:11] sil2100: and did we got new commits? [09:11] Saviq: don't see the need to be honest [09:11] does anyone ever use that format? [09:12] tsdgeos, #RGB is supported, so [09:12] i know [09:12] * sil2100 is still noobish here [09:12] my guess is that this is just because they need to support xpm [09:12] didrocks: yes, I checked and there were commits to the indicator.cfg [09:12] didrocks: but the last run of for instance cu2d-indicators-head-1.1prepare-indicator-datetime is 22 days old [09:12] tsdgeos, anyway, QML doesn't support #ARGB [09:13] tsdgeos, so that's that [09:13] exactly [09:13] didrocks: and that's why indicators is yellow, since it was executed when indicator-datetime was still using the raring 13.04 branch [09:14] sil2100: indeed, there is no time assigned [09:14] sil2100: I think nobody redeployed the stack [09:14] cyphermox: seems that you never redeployed the head/indicators.cfg stack. Doing it now ^ [09:14] \o/ [09:15] * sil2100 wanted to use the 'redeploy' word but wasn't sure if it's the right one [09:15] nice catch sil2100 ;) [09:15] heh :) [09:15] sil2100: basically I looked at the head configuration [09:15] sil2100: and didn't see the timer [09:16] tsdgeos, typo [09:16] tsdgeos, not ARBG but ARGB [09:16] I think it should indeed show the 'last run' timer in the main 'Head' view [09:16] Saviq: where? [09:16] tsdgeos, commented inline, qcolor.cpp [09:16] _p [09:16] sil2100: yeah, that's the issue to play with jenkins to control the whole chain… [09:17] ok [09:17] actually no, qcolor.cpp [09:17] dang [09:17] tx [09:19] didrocks: you fixed up the media stack problems already, yes? [09:21] sil2100: not the qtvideo-node, as said [09:21] What also confuses me in jenkins is that there is a Last Success column and Last Failure column, but sometimes I would simply need a 'Last Run' one - here I have to check both columns to get the definite answer on when the build was executed [09:21] sil2100: mind looking at it? [09:21] didrocks: aye! [09:21] sil2100: agreed it's confusing [09:35] mzanetti, did you see the jobs seem not published anymore? following the links in jenkins messages gives out 404 [09:36] Saviq: no... didn't see that yet... still trying to fix the broken test [09:36] mzanetti, k [09:36] mzanetti, do you know who do I ping in absence of mmrazik about that? [09:36] Saviq: I guess me [09:36] lol [09:37] Saviq: in the afternoon fginther or sergiusens too [09:37] mzanetti, k [09:37] Saviq: I'll check it out as soon as the test is fixed [09:37] mzanetti, thanks [09:42] launchpad is super slow here, only to me? [09:43] Cimi: works fine here [09:43] strange [09:44] btw [09:44] Cimi: maybe you're super fast today :) [09:44] oh definitely not :) [09:44] I mean via browser, pages take a minute [09:44] I'm in Italy btw for the weekend, arrived here late night [09:45] maybe we don't like launchpad xD [09:45] dednick, do you have any update on the categories branch you didn't push? [09:46] Cimi: nearly there [09:46] Cimi: for me internet is in general very slow in italy... Might be related to the mountain area I come from [09:46] mzanetti, all the rest of internet is fine [09:46] yeah, internets don't deal with mountain well ;) [09:46] just launchpad got slo [09:47] they get tired [09:48] Saviq, have you mentioned pulseaudio eating CPU all the time? [09:48] dandrader, yes [09:48] I'm getting the same now [09:48] dandrader, Saviq: what version of the package do you have? [09:48] ok chrome is slow, safari is fast [09:49] dandrader, Saviq: https://launchpad.net/ubuntu/+source/pulseaudio/1:3.0-0ubuntu6 is supposed to fix it [09:49] hmmm [09:49] seb128, 1:3.0-0ubuntu5 [09:49] DNS prefetch? [09:49] Saviq, upgrade [09:49] seb128, trying [09:49] seb128, 1:3.0-0ubuntu [09:49] 1:3.0-0ubuntu5 [09:49] dandrader, same, upgrade [09:49] maybe I'm running uk dns [09:50] didrocks: the problem with qtvideo-node is that the source package name is different from the branch name... [09:50] didrocks: since qtvideonode-plugin is the src package, while the branch is lp:qtvideo-node [09:51] At least hm [09:51] seb128, hmm no upgrade available yet [09:51] Saviq, https://code.launchpad.net/~dandrader/unity/phablet_remove_fakes_from_qml/+merge/158370 should be good to go know, although jenkins is having some difficulties [09:51] s/know/now [09:51] dandrader, yeah, mzanetti is on it [09:51] Saviq, change mirror? [09:51] Saviq, or use directly archive.ubuntu.com [09:51] seb128, archive.ubuntu.com [09:52] seb128, yeah, I'm using archive directly [09:52] didrocks: maybe we should change lp:qtvideo-node to lp:qtvideonode-plugin ? [09:52] Saviq, hum, 0ubuntu6 was uploaded yesterday afternoon, it should be available [09:53] Saviq, sudo apt-get update? [09:53] Saviq, odd I got the pulseaudio upgrade [09:53] * Saviq disables apt-cacher [09:53] sil2100: what's the project name? [09:54] * dandrader reboots to try it out the new stuff (including new kernel) [09:54] better [09:54] https://launchpad.net/qtvideo-node <- on LP that's like this [09:54] sil2100: I would rather go the other around, change the source package name to map to the project name [09:54] sil2100: as it's what was decided for opening the project [09:54] sil2100: also, can you please add some bootstrap commit message? [09:56] well, so far so good. seb128 thanks for the tip! [09:56] dandrader, great, thanks for confirming the fix ;-) [10:01] seb128, yup, helped, thanks! [10:01] Saviq, yw! [10:01] * Saviq wonders why apt-cacher-ng broke [10:02] didrocks: in case of changing the source package name... should I change all previous changelog entries, or just the most recent? ;) [10:02] Will do! [10:02] sil2100: most recent is fine === tvoss|exercise is now known as tvoss [10:10] Saviq: so, what exactly is not published? [10:11] mzanetti, actually, it seems to have just taken time [10:11] mzanetti, yeah, it's fine [10:19] MCR: which ones needed review ? [10:19] MCR: I don't see anything that's not marked resubmit, and the only other one has comments by me but no follow up [10:20] smspillaz, https://code.launchpad.net/~compiz-team/compiz/0.9.10/+activereviews [10:21] Sam, all that have been resubmitted (Resubmit: 1) [10:22] MCR: hmm, I thought when you marked something "resubmit" it meant you were going to resubmit it [10:22] in any case I can have another look [10:22] and if you have a solution to fix the screenshot thingy with fbos, you could take over the screenshot branch [10:22] MCR: we'll do it separately [10:22] ok [10:23] I have set the screenshot MP to WIP [10:24] MCR: hm? I meant we'll do the fbo thing in a separate mp [10:24] yeah, I understood [10:24] but I have not yet applied your suggestions [10:24] ok [10:25] in my MP implementing the overlay color selector for screenshot [10:25] MCR: oh, actually quesiton about the cube gears proposal [10:25] sure - go ahead [10:25] do we call glDepthMask (GL_FALSE); anywhere ? [10:25] (can't see it from the diff [10:26] I do not think so [10:26] MCR: oh, actually [10:26] MCR: never mind [10:26] MCR: I meant to ask - we're calling glDisable (GL_DEPTH_TEST); [10:26]