[07:43] morning [07:54] hey didrocks [07:56] evening darkxst [07:58] larsu: after fiddling more with symbolic icons, i can only say this is really worth it! it's super easy and super failsafe, since it falls back to non-symbolic icons automatically if there are non found. so to the worst you get your current behavior [08:01] didrocks, is the timezone master ;) [08:04] ahah [08:15] bonjour didrocks [08:15] hey darkxst [08:16] hey pitti [08:20] hey didrocks pitti ochosi darkxst [08:20] morning seb128 [08:20] bonjour seb128, ça va ? [08:20] pitti, oui, très bien, c'est vendredi ! et toi ? [08:20] seb128: j'ai un rhume, je me sense faible :/ [08:23] pitti, :-( [08:23] pitti, repose toi ! [08:24] pitti: http://fun.sdinet.de/pics/english/Fahrrad.jpg [08:24] *chuckle* [08:25] hey seb128 [09:10] hello! [09:10] hey Laney [09:10] soooooooooo does anyone else have to frequently restart n-m to get networking to work on resume? [09:11] as a bonus, this time restarting nm locked the machine up [09:11] hi didrocks! [09:12] hey Laney, wie gehts? [09:12] Laney: hum, not yet at least :) [09:12] Laney, no, didn't have n-m issues since brussels here [09:12] ofono was stopped btw [09:13] hey seb128 [09:13] not bad, got tea and I can see the sky [09:13] you? [09:14] I'm good, thanks :-) [09:18] Hello from another airport desktoppers [09:18] Did anything burn down this week? [09:19] hey willcooke! [09:19] hey seb128 [09:19] no, we know how to party without burning the house don't worry [09:19] ups, did I say party? no no, we worked hard this week [09:19] nothing to worry about :p [09:19] XD [09:20] how was WMC? [09:20] Thanks for yours and didrocks reply on that email thread about MWC - it looks like we're heading in a better direction now [09:20] MWC was AWESOME! [09:20] Everybody loves the convergence demo. Lots of people coming back to show their friends and colleagues [09:21] Some Ubuntu fans even gave it a round of applause [09:21] :D [09:21] willcooke: excellent! I hope you had fun :) [09:22] nice [09:22] willcooke: next step, adding a clapping sensors to transform in desktop mode! [09:22] It was fun, but really hard work. Standing up from 8am to 7pm with no proper breaks. [09:23] try to not get the ubuflu on the way back! [09:23] yeah, and standing the whole day, not used to that anymore working from home desk :) [09:23] mwcflu :p [09:24] :D [09:25] willcooke: enjoy seating in the plane at least ;) [09:27] :D [09:49] Laney, I need some advice regarding a package update. [09:49] Laney, Please could you cast your eye over my last reply in this bug please. [09:49] https://bugs.launchpad.net/ubuntu/+source/mate-menu/+bug/1427742 [09:49] Ubuntu bug 1427742 in mate-menu (Ubuntu) "mate-menu package needs updating" [Undecided,New] [09:50] Just want to know what I should do about the changelog entry for a version that was never release to Ubuntu. [09:53] hi flexiondotorg_, usually I'd update that one instead of adding a new one on top [09:53] so each entry corresponds to one upload in ubuntu [09:58] So merge the last code review into the new one and skip the release that never happened? [10:00] Sounds right, you want people to be reviewing the changes between today's archive and what you want to upload [10:00] Laney, Thanks. I'll do that now. === mzanetti_ is now known as mzanetti [10:36] larsu: doing a gtk upload to fix that messagedialog-icon thing, thought I'd take the commit to add a CSS class to title buttons too [10:36] are there any other ones you need? [10:39] hi folks, wonder if someone with indicator-location insight would look at this crasher: [10:39] https://bugs.launchpad.net/ubuntu/+source/indicator-location/+bug/1419405 [10:39] Ubuntu bug 1419405 in indicator-location (Ubuntu) "indicator-location-service crashed with SIGABRT in internal::ToBackend::exit_module()" [Medium,Confirmed] [10:39] seems to have been around a while, regular crash on startup [10:45] sabdfl: Not sure if this is meant to work on current desktops (it's not installed by default there) - should exit cleanly instead of crashing though if that's the case. [10:45] charles: ^? [10:47] Sounds unfriendly for platform-api to be aborting... [10:48] yes indeed [10:48] it may have come in with unity8 which i have on this machine [10:48] but still [10:50] likely [10:51] technically you should have a dummy location service running [10:51] it's aborting on some missing configuration file [10:51] hmm, tvoss isnt in here [10:51] see the message in the trace [10:56] * ogra_ wonders with ubuntu-location-* services are installed on desktop-next [10:57] (in /etc/init) [10:57] does it have this /etc/ubuntu-platform-api/application.conf file? [10:57] or maybe do you just not see