[01:27] Laney, yeah i have since realized that my nexus 7 build box had the old gtk too === alf is now known as alf_ [04:59] Good morning [05:33] morning all [05:34] morning didrocks, are you already around? [05:34] dpm: hey, sure! [05:34] :) [05:35] didrocks, a quick question: I'm writing a tutorial on scopes, and it depends on this libunity revision to have landed in the distro: http://bazaar.launchpad.net/~unity-team/libunity/trunk/revision/240 - do you have a rough idea when it might land? Or how can I track when it lands? [05:36] dpm: should land today as it was merged yesterday evening [05:36] dpm: if tests pass while doing daily releases :) [05:36] dpm: the best way to track is to look at saucy-changes ML [05:36] didrocks, perfect, thanks :) [05:37] yw :) [07:20] Morning! [07:20] didrocks: hi! If I remember correctly, you were also getting that gpu hang and compiz hang-up on your lenovo, right? [07:20] didrocks: are you still experiencing that one? [07:20] sil2100: yeah, I have it sometimes [07:21] I get it 2-3 times every day, and frankly it's starting to get irritating ;/ [07:23] sil2100: argh, well, poke upstream I guess :) [07:23] sil2100: it seems that all integration tests are timing out, I think autopilot is pretty broken [07:23] sil2100: I'm still working on Mir and the new packages, do you mind having a look there? [07:23] sil2100: I would even suggest a revert, we asked for 4 days for a fix, and still nothing [07:23] thomi: veebers $ [07:23] thomi: veebers ^ [07:24] sil2100: all stacks seem impacted? (but first, maybe look if it's not a setup/environment issue) [07:24] didrocks: aye! hm, but yesterday we ran unity integration tests, right? [07:24] sil2100: didn't they timed out? [07:24] sil2100: seems the QA stack itself timed out [07:25] I'll let you handle that anyway, back to Mir here :) [07:25] didrocks: let me look at that then [07:25] sil2100: dpm is waiting for the new unity to update some tutorial, mind letting him know? [07:25] didrocks: good luck! If you need some help on Mir later, just give me a poke - let me see what's up with AP [07:26] sil2100: thanks, I've some headaches on the packaging, but at least, it's needed :) [07:26] didrocks: by 'new' you mean from today? Since yesterday we did a manual publish [07:26] sil2100: yeah, today's one [07:26] (the merge was yesterday afternoon) [07:30] thanks didrocks, sil2100 :) [07:31] didrocks, sil2100 I noticed that too: "[10760.361675] compiz[25473]: segfault at f2f2fffe ip b7083aee sp bfea9dc8 error 5 in libgobject-2.0.so.0.3702.0[b7052000+51000]" [07:42] good morning desktopers [07:46] hey seb128! how are you? [07:47] didrocks, hey, I'm good thanks, a bit tired though, got carried on in some guest session debugging at 23h30, wrong time to pick the computer back to "quickly look to that bug" [07:47] ;-) [07:47] seb128: ah, fatal error! I thought after all those years, you would have learn :) [07:47] hehe [07:48] I didn't fall into that trap for quite some time [07:48] it was like in the good old times :p [07:48] heh [07:52] dpm: hi! [07:52] dpm: what fix exactly do you need to get releases? [07:52] sil2100, hey [07:53] *released [07:53] seb128: hi [07:53] dpm: it's Ted's fix for upstart ? [07:54] hi sil2100, here's what I was asking this morning to give some more context: [...] I'm writing a tutorial on scopes, and it depends on this libunity revision to have landed in the distro: http://bazaar.launchpad.net/~unity-team/libunity/trunk/revision/240 - do you have a rough idea when it might land? Or how can I track when it lands? [07:55] sil2100, it's not Ted's fix, it's just a minor unity scopes API update [07:56] dpm: ok, then libunity it is! We're fixing the autopilot/otto issues we're currently experiencing, once these are dealt with we should be able to release, so I say approx 2-3 hours? [07:56] sil2100, that's awesome, thanks! [07:59] hey, are