[00:00] yes? [00:00] lets try just a simple script, open up gedit [00:00] type: #!/bin/sh [00:00] for the first line [00:00] then go to a new line [00:00] unity --advanced-debug 2>&1 | tee unity-gdb.log [00:01] then save the file as unity-gdb.sh [00:01] and type chmod +x unity-gdb.sh [00:02] Luzbelito, if you're confused at all just let me know :) [00:02] * bschaefer goes to fast at points... [00:03] in which folder i save it? [00:04] Luzbelito, your home folder if fine [00:05] ok [00:06] chmod +x unity-gdb.sh this in the tty? [00:06] sorry, once you save the file as unity-gdb.sh exit gedit, then on the command line type chmod +x unity-gdb.sh [00:06] Luzbelito, nope anywhere, it changes the file to an executable [00:07] that way you can just do this from a tty: [00:07] ./unity-gdb.sh [00:13] and what are we expectign from this script? [00:14] to generates the doc with the info? [00:15] Luzbelito, to run unity in gdb from tty, which will generate a stack trace for me to look at to see where unity is crashing [00:15] Luzbelito, let me know when you are about to run the script [00:15] i'll run it from tty in 5, 4, 3... [00:16] from there we can either find a bug that related which might have fix on it [00:16] Luzbelito, wait [00:16] so once you run it [00:17] you'll wait for it to load up, then when you get a chance type "r" to run unity [00:17] from there go back to (Ctrl+alt+F7), [00:17] wait for things to crash [00:17] onces they crash go back to tty (ctrl+alt+f1) and type bt (backtrace) [00:17] which all this will be outputed into a log that is set up...once you've done that [00:18] you can use pastebinit unity-gdb.log which will give you a pastebin link that you'll link here :) === jhodapp|bbl is now known as jhodapp [07:25] MacSlow: I didn't get your problem from the mail you wrote [07:25] MacSlow: can you elaborate? what exactly does not work? [07:26] Saviq: ping [07:26] dednick, otp [07:28] mzanetti, I'm trying to get along using simple string-arrays "as" now. Initially I wanted to map a GVariant of "a(ss)" and "a(sss)" to Qt data-types equivalents... but it's probably not possible. [07:29] MacSlow: should be [07:29] Saviq: unping [07:30] MacSlow: do you have an example of such a a(ss) value? [07:30] dednick, pto [07:30] :) [07:30] MacSlow: I would assume you get them in a QVariant where you can call toMap().value("somekey).toString() for example [07:31] mzanetti, e.g. [("textentry", "Login name:"), ("passwordentry", "Password:")] [07:31] MacSlow: yeah... I thin you could access stuff like this: [07:31] MacSlow: variant.toMap().value("textentry).toString() => "Login name:" [07:32] MacSlow: or variant.toMap().value("passwordentry").toString() => "Password:" [07:33] mzanetti, I rather not have it be a map [07:34] * mzanetti is wondering why [07:36] mzanetti, it makes passing it to the frontend more difficult. [07:38] MacSlow, how so? [07:38] MacSlow, you'll be able to just access it via obj['key'] [07:38] MacSlow, instead of iterating over it and trying to find the key [07:41] dednick, ah, but that's about transitioning the tab *bar*, not the tab contents [07:41] dednick, I meant that there's no transition between indicator pages [07:42] Saviq: ah. pages. As far as i've been made aware, this is as designed [07:43] dednick, ok then, happroving [07:45] Saviq: whoop. thanks [07:54] Saviq, mzanetti: .canConvert>() fails/returns false when trying to pass a "a(ss)"-GVariant [07:57] isn't a(ss) an array of stringlists? like [[foo, bar],[baz]] [07:58] no, ss is exactly two strings [07:58] mzanetti, as far as I understand GVariant it's an array of string-tuples [07:58] tsdgeos, yes [07:59] i mean, yes two strings are a string list [07:59] of length two [07:59] but [[foo, bar],[baz]] is not a(ss) [08:00] mzanetti, there must be 2 strings in each entry... it cannot vary [08:02] mzanetti, I won't waste any more time on this and stick with "as" and its drawbacks... at least those I can properly access on the backend-side [08:04] garg [08:04] now