/srv/irclogs.ubuntu.com/2013/11/29/#ubuntu-desktop.txt

* robert_ancell heads to lunch, be back in 30-6000:00
ali1234i can report a bug, but what do you want me to reproduce with? trusty, ubuntu-desktop in virtualbox (assuming i can?)00:01
ochosiali1234: that would be great, i'm not in a good mindset anymore now to write something that makes sense..00:03
ali1234also... maybe we're only seeing it now because of my patches, which change the RetainPermanent stuff in lightdm-gtk-greeter, xfwm4, and xfdesktop00:03
ochosithink i'll head to bed and look at this again tomorrow00:03
ochosimakes more sense00:03
ochosiali1234: i'm not using your patches atm00:03
ali1234ah yes, good point00:03
ochosiand i had that issue way before you started working on this00:03
ochosis/that/these/00:04
ali1234i'll do some tests with clean isos in virtualbox00:04
ochosiso you're certainly not to blame ;)00:04
ochosicool00:04
ochosii'll check in again tomorrow00:04
ochosinight everyone!00:04
ali1234well, it didn't happen with trusty daily in virtualbox00:51
robert_ancellali1234, oh really? I could reproduce it reliably in trusty yesterday00:54
ali1234i reproduced it with the guest session00:54
robert_ancellseb128 said that downgrading to 1.8 seemed to make the problem go away00:54
robert_ancellali1234, but not lock screen?00:54
ali1234it's put a greeter on top of my user session :/00:54
ali1234i couldn't reproduce it by login, logout, login, lock00:55
robert_ancellthat sounds bad!00:55
ali1234switched to virtualbox console and the display has gone completely mad00:55
ali1234i didn't install the guest drivers either00:56
ali1234lightdm just bombed out and sent me back to the login screen00:56
ali1234i'm not convinced this is the same issue. i'm going to try on real hardware00:57
ali1234actually now i think about it, what i saw in virtualbox sounds like what someone else described with trusty01:29
ali1234and from looking at it, it looks like two xservers are fighting for the same framebuffer01:29
ali1234so a theory: on saucy the second x server simply fails to load because the first one owns the framebuffer. on trusty it somehow gets past that and they fight...01:30
ali1234or perhaps the result is X server implementation specific01:30
ali1234ah... again, i can only reproduce with the guest session01:34
ali1234and the result is the same: intel x server fails to start due to permission error01:35
ali1234it seems like locking the screen on unity/trusty does not create a second X server instance. i only have Xorg.1.log and not Xorg.1.log.old01:37
ali1234robert_ancell: what package do you want me to report it against?01:38
robert_ancellali1234, mark it against lightdm and I'll also mark it against xorg01:38
ali1234"the problem cannot be reported. this is not an official ubuntu package..."01:40
ali1234i just installed from the daily iso...01:40
ali1234on the console: cat: /etc/lightdm/lightdm.conf: No such file or directory01:41
ali1234i did touch /etc/lightdm/lightdm.conf and now the report is uploading....01:45
ali1234bug 125615001:48
ubot2Launchpad bug 1256150 in lightdm (Ubuntu) "Xorg guest session fails to start if the user has logged out and logged in again" [Undecided,New] https://launchpad.net/bugs/125615001:48
robert_ancellali1234, thanks!02:01
ali1234"log out" has now stopped working completely02:02
robert_ancell!02:03
robert_ancellI have to go, but I will have a look next week at this02:03
ali1234i'm going to wipe this install for trusty/xubuntu now, the bug should still be easily reproducable there, and i need to test a bunch of things02:03
ali1234ah... log out just took 4 minutes for no reason...02:04
pittiGood morning07:23
Mirvhello07:25
JackYumorning07:32
damianatorrpmGM all :)08:06
damianatorrpmlarsu: how are you :)08:07
larsudamianatorrpm: good morning. I'm good thanks. How are you?08:32
seb128good morning desktopers08:33
larsudamianatorrpm: I saw your emails. Do you get any other output on stdout?08:33
damianatorrpmlarsu: Thanks. I'm fine too, it's Friday :-)08:38
damianatorrpmlarsu: no other stdout output. I also started QtCreator from terminal to see for any more output, nada. I also tried qmlscene and run the main.qml08:39
damianatorrpmlarsu: stii no difference.08:39
damianatorrpm*still08:40
larsudamianatorrpm: hm, I'm starting to run out of ideas :-/08:43
damianatorrpmlarsu: Can I show you via teamviewer?08:43
larsudamianatorrpm: never used that. Do they have a linux version?08:45
damianatorrpmlarsu: yes :) .deb for 64 and 32 bit Ubuntu as well as .rpm for suse and fedora http://www.teamviewer.com/en/download/linux.aspx08:46
damianatorrpmlarsu: It's free of charge for personal use and works *always*08:46
damianatorrpm:)08:46
=== jhernand1z is now known as jhernandez
larsuand is not free software :P  (I'll install it, give me a sec)08:47
damianatorrpmlarsu: really appreciated. Thanks08:48
Laneymorning09:02
larsuhi Laney, how's it going?09:02