the crash ;-) [10:57] i dont have a desktop unity8 install here :) [10:58] im trying to think about the cause [10:58] in the phone there are some overrides [10:58] if indicator-location-service is started from the context of the unity8 session it should have the platform defined [10:58] https://errors.ubuntu.com/problem/375ff448c30eaac105e835eda789b285b14bf025 [10:58] this is in unity 7 [10:59] Just run it in your normal desktop environment [10:59] oh, do we seed it there ? [10:59] no [10:59] but if you have it installed for whatever reason ... [11:01] well, indicator-location seems to not have a dep on ubuntu-location-service [11:01] which makes me suspect you dont have a backend installed at all in this case [11:01] i.e. not even the dummy [11:03] thats not it [11:03] I definitely don't have this /etc/ubuntu-platform-api/application.conf file that it wants [11:03] right [11:04] whats the name of the desktop u8 session package? [11:04] you can just apt-get install indicator-location [11:04] ubuntu-touch-session [11:04] the desktop package [11:04] why do you need all of that? [11:04] it ships that file [11:04] this is a minimal package to reproduce [11:05] who cares? it's valid to install just indicator-location [11:05] nothing should crash [11:05] yea, no one is contradicting you [11:05] trying to understand why unity8 and applications find the plugin while indicator location doesnt [11:06] it'll be because something has pulled in ubuntu-touch-session [11:07] not really [11:09] fine, you take care of it then [11:10] will do indeed [11:11] Laney: so I remember a time when there was a unity8 override that provided the desktop session initialization [11:11] now all I see is different stuff in unity8-desktop-session-mir [11:15] To work it'd have to be applied to unity7 sessions too [11:16] ok so [11:17] indicator-location is now being started by the unity7 session? [11:17] if you have it installed, yeah [11:18] I think you want it to exit cleanly if it can't start the location service controller [11:18] but since platform-api aborts you in that case ... [11:18] kind of hard [11:19] is it started from the upstart job ? [11:20] as a quick fix you could just add a check for the file to a pre-start script [11:20] no no [11:20] http://bazaar.launchpad.net/~unity8-desktop-session-team/unity8-desktop-session/trunk/revision/51 [11:20] with the new scripts we basically dont have that env exported, as they are only exporting the QPA now (for unity8 session) [11:21] for the failing case under unity7, thats a different thing as we just dont have a need for the platform-api there [11:21] *however*, depending on how indicator-location is started we can just point it to the dummy provider [11:21] well, then you need a package dep [11:22] dep on what, platform-api ? [11:22] nothing guarantees you the location service is even installed currently [11:22] the location service being installed or not is irrelevant [11:22] platform api supports three platforms [11:22] it provides the dummy provider, no ? [11:22] touch_mir, desktop_mir and test [11:23] or do we have another internal one in p-api ? [11:23] imo unity7 doesnt fall on any of those three, so if unity7 starts the location it should just dunny out [11:24] so the question is, who starts indicator-location on unity7? does it ship an upstart job? [11:24] all indicators do nowadays [11:24] i think [11:27] indeed [11:27] one workaround would be to change the job of course [11:28] no, just wanted to know, changing it is wrong as its made to start on indicator-services-started or whatever the signal is [11:29] so if im correct it should also be crashing under the unity8 session [11:29] right? [11:29] depends where [11:29] in desktop-next ubuntu-touch-session is seeded [11:29] seriously? [11:29] why would they do that [11:29] if you plainly install unity8 it wont be dragged in [11:30] desktop-next uses (nearly) the same seed as touch [11:30] sure, but application.conf there clearly says touch_mirclient [11:30] which means hybris sensors, which means blow up [11:31] the only reason for which its not blowing up is that the qtubuntu plugin has extra checks in place for that [11:31] but any non-qt app that would like to use sensors would segfault [11:31] ogra@anubis:~/Devel/seeds/ubuntu-touch.vivid$ grep ubuntu-touch-session * [11:31] touch: * ubuntu-touch-session [11:31] I can only imagine there are other touch-specific bits in ubuntu-touch-session [11:31] false alarm ... not seeded [11:31] alright, good [11:31] so it should be crashing then [11:31] if it was then the error wouldn't talk about the file being missing [11:32] as the env var was dropped when desktop changed from the unity8-mir