there any DBus experts around? I'm writing a tutorial for scopes, and to make it simpler and easier to follow, I'd like to depend as little as possible on a build system to do variable replacement at build time. Could someone tell me if for DBus this will work? -> http://pastebin.ubuntu.com/5797817/ [08:00] mooooooooooooooorning [08:00] Laney, gooooood mooooorning to you ;-) [08:01] heelloooo laneeey [08:01] dpm, I wouldn't rely on that to work but I didn't try [08:01] dpm, what's the issue with using the build system? [08:01] hey Laney, mlankhorst [08:01] dpm, don't install in the multiarch directory if you want to avoid that... [08:02] most build systems make .in replacement like that quite simple [08:02] but I guess you're doing a tutorial? In that case maybe just avoid multiarch [08:02] or use dh-exec? [08:02] Laney, but not qmake :) [08:03] seb128, the issue is that I'll have to write a relatively complex qmake .pro file to do the variable replacement. It will work, but it will confuse the hell out of people. They should be focusing on the code, not the build system [08:03] dpm, well, they should just copy your .pro then? [08:04] seb128, yeah, that's what they'll do anyway, but I'd rather they use a very simple .pro file, rather than a complex one they might not understand [08:04] dpm, don't use the multiarch paths then? [08:05] seb128, yeah, I think that's what I'll end up doing. At the same time, I want to encourage installation on the multiarch path is this is where the other scopes are installed [08:05] seb128: are you planning to upload gtk3? [08:05] dpm, you want everything for 0 effort ... [08:06] We figured out yesterday that that was why the gcc-s testsuite was failing [08:06] Laney, not yet, still trying to figure out that armhf bug, why? [08:06] ah, I'm not surprised [08:06] seb128, of course, were you expecting anything else? :P [08:06] unity-greeter failed the same way [08:06] I'll do it if you want [08:06] Laney, sure, thanks ... just using -O0 then? [08:06] I wonder what that will do to performance [08:07] but if it's going to be very temporary then it might be ok [08:07] it's not like anyone was using gtk on arm [08:07] gtk doesn't work on the touch image and we killed the panda on [08:07] one [08:07] not sure we have any saucy arm desktop image atm [08:07] hmm [08:07] anyway we can do it and then the alternative is to build with gcc-4.7 if it goes bad [08:08] well, -O0 works here on my local build [08:08] no reason it shouldn't work [08:09] I mean in the sense of "argh you made $my_thing suck" [08:09] Laney, "we ought never to have disabled the gtk tests on armhf..." [08:09] (reading backlog from yesterday) [08:09] true enough isn't it? [08:09] Laney, we would still have proposed builds broken on armhf today for a week [08:09] now we have random stuff broken in confusing ways [08:09] due to arch mismatch on gtk between armhf and i386 [08:10] we also fixed the most frequent i386/amd64 segfault in saucy with that update [08:10] I still think the tradeoff was good, as said it's not like anyone was using gtk on armhf [08:10] We'd probably have arrived at this same place [08:11] well, the goal was never to let the tests disabled [08:11] I'm doing gtk builds for almost a week here to try to debug it [08:11] it was just to unblock others [08:11] and I still think it we would have more annoyed people if gtk rdepends were still blocked since a week [08:12] anyway, little point arguing there [08:12] let's just rebuild it with -O0 [08:12] that's slightly better, still not a toolchain fix though [08:12] but those toolchain issues are timesinks :/ [08:13] yeah [08:13] There's not many people who are efficient at debugging them [08:13] I'm definitively not [08:13] I'm just poking around at this point [08:16] Laney, great, I'm making progress on the way to reproduce I think ;-) [08:16] :-O [08:20] jibel: do