Laneypretty good, happy to have made it to friday ;-)09:02
Laneyyou?09:02
larsusame same :)09:04
Laneywhat's this dbus-daemon thing?09:07
didrocksLaney: ps4 on the way? ;)09:09
Laneydelivered 10:04-11:04 :D09:09
didrocks;)09:10
Laneydidrocks: did you go out for a midnight launch? ;-)09:11
didrocksLaney: no, I don't plan to buy it before a year at least09:11
Laneyoho09:12
didrocksI want games I'm interested in on the platform first ;)09:12
didrockslike watch dogs09:12
Laneyfair enough09:12
seb128back09:20
ogra_front09:20
seb128Laney, ogra_, hey, happy friday ;-)09:21
ogra_same to you :)09:21
Laneyhey seb12809:21
seb128Laney, how are you?09:21
seb128Laney, you made rb unhappy, no music today :-(09:23
Laneydoes banshee work? :-)09:23
seb128lol09:24
Laney(good thanks)09:24
seb128do I look like I use GTK2 apps still?09:24
Laneythat is one sexy GTK2 app09:24
seb128you are so old fashion... ;-)09:24
seb128(but yeah, banshee works ;-)09:25
seb128(you did that on purpose to made people use banshee, admit it!)09:25
Laney:-)09:25
Laneyanyway, I saw the bug, hopefully we can get a fix today09:25
Laneywe should get that i-power fix out too09:26
seb128yeah09:26
seb128hum, glib migrated to trusty09:26
seb128(it hadn't yesterday due to libffi but infinity fixed that)09:27
seb128Laney, btw did you see my ping about GTK build?09:27
Laneyyes09:27
Laneywant to understand what changed there09:27
seb128ok09:27
seb128just making sure it didn't get lost in the backlog09:27
seb128it seems suboptimal if we have to make glib-dev depends on dbus09:28
LaneyI don't see any changes in that code in glib09:28
seb128hum, maybe GTK started used something in glib that was not used before?09:29
Laneychecking that09:29
seb128speaking about indicator-power09:32
seb128larsu, did you ever get desrt to +1 or -1 your action change?09:32
Laneyah, I think it's a new Ubuntu patch09:33
Laneygtk-object-tests-run-under-local-environment calls g_test_dbus_up()09:33
seb128oh09:34
Laneywhich uses dbus-daemon, but that's nothing new09:35
Laneyso I'd probably add the BD09:35
seb128that seems wrong though09:35
seb128if glib provides an api09:35
seb128it should bring with it what is needed for the api to work09:35
seb128or handle the case where the depends are missing...09:36
Laneyit would be quite heavy though http://162.213.35.4/search?weighted=1&q=g_test_dbus_up09:38
seb128the indicators use dbus-test-runner so b-d on it, and that brings in dbus09:40
Laneyyeah09:40
seb128but yeah, options are09:40
seb1281- add a dbus depends to libglib2.0-dev09:40
seb1282- workaround it by adding a dbus b-d to packages using that api09:40
Laneydbus-x1109:40
larsuseb128: he said he needed to drink coffee first. And then he didn't ever look at it09:41
larsuI'll ping him again today09:41
seb1283- split the glib files out in a new binary which has the depends09:41
seb128larsu, thanks09:41
seb128I don't like any of the 3 options :/09:41
seb128but I guess the b-d is the less annoying one09:42
Laneysame09:42
Laneymaybe glib could improve its message in that case, but it's quite obvious currently anyway09:42
seb128let's do a ppa upload with dbus-x11 b-d09:42
Laneyk, ta09:42
seb128thanks for looking at it!09:43
Laneyxnox: Do you have some time to help out on those libtimezonemap reviews?09:43
LaneyI feel bad for not looking at them sooner, but I honestly have never looked at this code before09:44
Laneykind of got dumped in it there really09:44
xnoxyeah. i should review them. they do look good.09:44
Laneyhe seems to have done 1 commit per change-ish09:44
Laneyso I'll take some of them and cherry-pick them09:44
Laneyomg bzr-gtk went away?10:02
seb128yes, in saucy10:03
Laneyguess the fact that I only just noticed says something :P10:04
seb128it needed porting to the new version of something and nobody wanted to do the work iirc10:04
seb128desrt, wake up, I need you :p11:03
seb128desrt, n-m is having an issue similar to rhythmbox, http://paste.ubuntu.com/6493541/11:09
seb128desrt, I'm going to have a look at reverting this glib commit11:09
seb128Laney, ^ or do you want to do that?11:09
Laneyseb128: go ahead, would be good to confirm it actually does fix things before uploading it though11:11
seb128Laney, right, I'm not going to upload glib without testing, no worry11:12
seb128I'm going to wait for desrt to be up, but I can as well start testing building/running it11:12
Laneyseems unsafe anyway if we've discovered two applications in the first day11:12
Laneythey might want to rethink it upstream11:12
seb128right11:12
=== MacSlow is now known as MacSlow|lunch
seb128shrug12:15
seb128GTK build still not happy :/12:15
seb128https://launchpadlibrarian.net/157877308/buildlog_ubuntu-trusty-i386.gtk%2B3.0_3.10.5-0ubuntu1~build1.3_FAILEDTOBUILD.txt.gz12:15
seb128  /properties/GtkColorChooserDialog:12:15
seb128(/build/buildd/gtk+3.0-3.10.5/debian/build/shared/testsuite/gtk/.libs/lt-object:20493): GLib-GIO-ERROR **: No GSettings schemas are installed on the system12:15