my phone doesn't get connceted to wifi [08:04] and now it did [08:04] ah wait no [08:04] the ifconfig was local :D [08:07] i guess i'll phablet flash :-/ [08:07] tsdgeos: apt-get not working? [08:07] nope [08:08] :( [08:14] jamesh: I'll begin the renaming of hollywood in a moment - any objections? [08:14] sil2100: nope. [08:14] :) [08:15] sil2100: I've been testing a branch to rename the grilo plugin too, so that should be the last bit [08:20] dednick, can you clarify something for me? [08:20] Saviq: i'll do my best [08:20] dednick, Unity.Indicators2 is supposed to replace Unity.Indicators in the long run, right [08:21] dednick, but Unity.Indicators.{Messaging,Network} isn't going away any time soon [08:21] if ever? [08:21] Saviq: yes, it was. But I've changed Unity.Indicators to Unity.IndicatorsLegacy and Unity.Indicators2 to Unity.Indicators [08:22] dednick, problem is Unity.IndicatorsLegacy keeps everything like IndicatorsModel, IndicatorsManager et al, that's still being used, right? [08:23] Saviq: I've moved all that into Indicators, but the Messaging/Network ones are dependent on Legacy for now. We dont have a new backend equiv yet. [08:24] dednick, ah, then that might be ok [08:24] manager/model etc i moved last night into Unity.Indicators (new backend) [08:25] So now when we're done with Legacy, we can just remove. [08:25] dednick, ok, let me just try and fix the diff so that it shows what we really want, then [08:26] mzanetti: in stuff like https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-saucy/1059/consoleFull is it that test_rightEdgeDrag() is staying there "forever" and thus gets killed by the 60 min time limit? [08:31] dednick, aaargh I hate bzr so much ;( [08:33] Saviq: indeed. keeps picking up things as remove->add. I think the only way to do it may be by flattening the branch at this point. [08:34] dednick, you ok with that? [08:34] dednick, I'll do the grunt work, I just want the diff to actually show what's happening [08:34] Saviq: meh. not really bothered. [08:35] i think i did it awhile ago anyway [08:37] Saviq: you'll need to move the Network/Messaging and other old Legacy code if you dont want it to generate new file diffs. Which is where i ended up with 6k diff last night [08:38] dednick, yeah, I hope I'll manage :) [08:38] Saviq: you want me to move it to ~unity-team ? [08:38] dednick, nah, that's fine [08:38] dednick, let's see if I can actually succeed within a sane amount of time [08:39] tsdgeos: hmm... hard to say. I would need to see such a job while it runs [08:40] now i had another one being killed in dash [08:40] it'd be cool if we had timestamps [08:40] like the kde jenkins server has [08:40] so one could know where we lost all those 60 minutes [08:41] i.e. in http://build.kde.org/view/All/job/calligra_stable/1071/console you have the timestamps [08:41] someone to review https://code.launchpad.net/~unity-team/unity8/unity8.background-gsettings-fix/+merge/179884 ? [08:43] did s-jenkins just die? [08:45] it's back [08:46] Saviq, mzanetti is there a bugreport for u8 looping? [08:47] tvoss_, there is one, but we were hoping it was fixed :/ [08:47] tvoss_, which u8 version [08:47] ? [08:47] Saviq, updated yesterday [08:47] dednick, ↑ [08:48] ? [08:48] tvoss_: https://bugs.launchpad.net/touch-preview-images/+bug/1183065 [08:48] Launchpad bug 1183065 in touch-preview-images "Occassional severe battery drain" [High,Confirmed] [08:48] whats looping? [08:48] whose looping where? [08:48] asac: using a large amount CPU [08:48] no :) [08:48] hehe [08:48] i know what looping means [08:48] :) [08:49] ah... unity8 [08:49] jamesh: we'll have to wait with the detailed LP project renaming for Francis (he's in the US timezone), since we need him to switch the CI/merger machinery to the new name [08:49] etc. [08:49] mzanetti: i was wondering which thread is looping in unity8 [08:49] is that the main UI thread? [08:49] i wouldnt think so because the UI is pretty usable still :) [08:50] asac: We think its the indicator