you know why proposed-migration thinks pango1.0 is still running? [08:21] Laney, no idea, it ran successfully yesterday. I'll have a look [08:21] thank you! [08:22] Laney, btw could you update the panels blueprint to detail a bit more the workitem on what you did and what is remaining to do ... maybe put stuff blocked on toolkit work or the picker for july or august? [08:22] oh yes, will do [08:23] thanks [08:38] there we go [08:40] Laney, sorry, I edited mines and overwrote your changes...added them back ;-) [08:40] Laney, thanks for the update [08:40] heh [08:40] we need some locking there :P [08:40] ;-) [08:41] Laney, btw gsettings-qt is in saucy since yesterday [08:41] * seb128 needs to try that today [08:42] yeah saw that [08:54] sil2100, didrocks: I updated some of the WI on https://blueprints.launchpad.net/ubuntu/+spec/client-2013-06-touch-landing ... please check that the status is right [08:54] didrocks, I think the ppc and otto ones are DONE, but I was not sure so I let those [08:56] seb128: ACK [08:56] seb128: hum, indicator-client is not under dailies yet [08:56] seb128: let me revert some :) [08:58] seb128: ok done, thanks! [08:58] didrocks, thanks [08:58] didrocks, https://launchpad.net/ubuntu/+source/indicators-client seemed like it was, sorry [08:58] 0.31daily13.05.02-0ubuntu1 [08:58] the version tricked me ;-) [08:59] seb128: yeah, seems they did it once, then reverted because of some segfault IIRC [08:59] seb128: look at the configuration rather [08:59] ok [08:59] seb128: http://bazaar.launchpad.net/~cupstream2distro-maintainers/cupstream2distro-config/trunk/view/head:/stacks/head/indicators.cfg#L74 [09:00] didrocks, is there still much NEWing to do BTW? [09:00] didrocks, ok, " daily_release: False" but not comment on why [09:00] seb128: yeah, ~20 components at least [09:00] alright, keep that WI then ;-) [09:01] I'll ask robru about the progress on them :) [09:01] heh ;) [09:02] sil2100: resetting yours to DONE, the edit reverted it [09:03] seb128: you reverted mines :/ [09:03] didrocks, :-( [09:04] * seb128 hate blueprint handling of conflicting edits [09:04] :D [09:04] seb128: same feeling… [09:04] seb128: turn all them to DONE, not cyphermox's one as he still has some components to land [09:04] didrocks, I'm done editing, feel free to change them, sorry about the revert [09:05] ok fixed [09:05] thanks ;-) [09:12] Laney, pango ran successfully yesterday against glib2.0 2.37.2-1ubuntu1, I cannot find what went wrong with the results and why it is still marked running. [09:13] jibel: so you think it could be britney-side? [09:14] Laney, no because I do not find any result file with a PASS, it would be on jenkins side. [09:15] hrm === tvoss is now known as tvoss|quick_lunc === MacSlow_ is now known as MacSlow [11:04] tkamppeter_: I was hoping I didn't need to explicitly mention it but the touch bug is not magically fixed by updating to a newer xserver, although the newer xserver does make it easier to pull in the current touch fixes, which you also tested and still didn't fix the touch bug === MacSlow is now known as MacSlow|lunch [11:25] is there any documentation for gsettings-qt? === ara is now known as Guest5314 [11:49] Laney, http://bazaar.launchpad.net/~larsu/gsettings-qt/add-basic-testing/revision/5/tests/tst_GSettings.qml [11:49] yay, ta [11:49] yw [11:50] Laney: I'll add some proper examples this week :) [11:51] larsu, hey ;-) [11:51] seb128: hi! How are you? [11:51] larsu, I'm good, thanks! How are you? [11:51] seb128: good as well, thanks :) [11:52] seb128: the tests you linked to fail when they are run in dpkg-buildpackage (but work fine when running `make check` manually), because Qt cannot connect to X. Do you have any idea why this might happen? [11:53] larsu, do you run them under dbus-test-runner? [11:53] it shouldn't need to connect to X, but the qmltest thingy depends on QtGui and