seb128FAIL12:15
seb128I guess it's looking in the system location, not the source12:16
seb128larsu, ^12:16
seb128brb, session restart with glib and the constructore change reverted12:20
Laneyxnox: reviewed a few of the changes, left some for you12:26
ali1234seb128: so, you know this lightdm/guest session bug... apparently it went away when you downgraded lightdm to 1.8? but i can reproduce it in saucy, which uses 1.8.412:27
seb128ali1234, hey, weird ... it's rock solid for me with 1.8.412:27
seb128ali1234, it maybe is a timing issue and I just happen to be one side of the timing issue with 1.9 and on the other side with 1.812:28
ali1234we reproduced it in saucy not with guest session, but with the screen lock dialog, which xubuntu opens on :1, causing an identical problem12:28
ali1234it's not a timing issue, lightdm actually tells Xorg to run on the wrong VT12:28
seb128ok12:28
seb128is that the same one?12:28
ali1234well, maybe, maybe not12:29
ali1234but it's reproducable the exact same way12:29
ali1234if you've logged out and logged in, and then ask lightdm to start a display on :1, it will attempt to start it on VT7, which is already used by :012:29
seb128in trusty I get xorg hitting a "no screen found" error12:29
ali1234the result depends on driver. you get no screen found with intel12:30
ali1234with others, you get two xorgs on the same VT and they fight each other12:30
seb128right, I'm using intel12:30
seb128right12:30
seb128for me it's like a get a vt over my xorg12:30
ali1234but checking lightdm logs, it's explicitly telling Xorg to use the wrong VT12:30
seb128e.g my vt7 is blank, but I can move the cursor and see it change shape where there are windows from my session12:30
seb128ok, good12:30
seb128seems like a bug for robert_ancell to fix then12:30
seb128ali1234, thanks for figuring that out!12:31
ali1234alreday reported. he's the one who told me you said 1.8 fixes it12:31
seb128well, it does for me12:31
ali1234perhaps it handles guest and "lock" slightly different12:31
seb128could be12:31
ali1234and only one got fixed12:31
seb128what I do is12:31
seb128- log in12:31
seb128- log out12:31
seb128- log in12:31
seb128- start a guest from indicator-session12:31
seb128that works in saucy12:32
seb128but screws my session with 1.912:32
ali1234i haven't tried that on saucy, what we do is same: log in, log out, log in, lock screen with light-locker (that tries to start :1 and does exactly the same thing in saucy and trusty)12:32
ali1234guest session also does it in trusty, but never tested that in saucy cos we didn't know that was also an issue at the time12:33
seb128ok12:33
seb128ali1234, yeah, I can confirm that in the log I sent to robert_ancell12:35
seb128DEBUG: Launching process 4073: /usr/bin/X -core :1 -auth /var/run/lightdm/root/:1 -nolisten tcp vt7 -novtswitch12:36
seb128well, let's see if resolving that issue fixes things for the different scenarios12:36
ali1234yeah. bug 1256150 btw12:36
ubot2Launchpad bug 1256150 in xorg (Ubuntu) "Xorg guest session fails to start if the user has logged out and logged in again" [High,Confirmed] https://launchpad.net/bugs/125615012:36
seb128right, I saw it in my bug emails box this morning12:38
=== MacSlow|lunch is now known as MacSlow
larsuseb128: seems related to my patch for https://bugzilla.gnome.org/show_bug.cgi?id=71171513:09
ubot2Gnome bug 711715 in general "gtk object tests: run under local environment" [Normal,New]13:09
larsuseb128: (sorry, I was out for lunch)13:09
seb128larsu, no worry, I tested the hack debian is doing13:10
seb128larsu, https://launchpadlibrarian.net/157881868/gtk%2B3.0_3.10.5-0ubuntu1~build1.3_3.10.5-0ubuntu1~build2.diff.gz13:10
seb128larsu, tested -> I'm testing rather (it's building)13:11
larsuseb128: ya that'll probably work. I'm waiting for desrt to tell me how to do it properly for the tests13:14
=== psivaa is now known as psivaa-lunch
seb128larsu, I hope desrt has coffee before looking at IRC, seems there is a stack of questions queued for him today13:16
larsuseb128: he should move to a better time zone13:18
seb128we have the best tz and the best cities, no reason to not come ;-)13:19
larsuhehe :)13:20
seb128larsu, btw Cimi reviewed on overlay-scrollbar changes and just pinged me on #ubuntu-devel, I asked him to join here/on a channel where you are as well13:22
seb128larsu, seems like he picked another channel...13:22
desrthello hackers!!13:48
desrtseb128: et tu, n-m?13:48
seb128desrt, hey, how are you?13:48
desrtokay13:49
desrti got coffee beside me :)13:49
seb128desrt, should I revert the 2 commits from dan? just 1? just comment the g_error?13:49
desrti'd do both13:49
seb128I did that13:49
seb128just wanted to check with you before uploading13:49
seb128though it refuses to build on amd64 in the ppa13:50
desrtsigh13:50
seb128FAIL: gwakeup13:50
seb128FAIL: gwakeup-fallback13:50