upstart job [11:32] unless there are that many people installing it bare [11:32] ogra@anubis:~/Devel/seeds/ubuntu-touch.vivid$ grep session * [11:32] desktop: * unity8-desktop-session-mir [11:32] touch: * ubuntu-touch-session [11:32] there we go [11:32] Laney: could you confirm that it crashes on unity8 as well? [11:32] not easily, maybe seb128 can, afaik he has a desktop-next installation [11:33] seb128: ^ could you give us a hand with that? [11:34] if test -z "$DISPLAY" && [ $DESKTOP_SESSION != "unity8-lxc" ]; then [11:34] export QT_QPA_PLATFORM=ubuntumirclient [11:34] export UNITY_INDICATOR_PROFILE=desktop [11:34] from unity8-desktop-session [11:35] yea, thats why I say it should crash too [11:35] from first glance, the UBUNTu_PLATFORM_API_BACKEND var was dropped [11:36] when it moved from unity8-mir.conf to these helper scripts [11:36] hmmm, is gtk3-icon-browser not packaged at all? [11:36] should be a part of gtk3.14, but i can't seem to find it at all [11:38] ricmm: yeah, okay, just pulled out the inspiron & it is crashing there [11:38] Laney, ricmm, want me to confirm something? [11:38] no worry [11:41] ochosi: no, can probably put it in gtk-3-examples tho [11:42] Laney: that would be really great! [11:42] maybe along with gtk-encode-symbolic-svg ? [11:42] those two are really important for porting stuff to gtk3 and symbolic icons [11:42] what is that? [11:42] symbolic icons you mean? [11:42] or the encoder [11:42] no [11:43] the encoder creates pngs from symbolic svgs which are still recolorable, but are loading faster [11:44] so you can use that in a script and convert your symbolic icons in humanity or wherever you keep them to .symbolic.png icons [11:44] gtk3.14 loads them automatically, and faster than with svg [11:45] https://developer.gnome.org/gtk3/stable/gtk-encode-symbolic-svg.html [11:46] ok, yeah, sounds like we can install that too [11:47] sweet [11:47] thanks a lot! [11:47] we'll use that in xubuntu for sure [11:48] so the unity8 crash is a regression as that env var was dropped from the environment [11:48] however what we could do instead of provide the module config files from the platform-api module package directly [11:48] install to /usr/share/ubuntu-platform-api/ or something [11:48] and make touch/desktop conflict/replace [11:48] how do you know in which env you are then ? [11:49] what? [11:49] well, the config changes based on your environment, no ? [11:50] no, env var can override whats in the config file [11:50] you either install the touch plugin, or the desktop plugin [11:50] ah, k [11:50] but not both [11:50] yeah [11:50] right now we are relying on either the env var being there (which it used to be in utopic for unity8-desktop-session-mir) or the application.conf file being there [11:51] so my suggestion is to ship the file with each plugin, and make both packages conflict [11:51] so if some -next dependency pulls in something that uses platform-api, it will guarantee that the module conf file is in place [11:51] Laney: do you want me to report bugs about those two to keep track of them? [11:51] no [11:51] I was doing a gtk upload alraedy [11:51] oh cool [11:52] Laney: do you agree with that approach? [11:52] Laney: thanks a lot! [11:56] ricmm: I don't really like having packages which both supply the same file forever [11:56] Would be best if this could be dynamic - i.e. try to initialise each module in turn until one works [11:56] but in the absence of that I probably prefer setting environment variables [11:57] its gonna have to be an env var then [11:57] UBUNTU_PLATFORM_API_BACKEND=desktop_mirclient needs to be part of both unity7 and unity8 sessions [11:58] it used to be part of the unity8 one, so needs to get back in that file ogra pasted earlier [11:58] it works on x? [11:59] yes, it works, the UI bits are unused outside of the unity8 session [12:00] ok so you also have to have a dependency which pulls it in [12:00] it is pulled in by the unity8-desktop-mir-session [12:00] you don't get this if you just install indicator-location [12:01] indicator-location depends on ubuntu-application-api [12:02] yeah, that only gets you 'test' afaics [12:02] yea, it is up to the session to set up its environment if you want more than that [12:02] consideri t like QT_QPA_PLATFORM=dummy [12:03] Qt doesnt have the depend on qtubuntu-desktop or qtubuntu-android, thats up to the session to provide and setup [12:03] no? [12:03] I don't really care if it doesn't work without setup [12:04] I care that it doesn't crash though [12:04] and particularly that it doesn't crash on startup [12:08] Laney: easiest way is to add the env to the unity8 session, and have unity7 default to the dummy plugin [12:09] if