initilizes it :( [11:53] seb128: no [11:53] try that [11:54] override_dh_auto_test: [11:54] dbus-test-runner -t dh_auto_test [11:54] in debian/rules [11:55] trying now, thanks! [11:55] seb128: same problem :( [11:55] QXcbConnection: Could not connect to display [11:55] hum [11:56] what if you use xvfb-run? [11:56] seb128: \o/ [11:56] thanks! [11:56] yw [11:56] larsu, should that work? [11:56] GSettings { [11:56] id: settings [11:56] schema: "org.gnome.desktop.interface" [11:56] } [11:56] Label { [11:56] text: settings["iconName"] [11:57] } [11:57] ? [11:57] yes [11:57] settings.iconName as well though [11:57] tcase.qml:19: Unable to assign [undefined] to QString [11:57] 19 is my "text: ..." line [11:57] hum [11:57] seb128: there's no such key in that schema... [11:58] there's icon-theme [11:58] bah [11:58] yeah, thinko... [11:58] (aka iconTheme) [11:58] :) [11:58] do I need to add xvfb to the build deps? [11:58] same problem though :/ [11:59] larsu: yes [12:01] seb128: that's weird. I'm trying to reproduce [12:01] larsu, does it work for you? [12:01] larsu, what qt source needed a fix? I want to check I've the right version [12:02] ken uploaded to saucy but I want to make sure that built/got published === tvoss|quick_lunc is now known as tvoss [12:03] seb128: libqt5qml5 5.0.2-1ubuntu2 [12:03] seb128: I can reproduce it, let me investigate [12:03] larsu, thanks [12:04] ARGH!!! [12:04] same thinko that you had [12:04] I cannot reproduce, works for me :) [12:04] seb128: do you have the right package version? [12:05] larsu, no, that's my issue I guess [12:05] 5.0.2-2ubuntu1~saucy1~test1 [12:05] blaming ppas... [12:05] I'm downgrading [12:05] ya, this error is exactly what you'd get without mardy's fix [12:06] larsu, works \o/ [12:06] :) [12:06] will teach me to opt in for ppas :p [12:06] larsu, danke ;-) [12:06] hehe [12:06] seb128: yw! [12:07] larsu, oh, and great work, being able to use gsettings that way rocks ;-) [12:07] thanks :) [12:08] the only problem is writing keys back - you have to use a function for that because qml only does one-way bindings [12:08] CheckBox { checked: settings.something; onClicked: { settings.something = value; } } [12:08] (onClicked is probably wrong syntax , but you get the idea) [12:08] that's easy enough [12:09] true :) === MacSlow|lunch is now known as MacSlow [12:25] good morning [12:26] hey desrt! [12:26] didrocks: hey :D [12:26] desrt, hey [12:26] Laney: hopefully you did not do glib yet [12:26] because i want to squeeze in one more patch today [12:27] desrt, he did... [12:27] ah crap [12:27] i knew i should have done it last night before bed :p [12:28] is that a blocker for the other bits to land? [12:28] no. not at all [12:28] k, so that's something at least ;-) [12:28] it'll be blocked for alpha 1 anyway [12:28] this one is a weird feature for some crazy guy working on qt stuff [12:28] * larsu heard that [12:29] oh. hi larsu [12:29] you use this channel too? [12:29] desrt: morning :) [12:29] well, yeah..... [12:30] Laney: blocked, how? [12:30] as in, no new uploads after this morning? [12:31] just will be held in saucy-proposed [12:31] larsu: good enough? [12:31] standard procedure around milestone releases [12:32] desrt: so I've hit another problem just now. I might need to have a list of things a custom widget supports after all [12:33] unless you have a good idea on how to solve this: I don't know when an attribute contains an icon [12:33] larsu: so maybe you don't need my patch? [12:33] ya, maybe [12:33] ah. tricky. [12:33] you need to hint the type conversion, you mean [12:33] exactly [12:33] right... [12:34] schema man schema man [12:34] does whatever a schema can [12:34] so I've been thinking to either have a .getAsIcon() in qml, which is ugly [12:34] or have a schema [12:34] right [12:34] somehow I guessed you might like the schema idea :P [12:34] it's the least ugly option, i