popeylooks like we have the "hud-service eating my RAM" issue back again in 13.10.13:50
seb128but of course no details in the build log13:50
seb128popey, talk to pete-woods on #ubuntu-touch13:50
desrtseb128: that seems highly suspicious13:50
popeybug 125359313:50
ubot2Launchpad bug 1253593 in hud (Ubuntu) "hud memory usage grows over time" [Undecided,Confirmed] https://launchpad.net/bugs/125359313:50
popeyok13:50
seb128popey, wrong channel13:50
desrtseb128: considering your reverts are in libgobject13:50
popeythis is on the desktop though ☻13:50
seb128popey, you want ted or pete, ted is off for thanksgiving13:51
popeynot phone.13:51
desrtseb128: and the gwakeup test doesn't link against that13:51
seb128desrt, right, so maybe another reason/something with the ppa13:51
seb128desrt, https://launchpadlibrarian.net/157875681/glib2.0_2.39.1-0ubuntu1_2.39.1-0ubuntu2~build1.diff.gz13:51
desrtseb128: maybe a new kernel version?  wouldn't be the first time this testcase found a kernel bug...13:51
seb128desrt, can you sanity check? (ignore the "git" file in there)13:51
desrt+++static inline gboolean13:52
seb128desrt, well, it's basically reverting those 2 commits13:52
desrtyou said SANITY? ;)13:52
seb128lol13:52
seb128desrt, yeah, not a lot to check, that are the 2 upstream commits13:52
seb128desrt, you saw the pastebin in the backlog (from nm)13:52
seb128?13:52
seb128desrt, can you upstream that one as well?13:53
desrtno13:53
desrtgot it13:53
seb128cool13:53
desrtthe good news is that these projects were legit buggy/leaking13:53
desrtand now we find out about it13:53
* desrt files a bug for the other danw13:54
desrtseb128: of course everyone in USA is on their 5-day eatfest/buyfest14:01
seb128desrt, oh, right14:01
desrtthat includes both danw and dcbw, i guess14:01
seb128desrt, well, anyway with the revert that can wait14:01
desrtkinda curious about this gwakeup issue14:01
desrtthe irony builds14:03
desrtthe last person to touch this file.... you guessed it.... danw14:03
desrt(in fairness, though, dcbw wrote the offending code)14:04
didrocksseems like a running joke14:04
didrocks;)14:04
desrtdidrocks: the joke started last night14:04
didrocksdesrt: yeah, seb128 told me :)14:04
didrockswith your bug report?14:04
desrtwith seb seeing rhythmbox crashes14:04
didrocksyep14:05
didrocksand who wrote the libsoup code14:05
didrocks:)14:05
seb128larsu, gtk 3.10 built in the ppa \o/14:13
larsuseb128: yay!14:14
seb128everyone: ^ new gtk in the desktop team ppa if you want to give it a try14:14
Laneyexciting!14:16
seb128I'm going to follow up on the list with that and the list of known issue14:17
seb128glib with the gobject constructor restriction reverted uploaded to trusty as well14:17
seb128would be nice if people could grab it/test in half an hour when it's built14:17
seb128didrocks: ^ that should fix nm on the touch image14:18
=== psivaa-lunch is now known as psivaa
seb128bbiab14:24
desrtbumpy ride14:26
desrti bet seb's appetite for a taste of gtk 3.12 has subsided14:26
larsudesrt: morning!14:27
desrtlarsu: hi :)14:27
desrtyou finally put o-s to bed?14:27
Laneyah, such fun14:27
larsudesrt: let me know when you had coffee. I have two things for ya14:27
desrtlarsu: i'm coffeed14:27
larsudesrt: no...14:27
larsuI tried hard!14:27
larsudesrt: https://code.launchpad.net/~larsu/indicator-power/use-gsettings-actions/+merge/19706614:27
desrtoh.  that.14:28
desrti forgot to actually comment on that14:28
desrti see lots of red.  oddly enough, that means that i'm in a good mood.14:28
larsuis it okay?14:28
desrtmy mind is blown by how bad this code was before14:29
desrtwhat in the hell....14:29
larsudesrt: ya, we found it because of the non-stateful actions that had their state set later on14:30
larsuthanks for putting that check into gio ;)14:30
desrtwhich was only possible if using g_object_set()...14:30
desrt...which you were ;)14:30
larsuI?14:30
desrtthe collective 'you'14:30
desrtwho wrote this?14:30
* larsu is innocent!14:30
larsucharels14:30
* desrt shakes fist14:31
desrtcharles_: if you're using my APIs and you ever find yourself having to do something so ugly again, please talk to me about it :)14:31
larsuwell, binding the settings isn't that ugly14:31
desrtthe gvalue shuffling...14:31
larsuthat is, definitely14:32
desrti'm actually trying to understand what the old code is doing.  don't see it yet.14:32
desrtokay.  got it.14:32
desrtthe binding is directly manipulating the state of the action from the outside14:32
larsuyes14:32
desrtyou're right... it's not actually _that_ bad14:32
larsuit's very reasonable if you don't know about GSettingsAction14:33
desrtand about GAction...14:33
larsuha, yeah :(14:33
desrtthe API very much has a 'state comes from within' feel about it14:33
larsuwell, simpleaction goes against that14:34
larsuand if you're used to using that....14:34