you dont care about it working, just not crashing [12:11] I don't think it's the end of the world to have to install an extra backend to get it working [12:11] does dummy provide fake data or just no data? [12:11] also if I run UBUNTU_PLATFORM_API_BACKEND=test /usr/lib/x86_64-linux-gnu/indicator-location/indicator-location-service I get an assertion failure [12:12] so? but it doesnt crash [12:12] cant expect it to work with the dummy [12:12] nope [12:12] I'm just saying [12:13] if you want it to *work* it needs desktop_mirclient [12:13] but its illogical for idnciator-location to depend on that [12:13] as it runs on both touch and desktop seeds [12:13] if you want it to not crash, test is enough [12:14] I'd rather it didn't crash in the no configuration case too [12:14] if that was the case then we would only have to fix unity8 [12:14] yea if we are going to do test I'l just make it default to test if theres no other selection [12:14] indeed [12:15] ok, will do it that way [12:17] might be cleaner to get _create_controller() to return nullptr instead of aborting, then i-location could handle the case cleanly and just exit(0) [12:17] thats what it does when "test" is selected [12:17] ah [12:18] well, indicator-location can do whatever it want, currently it doesnt exit but it stays running [12:18] but it doesnt crash [12:18] Indeed that assertion shows it does return null then [12:25] Laney: ok I'll get a silo for platform-api, who can add the env var to the unity8 mir conf? [12:31] ricmm: hmm, there's no separate desktop job any more, just uses the main unity8 one [12:31] theres a script that exports the QPA [12:31] thats where this var should go [12:31] which job is that? [12:32] http://bazaar.launchpad.net/~unity8-desktop-session-team/unity8-desktop-session/trunk/revision/64 [12:33] lightdm-unity8-session [12:33] ok, not upstart [12:33] yea === MacSlow is now known as MacSlow|lunch [13:00] moin! [13:02] Ah, seb128 is cleverly hiding from me as I have a LibreOffice upload for him ... ;) [13:02] well, friday uploads are a bad idea anyway. [13:14] Laney: so whos the right guy for the desktop session change? [13:14] did you see the MP? [13:15] looking [13:15] bregma presumably [13:15] Laney: could you confirm on the inspiron that with that change it doesnt crash in the unity8 session? [13:17] we'll need ricmm's approval that it's a sane change then we'll test it in our lab and toss it in a silo [13:17] 'lab' sounds scientific [13:18] but it makes me sound like a professional [13:18] well I've added to a silo I requested for the platform-api change [13:18] but yea it looks fine, if someone could confirm on the desktop session that the indicator wont crash [13:18] one of the test machines in my lab sits on a stack of 3 dead laptops so its display is at eye height [13:19] Laney, is there also a fix for the indicator-location crash under Unity 7? [13:20] theres a platform-api in the same silo-in-wait [13:20] that makes it default to the dummy [13:23] ricmm: just tried & i-l-s is running now [13:23] so seems ok [13:23] ok [13:23] bregma, Laney: I'll ping you with the silo once platform-api is there too to try the unity7 one [13:23] I've tried here and looks fine [13:24] well, looks fine in the sense that it doesnt crash just uses the dummy (so no functionality) [13:24] 'k === MacSlow|lunch is now known as MacSlow === jorge_ is now known as jcastro [14:10] didrocks: ping [14:10] didrocks: i just wanted to remind you about how great i think you are [14:10] didrocks: keep it up [14:35] you're right, didrocks is a great guy [15:19] Laney: vivid 13 has the two packages [15:19] if you'd like to give it a spin on your test laptop === jhodapp_ is now known as jhodapp [16:44] Laney, thanks for https://code.launchpad.net/~laney/unity8-desktop-session/platform-api-export/+merge/252105 [16:45] charles: np [16:45] might want to consider making i-location exit if it can't init the service === jfunk is now known as jfunk-lunch === alan_g is now known as alan_g|EOW [18:06] Laney, not sure how I could make i-location pre-empt that abort() call; it's happening in platform-api when i-location calls ua_location_service_create_controller() [18:06] ie, there's not a way to test first [18:07] ricmm's platform-api change makes it return nullptr now [18:07] IIUC [18:08] charles: yes, if you'd like to test silo 13 vivid it should be fine now [18:08] sure enough, https://code.launchpad.net/~ricmm/platform-api/no-abort-default-dummy/+merge/252108 [18:08] as it will default to the dummy if theres no module selected [18:10] * Laney is off [18:10] sorry I didn't get to test, will do monday if you still need it then === jfunk-lu_ is now known as jfunk