think [12:34] but where should they live? [12:35] installed somewhere? [12:35] or passed in when constructing the model? [12:35] well... your usecase is that you want to be able to add new types of widgets without having to update some centralised list of attributes [12:36] _and_ your setup is structured such that you don't load widgets until after the model is constructed [12:36] so you have like... 3 options? [12:36] 1) have some .d directory somewhere with xml or something [12:36] 2) prescan/preload the modules [12:37] 3) have some mechanism for getting a 'raw data' object out of the model and wrapping it with something nicer from its user (who knows the required information by this point) [12:37] "with xml or something" [12:37] i love the smell of angle brackets in the morning [12:37] haha [12:38] (3) could be something like item.getCustomData(schema) [12:38] i kinda like 3, fwiw [12:38] where schema is a hash table [12:38] ya... at that point you have the ability to do anything at all [12:38] { 'primary-icon': 'icon', 'min-value': 'uint32' } [12:39] sounds good to me [12:39] Laney: no patch required :) [12:39] desrt: I'm kind of glad you didn't land it yet - you'd have been mad now :D [12:41] desrt, Laney: I'm getting closer of that gtk on arm breakage, testcase down to [12:41] cd gtk; touch gtkcssstylepropertyimpl.c; make; xvfb-run ./tests/builder === tkamppeter_ is now known as tkamppeter [12:42] seb128: nice :) [12:42] so this file is the one that new gcc is miscompiling? [12:42] CC libgtk_3_la-gtkcssstylepropertyimpl.lo [12:42] seems to be that object [12:42] yes [12:42] not sure what to do next [12:42] lemme look [12:43] nice! [12:43] ? [12:43] you picked a nice and small file [12:43] Laney, I found the problem with pango1.0 marked running [12:43] seb128: got a backtrace of the crash? [12:43] seb128, is doko breaking things again with his crack compiler? [12:44] Laney, libpango1.0-doc has a dep on lynx|www-browser but lynx doesn't directly provides the virtual package www-browser (it's lynx-cur) so the test can be triggered by any package providing www-browser [12:44] chrisccoulson, he is! [12:44] Laney, in the case of pango1.0 the test has been triggered by rekonq which also provides www-browser but when the test runs it installs lynx as preferred dep and the conditions of the test are never satisfied [12:44] mlankhorst, I simply tried it to see whether the patches are in and whether they help and have given my feedback. [12:44] desrt, wait, let me change the makefile back and run make [12:44] Laney, and we cannot match the request and the test results [12:46] Laney, a solution would be to enforce installation of the dependency that triggered the test or improve dependency resolution by going deeper in the reverse dep tree for virtual packages [12:46] I'll see how I can improve this [12:48] desrt, http://paste.ubuntu.com/5798418/ [12:50] i start to have a theory [12:50] desrt, let me know, but I'm lost on what would be the next step therer [12:50] out of looking at the generated asm [12:50] but there lie the path to crazyness :p [12:50] so follow this logic... [12:51] you are on the path you're on because of [12:51] if (!keyframes_set_value (keyframes, k, GTK_CSS_STYLE_PROPERTY (property), value)) [12:51] which happens because of: [12:51] _gtk_css_parser_error (parser, "Cannot animate property '%s'", _gtk_style_property_get_name (property)); [12:51] if (!