desrtya... strictly speaking, if implementing a simpleaction (as this code is doing) then it's legit to say that the binding is part of the implementation itself14:34
desrtas it is here...14:34
* desrt always perceives bindings as something outside of an object14:34
larsuI agree, it should have triggered the smelly code detector14:35
desrtmy complaint, specifically: GAction::state is readonly and the binding is modifying the state of a GACtion14:35
desrtbut... it's also a GSimpleAction, so hey...14:35
larsuyeah14:35
larsuthanks for having a second look14:35
desrtso why do you keep the action around?14:35
larsuI don't know14:36
larsuit's charles' code, I didn't want to mess too much with it14:36
desrtthere's almost nothing that you can meaningfully do with a gsettingsaction beyond adding it to the map14:36
larsubut he's on tg holiday14:36
desrtcommented14:37
desrtin summary: i want to see more red :)14:37
larsuthere is more code to be cut, but please let's fix the bug first14:37
larsuI know what you'll say now.14:38
larsu"You'll never get around to removing that anyway"14:38
desrtalso: it will take you less than 5 minutes to do what i asked14:38
desrtand most of those 5 minutes will consist of tapping your 'd' key in a very satisfying way14:38
larsuwell, there are other actions kept around as well14:39
larsuanyway, I'll have a look of you do me a second favor and have a look at how I install the schema in https://bugzilla.gnome.org/show_bug.cgi?id=71171514:39
ubot2Gnome bug 711715 in general "gtk object tests: run under local environment" [Normal,New]14:39
larsuit's causing problems on the builders14:39
larsubut I didn't want to fix it because you probably have a very specific opinion on that14:40
* desrt has no idea what he's looking at14:40
* desrt becomes curious14:40
desrti guess this is some fallout from your gtk 3.10 work?14:40
larsudesrt: yes. Running the tests in the session didn't work14:40
larsubecause schemas were missing if you didn't have 3.10 installed14:41
desrtahhhh14:41
desrti'm very surprised that gnome-continuous didn't pick that up14:41
larsusame problem on the builders14:41
larsubecause its using installed tests14:41
larsuit's14:41
desrtright.14:41
desrtprobably also because it doesn't do clean rebuilds14:41
desrtso even if it did 'make check' it would have a system image there14:42
seb128desrt, glib built on amd64 on the buildders, I'm going to do another ppa upload with VERBOSE=1 just to see14:42
desrtseb128: blame the kernel!!14:42
larsuseb128 is back!14:42
seb128larsu, yeah ;-)14:42
larsuseb128: good news, giving only GtkPaned a background fixes the theme issues14:42
Laneydamn that colonel14:42
larsubut I don't know if there are more14:42
larsuso testing is in order14:42
seb128larsu, \o/14:42
desrt$(GLIB_COMPILE_SCHEMAS) $(top_srcdir)/gtk --targetdir=$(builddir)14:43
desrtoh man...14:43
seb128Laney, please test https://launchpad.net/ubuntu/+source/glib2.0/2.39.1-0ubuntu2/+build/528115114:43
seb128Laney, thanks ;-)14:43
desrtlarsu: strictly speaking, this is no longer valid....14:43
* seb128 doesn't want to be the one who screwed glib on a friday evening14:43
Laneyseb128: anything in particular?14:43
larsudesrt: why not?14:43
larsuwhat's you issue with it?14:43
larsu*your14:43
desrtlarsu: gsettings now assumes the xml will be in the same directory as gschemas.compiled, so this will break out-of-tree14:43
seb128Laney, just make sure rhythmbox & sessions are working on amd64 (I tested on i386)14:43
Laneyok, sec14:44
desrtbut it only breaks if you call _get_description() or _get_summary() in the testcase14:44
desrtwhich i assume you're not14:44
larsudesrt: so ... --target-dir is now useless?14:44
seb128Laney, thanks14:44
desrtlarsu: i was going to remove it... that's why i mention it14:44
desrti stopped using it from the glib testcases14:44
desrtmaybe i'll keep it around if others use it, though14:44
larsuno, please remove it14:45
desrtlarsu: as long as you're not looking at summary/description, you're in the clear...14:45
larsudesrt: I'm not, but this line seems to be causing issues anyway14:45
larsuseb128 is now compiling with this https://launchpadlibrarian.net/157881868/gtk%2B3.0_3.10.5-0ubuntu1~build1.3_3.10.5-0ubuntu1~build2.diff.gz14:46
desrtlarsu: fwiw, i'd also advise not to give a dir like you did there but rather specify the individual schema files14:46
desrtalthough maybe it makes sense this way if people add other tests that use schemas14:47
larsuthis is for the generic test14:47
desrtbut in general, having wildcards in makefile depends is a _big_ no-no14:47
=== Ursinha is now known as Ursinha-afk
larsuI agree14:47
desrti think the gmake manual has a section about why not to do that :p14:47
larsubut I thought people won't think about addind their new schemas to this test14:47
desrtlarsu: they will when the test stops working?14:48
larsubut then, they probably will after running `make check`14:48
larsuya14:48
desrtthere is some --schema-file= flag that you can use14:48