that constantly tries to read from the network service despite that being crashed [08:50] asac: check out the bug report. there's a strace while it happens: https://bugs.launchpad.net/touch-preview-images/+bug/1183065 [08:50] Launchpad bug 1183065 in touch-preview-images "Occassional severe battery drain" [High,Confirmed] [08:50] ok [08:50] from our last encounter, its objects that werent being deleted (animations kept running) [08:52] so is it still draining? or not? [08:53] * greyback rebooting [08:54] come on [08:54] what's happening with the qmluitests, they are all timing out [08:54] is it just that we got slower machines now? [08:54] or we have too many tests? [08:55] https://jenkins.qa.ubuntu.com/job/unity-phablet-qmluitests-saucy/1071/consoleFull this one just timedout after a successfull "********* Finished testing of qmltestrunner *********" [08:55] this is getting a bit silly :-/ [08:56] mzanetti: where/how does the indicator do the polling? [08:56] sil2100: okay [08:56] asac: I don't know exactly where it happens. dednick debugged it. he might be able to give more details [08:56] mzanetti: is that implicit through using something like dbus framework, or does it poll directly thourhg select etc.? [08:56] guess the former [08:56] asac: yes [08:57] mzanetti: are the indicators in lp:unity8? [08:57] asac: yes [08:57] asac: dbus [08:57] tsdgeos, remember we're running multiples of qmltestrunner [08:57] Saviq: concurrently? [08:58] tsdgeos, no [08:58] dednick: did you figure which indicator? [08:58] dednick, MenuActionBinding.qml, does it have any corresponding thing or is it just gone from the new Unity.Indicators? [08:58] Saviq: then i don't understand what you mean [08:58] tsdgeos, ok, nvm me [08:59] bregma: hi! [08:59] asac: last time i saw the issue, it was because objects weren't being deleted due to a quicrk in qt where calling back into qt directly from a glib callback wouldnt action an deleteLater on objects. In the case of the network indicator, animations were being deleteLatered, and because they were infinite animations and not being deleted, they would continue running forever. [09:00] dednick: yeah. i assume all that logic to stop polling is probably nicely hooked up to delete etc. [09:00] Saviq: removed. menu items signal delegate to do the updating of backend. [09:00] dednick, k [09:00] so memory leak will trigger it [09:01] it's possible i didnt catch all the instances of the issue (my fix was a workaround). i proposed a fix to qmenumodel for a more permanent solution. [09:02] asac: but it hasnt been approved [09:02] dednick: rejectred or waiting for reviewers? [09:02] asac: waiting [09:02] unfortunately larsu is on holiday now [09:02] dednick: do you have a test package with that included? i certainly see the looping and would be happy to test [09:02] dednick: noone else can review? [09:02] bregma: we seem to getting a failure in one of our important merges to lp:nux - it seems to have problems finding the check-headless target, even though this branch didn't do any change for that [09:02] MacSlow: hi! [09:03] bregma: https://code.launchpad.net/~sil2100/nux/unity_support_test/+merge/179666 [09:03] pete-woods, what's up? [09:03] bregma: can you get someone from nux to look at it? [09:03] MacSlow: I was checking up on the "system dialogue" stuff [09:04] pete-woods, nowhere near completion [09:04] asac: https://code.launchpad.net/~nick-dedekind/qmenumodel/QTBUG-32859/+merge/179344 [09:04] unfortunately it doesnt look like qmenumodel ci does the packages. [09:04] pete-woods, I would have pinged you otherwise [09:04] MacSlow: is there any sort of API for interacting with them I can start to look at? [09:04] seb128_, Saviq mumble? [09:04] asac: but i can build one if you wish [09:04] pete-woods, no... [09:04] Cimi, nothing in my calendar ;) === seb128_ is now known as seb128 [09:05] neither in mine [09:05] MacSlow: okay, thanks, just wanted to check :) [09:05] dednick: package would be nice [09:05] :) [09:05] pete-woods, currently