_gtk_css_style_property_is_animated (property)) [12:51] return FALSE; [12:52] this is just a normal property getter [12:52] and observe: [12:52] node = g_object_new (GTK_TYPE_CSS_STYLE_PROPERTY, [12:52] "value-type", value_type, [12:52] "affects-size", (flags & GTK_STYLE_PROPERTY_NO_RESIZE) ? FALSE : TRUE, [12:52] "affects-font", (flags & GTK_STYLE_PROPERTY_AFFECTS_FONT) ? TRUE : FALSE, [12:52] "animated", (flags & GTK_STYLE_PROPERTY_ANIMATED) ? TRUE : FALSE, [12:52] "inherit", (flags & GTK_STYLE_PROPERTY_INHERIT) ? TRUE : FALSE, [12:52] !pastebin [12:52] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://imagebin.org/?page=add | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [12:52] "initial-value", initial_value, [12:52] "name", name, [12:52] NULL); [12:52] that's a long varargs function [12:52] i wonder if the "animated" is not getting passed in properly [12:53] seb128: wanna try something for me? [12:53] desrt, sure [12:53] go to gtkcssstyleproperty.c around like 278 [12:54] you should see the registration of the "animated" property [12:54] change the CONSTRUCT_ONLY to just CONSTRUCT [12:55] then just after the node=g_object_new() massive thing i pasted above, do this: [12:55] if (flags & GTK_STYLE_PROPERTY_ANIMATED) g_object_set (node, "animated", TRUE, NULL); [12:57] then add some of this before/after the set line you add: { gboolean x; g_object_get (node, "animated", &x, NULL); g_print ("Name: %s, animated: %d\n", name, x); } [12:58] desrt, ok, one sec, that panda is quite slow... [12:59] no n7 for you? [13:00] didrocks: one more quickie: https://code.launchpad.net/~sil2100/cupstream2distro-config/fix_packages_list_graphicaleffects/+merge/171297 [13:00] desrt, I've one, but io is slower than on this porter box [13:01] ah. tricky. [13:02] * larsu shouldn't have rebooted. Can't log in to a a sesion anymore :-( [13:02] Laney, you rock! I assume that gtk upload should fix the libsignon-glib tests :) [13:03] larsu, what's broken? [13:03] kenvandine, I was trying to determine if gtk was to blame, but the porter box gtk was outdated... [13:03] kenvandine, hey btw [13:04] seb128: I don't know yet. Typing my password shows a blank screen for a couple of seconds and throws me back into lightdm [13:04] larsu, what is in ~/.xsession-errors and ~/.cache/upstart/gnome-session.log ? [13:05] kenvandine: hi! [13:05] kenvandine: are you free for a quick review? :) [13:05] https://code.launchpad.net/~sil2100/cupstream2distro-config/fix_packages_list_graphicaleffects/+merge/171297 [13:05] seb128, yeah... my old broken n7 wasn't updated either :) [13:05] i thought it was... [13:06] sil2100, sure [13:06] kenvandine: btw. did I hear something about fixed libsignon-glib tests? :D [13:06] Laney fixed gtk [13:06] Laney: can I hug you? [13:06] hehe :) [13:07] whew... that packages list is long :) [13:07] seb128: nothing overly critical I would say... the lightdm logs tell me that org.gnome.SessionManager is not running, though [13:07] kenvandine: it's just a switch from libqt5graphicaleffects5 to libqt5qml-graphicaleffects [13:07] kenvandine: since libqt5graphicaleffects5 is no more ;) [13:07] larsu, does startx -- :1 from a vt works? [13:08] desrt, http://paste.ubuntu.com/5798481/ [13:08] seb128: yes, and interestingly enough, so does the guest session [13:09] seb128: neat! [13:09] larsu, can you pastebin the logs I asked for earlier? [13:09] i wonder if my background-image is animated :) [13:09] larsu, with lightdm.log [13:12] seb128: at the spot you added the if (flags & GTK_STYLE_PROPERTY_ANIMATED) g_object_set (node, "animated", TRUE, NULL); can you add if (flags & GTK_STYLE_PROPERTY_ANIMATED) g_print ("animating!\n"); ? [13:12] i think flags is not getting set properly.... [13:13] Name: background-image, animated: 1 [13:13] should be this.... [13:13] desrt, do you want me to try with -O0 as well to see the difference? [13:13] sure... [13:14] make running [13:14] seb128: in fact, if you could add an unconditional g_print ("name: %s flags: %x\n", name, (guint) flags)); ... [13:14] would be most useful [13:15] seb128: http://paste.ubuntu.com/5798489 http://paste.ubuntu.com/5798491 http://paste.ubuntu.com/5798492 http://paste.ubuntu.com/5798494 [13:15] * larsu is on a vt and typed those by hand... [13:16] larsu, init: gnome-session main process (6895) killed by SEGV signal [13:16]