desrtyou can probably use $(addprefix,--schema-file=,$(files)) or something to generate what you need14:48
larsuokay, thanks14:49
larsudesrt: any idea why the test still fails?14:49
desrtlet me apply it14:49
larsuhttps://launchpadlibrarian.net/157877308/buildlog_ubuntu-trusty-i386.gtk%2B3.0_3.10.5-0ubuntu1~build1.3_FAILEDTOBUILD.txt.gz14:49
desrti still don't fully understand what's going on here14:49
larsudesrt: I'm installing the things and then set GSETTINGS_SCHEMA_DIR from inside the test (with g_test_build_filename)14:50
mhr3is trusty screwed?14:54
mhr3https://code.launchpad.net/~ubuntu-unity/+archive/daily-build-next/+recipebuild/59616914:54
mhr3seb128, ^?14:54
seb128mhr3, ask on #ubuntu-devel14:56
seb128seems like a launchpad/soyuz issue14:56
seb128maybe those builders are having a too old apt/dpkg or something14:56
mhr3will do, thx14:56
seb128yw14:56
ali1234so i did a make install of lightdm, and it totally broke the system. what's the fastest way to get test builds installed?14:58
seb128ali1234, build a deb, install it15:00
ali1234lame. that takes ages :(15:00
seb128ali1234, you can also "debuild" once, so it's built with the right configure options, etc15:00
seb128and hack; make; sudo cp binary15:01
seb128if there is one specific binary you hack on15:01
seb128like the main lightdm binary15:01
ali1234that sounds like a plan, thanks15:01
seb128ali1234, yw15:01
desrtlarsu: so it's the installed case that's failing?15:08
larsudesrt: no15:08
larsuI don't think the gtk debian package does installed tests15:09
desrtso "make check" is the fail15:09
seb128yes15:11
larsumake check on an out-of-tree build15:11
larsuseb128: right?^^15:11
seb128yes15:11
larsuseb128 is very efficient15:11
Laneywould be nice to add installed tests too ;-)15:12
larsuLaney: yup, but let's make it build without hacks, first ;)15:13
* Laney fast forwards 60 years to himself as a skeleton in a library covered in dust writing a changelog entry "Enable installed tests"15:13
Laney:P15:14
larsuhaha15:14
* larsu would definitely do different things if he were a skeleton in a library15:15
desrtlarsu: so you have an interesting problem here15:17
larsuno brain?15:17
desrtno...15:17
desrtit's a bit of an annoying catch 2215:17
desrtyou can't use g_test_build_filename() until after you call g_test_init()15:18
desrtbut in these tests, it's gtk_test_init()15:18
larsubah15:18
desrtwhich calls gtk_init()15:18
desrtwhich presumably initialises the schemas sources15:18
larsuright15:18
desrtso by the time you set GSETTINGS_SCHEMA_DIR it's too late15:18
seb128desrt, larsu: as much as I like you figuring that out, can we get the indicator-power fix approved so it lands before the w.e?15:18
seb128then we can go back to fixing stuff that are not hitting users15:19
larsuseb128: good point. Let me fix it up for desrt15:19
seb128larsu, thanks15:19
desrtlarsu: i'll see if i can figure out a way around this.  i have an idea.15:19
* larsu already has it open, was just distracted15:19
larsudesrt: is this the problem why the build fails?15:19
desrtthe _build_ fails?  or make check?15:20
larsumake check15:20
desrtmaybe15:20
desrti'm checking my assumptions and finding them to be false, in fact15:20
desrtit looks like gtk_init() doesn't bring up the schemas sources15:20
larsuseb128: in the meantime, do you have time to test/approve https://code.launchpad.net/~larsu/overlay-scrollbar/fix-for-3.10/+merge/19692015:21
seb128larsu, sure15:21
* seb128 discovers that gnome-keyring dialogs also stopped wrapping with the new GTK15:22
larsuseb128: I have that fix open somewhere as well. Crazy Friday.15:22
seb128larsu, crazy week you can say, it feels like every day has been like that15:23
seb128soon the w.e though!15:23
larsudesrt: now I remember. Charles kept the actions around to disconnect from the notify signal15:23
desrti'd rewrite that using g_settings_bind()15:24
larsuhm?15:24
desrti assume you're using the notify to update some sort of [something] when the value changes, right?15:24
larsuyes, but bind() binds a property, no?15:25
desrt(which means, btw.... your patch may be introducing a bug... since the first notify:: will no longer come as it would have before your patch)15:25
larsudesrt: he's calling the same function after anyway15:25
* larsu checked for that15:25
desrtanyway... the point is maybe a bit theoretical, but it's a good point:15:25
larsuisn't connecting to changed::show-time the right approach?15:25
desrtwhatever that's tied to is interested in what the gsettings say15:26
desrtthe fact that the gsettings is changed via a GAction tied into some UI somewhere is a totally separate point15:26
desrtso to have these two things tied together is inappropriate coupling15:26
larsuyes, I agree. Still, I'd connect it to 'changed' instead of using bind()15:26
desrtie: don't connect to the notify:: signal on the action... connect to the changed signal on GSettings15:26