most of it is done (intended to be done) via notification-hints [09:05] Cimi: i'll have a look at that review [09:05] dednick: just put it somewhere on people.canonical.com or so [09:05] pete-woods, and even that isn't set in stone [09:06] Saviq, asac: on a related note, i have seen instances of the other network related components (wpa_supplicant?) going to full cpu and not going down. [09:06] seb128, Saviq check again :P [09:06] MacSlow: okay, are there any good examples of using the notification API? I haven't even used that yet, so at least I could kinda get into the mindset then [09:06] dednick: maybe... right now i see clearly the unity8 process looping :) [09:06] i havwent seen wpasupp for a while [09:06] asac: ah. ok, wasnt sure if you had actually checked the process. [09:07] sure i did :) [09:07] pete-woods, bzr branch lp:unity-notifications and look into the examples-directory [09:07] MacSlow: thanks! [09:07] pete-woods, there are C++- and Python-examples [09:07] asac: does showing top threads give any more info on where it's stuck? [09:08] dednick: what should i run? [09:09] exactly? [09:09] Cimi, Saviq: is that going to be hangout or mumble? [09:09] pete-woods, btw... the sd-example-incoming-call.py can take multiple actions (more than 2) [09:09] seb128, Cimi, your call [09:09] MacSlow: I only see Python examples, but maybe I can look at the unit tests to see C++ ones [09:09] asac: I just ment running 'top', and "shift+H" to show threads. [09:09] dednick: ah [09:09] seb128, you wanna know if you need to put your pants on? ;) [09:09] yeah found it through man [09:09] dednick: QQuickRenderThr [09:09] thats the thing looping [09:09] or consuming constant 50% [09:10] asac: yep, just keeps redrawing. [09:10] pete-woods, just uncomment the other n.add_action(...) calls [09:10] dednick: right. what triggers that? i assssume you havent tackled that? [09:10] Saviq, I rather want to know if I need to figure out if mumble is still working, I didn't use it in ages and last time I did the push-to-talk wasn't working for me [09:10] i certainly dont touch the UI :) ... nor do i have the screen on, so this surely shouldnt consume 50% :) [09:10] seb128, I'm good with hangout [09:10] pete-woods, doh... yeah... only Python-examples there [09:10] cool [09:10] asac: last time it was animations that were leaking. [09:11] Saviq, and you can do audio only if you don't want to put your pants on :p [09:11] ;) [09:11] MacSlow: I'm really pleased to see what looks like a nice fairly simple Qt API, though :) [09:11] pete-woods, I didn't have the time to port all the C-examples from lp:notify-osd yet [09:12] pete-woods, it's libnotify really... nothing Qt-specific [09:12] dednick: animations stopping their timeout when they get freed, righ? [09:13] dednick, https://code.launchpad.net/~unity-team/unity8/indicators.unitymenumodel/+merge/180070 [09:13] dednick: so your package might fix it? [09:13] dednick, the diff is bigger, but having in mind that you can ignore IndicatorsLegacy, it's not, really [09:13] let me know if you have some (not urgent) happy to be the experiement [09:13] dednick: te [09:13] btw, did the software store merge land already? [09:14] asac, you mean app preview? [09:14] asac, then yes, it was just released [09:14] asac: the animations are never stopping because they are infinite and arent deleted. [09:14] Saviq: everything ... scope i think is the main part [09:14] asac, scope was there for some time already [09:15] Saviq: but i dont have it on my phone [09:15] asac, guys are fixing some issues still on the backend side [09:15] asac: https://bugreports.qt-project.org/browse/QTBUG-32859 [09:15] so its not in the daily-release branch? [09:15] or is it disabled? [09:15] asac, see description of https://code.launchpad.net/~unity-team/unity8/app-preview-data/+merge/176941 [09:15] asac, or https://code.launchpad.net/~unity-team/unity8/app-preview-data/+merge/179348 rather [09:16] Saviq, seb128 mumble or hangout? [09:16] Cimi, h'out [09:16] Cimi, hangout, I'm coming [09:16] Saviq: hey, do you know if there is a method in qtdbus to know that a certain interface is available? (like a daemon waking up?) [09:17] hmmmm [09:17] pete-woods, you know that ↑ [09:17] Saviq: ok so it just landed so might be in todays image [09:17] didn't mterry fix the problem with the ownership of MockLauncherModel? [09:17] if we can get that pass our tests [09:17] it's still segfaulting here [09:17] Saviq: cool. thanks [09:17] asac, yeah, will be there soon (scope disabled for now, though) [09:18] Saviq: I thought you used http://qt-project.org/doc/qt-5.0/qtdbus/qdbusservicewatcher.html for that? [09:18] didrocks, ↑ [09:18] Saviq: when will it be enabled? [09:18] Saviq, we're waiting the master :P [09:18] what is holding it back? [09:18] Cimi, ah you added to the invite, going [09:19] asac, question for ralsina [09:19] pete-woods: Saviq: exactly what I needed, thanks! [09:20] asac: https://dl.dropboxusercontent.com/u/85539674/qmenumodel-qml.zip [09:21] dednick: cool it installed cleanly... rebooting :) [09:22] asac: for info, when last did you update unity8 ? [09:23] asac: or flash... [09:23] sigh. i dont think this is binary compatible... [09:23] hehe [09:23] shit happens [09:23] bugger [09:24] guess i wont have to wait for my UI ... problem solved. no looping unity8 :) [09:24] lol [09:24] asac: give me a few minutes and i can compile you a unity8 to go with it if you like ;) [09:25] dednick: btw, the libs dont need to be libs if noone beyond unity8 would use them... [09:25] in that way you dont need to bother about API/ABI [09:36] galaxy nexus is sooo slow! [09:36] give me an edge! [09:49] dednick, if I drag the (real) sound indicator's volume down to 0, it starts wobbling somewhat - might be the 0 vs 0.0 issue? [09:50] Saviq: do you remember what was failing when an invalid background was set? Was it the autopilot test? [09:50] tsdgeos, all the ap tests, yes [09:50] tsdgeos, as greeter has status == Image.Ready test to consider itself loaded [09:51] tsdgeos, and ap tests were waiting for greeter loaded to unlock [09:51] ok [09:51] let's try that with the new MR [09:52] dednick: GN is faster once we dont loop anymore :) [09:52] lol [09:52] dednick: guess that means unity8 is still building? [09:52] asac: 70% :) [09:53] Saviq: er, might be. possibly the backend deciding it doesnt like the value and bouncing back a different one. [09:54] dednick, also, between the two sound indicators, the "real" one's icon is smooth [09:54] dednick, the phone one is non-antialiased [09:55] Saviq: panel icon? [09:55] dednick, yes [09:56] dednick, MenuContent.qml:64:19: QML Loader: Binding loop detected for property "contentActive" [09:56] Saviq: give me a few minutes. cant use phone at the moment. odd though. they should be the same unless there's something up with the theme [09:57] dednick, that's on the desktop, not on phone [09:57] dednick, on phone I only get one (should I install the "real" ones btw?) [09:58] one what? [09:59] dednick, one sound indicator icon [09:59] Saviq: yeah, you'd need to install indicator-sound, indicator-power, etc. [09:59] indicator-time [09:59] *datetime [09:59] dednick, there's no bluetooth icon, too [10:00] Saviq: indeed. i'm in process of adding [10:00] dednick, ok [10:00] Saviq: although it doesnt seem to be working properly on the phone anyway [10:08] dednick, wow, it actually launched the settings app :D [10:10] Saviq: really? never even tried that. lol [10:10] dednick, it won't show up as it's not launched with the desktop file hint, but it did start ;) [10:11] E_TOOMANYINDICATORS [10:12] lol [10:12] yeah [10:12] asac: https://dl.dropboxusercontent.com/u/85539674/unity8.zip [10:12] cross fingers [10:12] (worked for me) [10:13] mine dont fit on screen... [10:19] Saviq: the icon antialiasing thing is a image://[gicon/theme] thing. I'm guessing we're loosing resolution in the GIconProvider pixmap loading process. [10:19] dednick, mhm [10:19] it's using the available