desrtya.  that's totally fine too.15:27
larsuI think we talked about different things but came to the same conclusion :)15:27
larsuthanks15:27
desrtya15:27
desrti just had bind() in my head because i always try to do that first15:27
larsuright15:27
seb128desrt, https://launchpadlibrarian.net/157891618/buildlog_ubuntu-trusty-amd64.glib2.0_2.39.1-0ubuntu3~build1_FAILEDTOBUILD.txt.gz15:28
seb128desrt, "GLib:ERROR:/build/buildd/glib2.0-2.39.1/./glib/tests/gwakeuptest.c:109:context_clear: assertion failed: (ctx->pending_tokens == NULL)"15:28
desrtlarsu: i don't understand your problem with this 'make check' failure15:28
seb128desrt, just mentioning it, don't bother about it today, it only happens on the ppa builders for some reason15:28
desrti mean... i found the bug, but i don't understand why you didn't find it15:28
desrtlarsu: specifically... you set GSETTINGS_SCHEMA_DIR from the defaultvalue test... and it worked15:29
desrtyou simply did not do the same from the object test...15:29
desrtdo that, and it works too15:29
larsudesrt: the object test failed?15:30
* larsu looks15:30
* larsu slaps forehead15:30
larsudesrt: sorry :-/15:30
desrtlarsu: happy to help :)15:30
larsuthanks a lot. I should maybe stop doing 10 things at once15:31
seb128larsu, sorry about the crazy day/week, I hope it's better next week15:32
larsuseb128: no worries :(15:32
seb128desrt, saw the build link? (just mentioning it so I'm sure it doesn't get lost in the middle of crossed discussions)15:32
larsus/:(/:)/ of course15:32
seb128larsu, speaking of doing 10 things at once, scrollbars work with gtk2 3.8.7 and 3.10.515:33
seb128larsu, I can't change the status on that project though, so I comment approve it15:33
larsuseb128: I love how small the patch has become after cimi reviewed the patch15:33
seb128larsu, ;-)15:33
seb128larsu, so you spent 10 days to write 3 lines of code?15:34
larsuseb128: I can't approve either15:34
* seb128 likes that joke15:34
seb128larsu, sorry, it has been a long week ;-)15:34
Laneyaww15:34
seb128Laney, feel unwell? your ps4 arrived? :p15:35
larsu:)15:36
Laneyfeeling hurt by your mean joke!15:36
seb128oh15:36
seb128lol15:36
seb128it's friday, I need to do some trolling15:37
didrocksseb128: \o/15:37
seb128didrocks, glib works?15:37
didrocksseb128: no, just read your message15:37
Laneygo troll on your menu gnome bugs ;-)15:37
seb128didrocks, or just liking trolls? ;-)15:37
didrocksand back from running :p15:37
seb128didrocks, nice!15:37
didrockstrolls \o/ is just a coincidence15:37
seb128Laney, that's a good point, I spent monday trolling this week15:38
didrockscan't type, need to do one letter after another15:38
seb128didrocks, freezing out there?15:38
* Laney gets back to adding the nicer looking axes to the battery graph15:38
seb128didrocks, or is it raining?15:38
seb128Laney, \o/15:38
didrocksseb128: just freezing15:38
didrocksso my hands can't type15:38
Laneystill don't look that good though15:38
didrocksit's like if I typed in slow motion :p15:38
seb128didrocks, hot shower time:15:38
seb128!15:38
didrocksI guess so ;)15:39
LaneyI look with envy at the gnome-system-monitor graphs15:39
desrtlarsu: we're good?15:39
seb128Laney, those which make the system monitor uses 25% of the cpu displaying you what is using cpu? :-)15:40
larsudesrt: yes, thanks a lot. If you want another look, I just pushed the fix we talked about: https://code.launchpad.net/~larsu/indicator-power/use-gsettings-actions/+merge/19706615:40
Laneyhaha, they seem alright here (but then again I have a nice i7)15:40
Laneyeven so they look damn sexy while doing it :P15:40
seb128(it's better nowadays)15:40
larsudesrt: more red \o/15:40
ali1234back in the day it was more like 75%15:41
* desrt resists urge to complain about connect_swapped15:44
larsudesrt: I hate doing unrelated changes in one commit15:45
desrtone small issue15:45
larsuseb128: o-s is approved by cimi. Does it have auto landing?15:45
desrtyou introduced a race15:46
seb128larsu, looks to the bzr log it seems so15:46
desrtLaney: here's the other one: https://bugzilla.gnome.org/show_bug.cgi?id=71955515:47
ubot2Gnome bug 719555 in general "nm-device constructor() wrongly finalizes object during construction" [Critical,Unconfirmed]15:48
larsudesrt: it's always about races with you!15:48
larsudesrt: which one?15:48
* larsu doesn't see it15:48
desrtlarsu: if a signal comes in while you unref the gsettings object, the signal will still be delivered after the unref because the worker thread will have picked up an extra ref15:48
seb128larsu, just put a sleep(1) in there, that ought to fix a race if there is one15:48
larsulol15:48
seb128;-)15:48
Laneydesrt: ta, subscribed15:49
desrtlarsu: ie: you cannot assume that just because you unref() your [presumed only] copy of the gsettings object, it will stop firing signals15:49
larsudesrt: how did I introduce that? (And why is it a problem?)15:49