sizes of the image, but it's an svg... [10:20] Saviq: result = themedIcon.pixmap(themedIcon.availableSizes().last()).toImage(); [10:20] dednick, ok, that should "fix" itself later, then [10:21] with larsu's new provider [10:21] right, we need to land that... [10:21] Saviq: ya [10:22] Saviq: anyway, it's the old indicator that's wrong isnt it? [10:23] dednick, well, the new one has the icons in the page (the slider between low volume and high volume) [10:23] Saviq: yeah, the new one has the proper panel icon i mean. [10:24] dednick, panel icon... it seems blurred, shouldn't be, AFAIK [10:24] dont know wtf is going on with the datetime indicator time format. sometimes 11:25, sometime 11:25AM , others 11. [10:25] dednick, https://launchpadlibrarian.net/147541416/shot.png [10:26] dednick, the formatting depends on locale/settings, so indicator-datetime should handle that internally [10:28] Saviq: hm. it is blurred. turned up to 30px esspecially. also icons in panel [10:28] hum. that's using image://theme . [10:29] dednick, anyway, the only potentially remaining thing is the "drag volume to 0 → E_CRAZY", did you remove the related code? [10:30] Saviq: does that only happen on phone? [10:30] dednick, no, only on desktop [10:30] Saviq: hm. not happening for me. [10:31] Saviq: i didnt remove the 0.0001 blah code. === alan_g is now known as alan_g|tea [10:32] Saviq: ah. yeah.... the icon goes funky [10:32] Saviq: let me check that out [10:33] dednick, yeah, if I then drag it back up, it will get death-metally, and not funky, even [10:33] the slider will do pogo-dance for a second or two [10:34] cd .. [10:34] bugger [10:35] i'm supprised i havent typed my su password in irc yet === alan_g|tea is now known as alan_g [10:46] Saviq: qvariant convertint double to int if on int boundary. going to have to make some changes to unitymenumodel and unity8 to fix. [10:46] Saviq: no idea why! [10:47] dednick, ok, approving in the mean time? [10:47] Saviq: yes please [10:47] it's not a critical thing [10:54] Cimi: the autopilot tests have started failing [10:54] seems to be exactly the same problem we used to have? === MacSlow is now known as MacSlow|lunch [11:05] Saviq: what is the ubuntu-touch package used for? [11:08] dednick, it's the metapackage/seed [11:08] dednick, it depends on everything the ubuntu-touch image is supposed to have [11:32] tsdgeos, Cimi, re: ap tests with background - looking at the video it does not fall back to the default background still [11:32] it is weird [11:32] or *again* [11:32] mzanetti, ^ ? [11:33] becuse the test actually tries that [11:33] Saviq, I didn't test them [11:33] and seems to work [11:33] the qml test i mean [11:34] let me run the qml test again [11:35] it works :-& [11:35] which means the test is not so good probably :D [11:38] which means the test is not so good probably :D [11:38] wops, worng screen for ↑+enter [11:39] seb128, shall I propose the inclusion when it's empty or when it's complete? [11:39] seb128, https://code.launchpad.net/~cimi/ubuntu-system-settings/welcome-wizard [11:40] Cimi, you can proposed it like that, I still want to discuss it with mardy/ken but we can do that on the merge request [11:50] mzanetti: greyback: Saviq: https://code.launchpad.net/~aacid/unity8/test_shell_crash/+merge/180108 this should fix the problems we have with the testShell test haging i think [11:50] lunch! [11:50] tsdgeos, thanks! [11:55] thanks tsdgeos === alan_g is now known as alan_g|lunch === MacSlow|lunch is now known as MacSlow [12:11] F*CKING Qt docs! [12:18] :D [12:33] mzanetti, tsdgeos, opinion: roleNames(), should return pre-populated m_roles or dynamically created QHash? [12:33] hmm... [12:33] it's a tricky question, too! :) [12:34] one saves us CPU, the other RAM [12:34] I think they are queried only when a view is attached [12:34] so only once [12:34] => saving RAM [12:34] which might differ for indicators or the like where views are constantly attached/detached [12:34] yup, my gut feeling, too [12:48]