desrtlarsu: you don't disconnect from the change signal on the gsettings...15:49
larsudesrt: what? This call was there before, I simply changed it15:50
larsuI assumed it did the unconnect as well15:50
desrti see addition of a connect, but no addition of a disconnect15:50
larsuI don't see an addition of a connect15:50
larsuand I was right it's disconnected from15:51
larsucharles_ didn't dissapoint!15:51
desrtoooh15:51
desrt- g_signal_connect_swapped (p->settings, "changed::" SETTINGS_ICON_POLICY_S,15:51
desrt121- G_CALLBACK(rebuild_header_now), self);15:51
desrt122+ g_signal_connect_swapped (p->settings, "changed", G_CALLBACK(rebuild_header_now), self);15:51
desrtyou just expanded the scope15:51
larsuyep15:51
desrtand it already had a disconnect?15:51
larsuyes15:52
desrtsorry.  misread the patch.15:52
larsujust checked15:52
larsuno worries. Can you approve it please?15:52
larsuat the top, I mean15:52
desrtalready did15:52
larsuthanks!15:53
desrtseb128: larsu didn't need sleep()15:53
desrthe fixes his race conditions using IRC15:53
larsuI fix race conditions by explaining them away15:53
larsumuch better than using sleep()15:53
desrtthis patch now looks very nice :)15:54
desrt+10/-7015:54
larsuI agree. This review thing kinda works15:54
seb128desrt, larsu: thanks ;-)15:54
desrtnormally i like lots of green15:54
desrt(coverage reports, test output, jhbuild results, damned lies)15:55
desrtbut today i like red15:55
larsudesrt: the original bug could have been solved with +2/-015:55
larsuwait, +2/-215:55
desrtadding an initial state of the correct type?15:55
larsuyep15:55
desrtthanks for not doing that :)15:55
larsutwice, for the two actions15:55
larsuya..15:55
larsuI knew you wouldn't let _that_ slide :P15:55
* larsu wouldn't have shown it to you, probably15:56
seb128desrt, I'm asking again in case, saw the glib build log a bit earlier?15:56
seb128too much going on today :p15:56
desrti think i missed that one :p15:56
larsudesrt: speaking of which, did you see the glib build log that seb128 pointed you to earlier?15:56
seb128desrt, https://launchpadlibrarian.net/157891618/buildlog_ubuntu-trusty-amd64.glib2.0_2.39.1-0ubuntu3~build1_FAILEDTOBUILD.txt.gz15:56
seb128 desrt, "GLib:ERROR:/build/buildd/glib2.0-2.39.1/./glib/tests/gwakeuptest.c:109:context_clear: assertion failed: (ctx->pending_tokens == NULL)"15:56
desrtoh15:56
desrtyou said not to bother looking at that one :p15:56
seb128desrt, that's only happening for ppas, not on the archive builder (or it's not constant and ppas got unlucky)15:57
desrtseb128: there was a kernel bug at one point that was causing this15:57
seb128desrt, right, I just wanted to make sure you saw it15:57
desrtmaybe the ppa builders still have that old kernel?15:57
seb128could be15:57
seb128I've the feeling we had the same conversation like a year ago15:57
Laneyyou can see the kernel version at the top15:57
Laneylooks kinda old15:57
seb128that's a recurring topic :p15:58
desrt2.6?  WHAT YEAR IS IT?!?15:58
desrtjanuary 2008.  christ.15:58
=== Ursinha-afk is now known as Ursinha
Laneythey're probably running hardy or something15:59
seb128lol15:59
desrtya.... i hope you don't mind if i don't look further into this one :p15:59
LaneyBuildd toolchain package versions: launchpad-buildd_119~0.IS.08.04 python-lpbuildd_119~0.IS.08.04 bzr-builder_0.7.2+bzr156-0ubuntu1~1.IS.8.04 bzr_2.4.0-0ubuntu2~11.IS.8.04.15:59
Laneyyeah ...15:59
seb128desrt, ;-)15:59
Laneysee those version numbers encoded there15:59
desrtpretty sure this was before the kernel bug got fixed15:59
desrtalthough...15:59
desrtthat kernel may be old enough that it lacks eventfd15:59
seb128desrt, at least I can discard that as being an issue now15:59
desrtin which case this could be a legit bug with our pipe fallback16:00
desrtnope.  that kernel had eventfd.16:01
desrtoh.  curious.16:01
desrtwe create the eventfd with EFD_CLOEXEC | EFD_NONBLOCK16:02
desrtwhich didn't exist until 2.6.2716:02
desrtfailing that, we assume eventfd is not supported and fallback to pipes16:02
xclaesseA regression in 13.10 that's getting on my nerves is that every time I suspend when the laptop is docked, if I resume when undocked the screen stay black and the only thing I can do is long-press the shutdown button16:02
desrtso it could be that we're seeing a failure of the pipe fallback here (and we don't see it on newer kernels because that codepath doesn't get hit)16:02
desrtthis would impact non-linux systems too, so it may be worth looking into16:03
=== shengyao is now known as shengyao_afk
desrtthe weird thing though, is that we have a separate testcase to exercise this fallback, and it doesn't seem to be failing....16:03
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
desrt...but even in the fallback, we use pipe2()... which is also unsupported on that kernel version...16:07
larsuseb128: GtkInfoBar. This is a hack though. https://code.launchpad.net/~larsu/ubuntu-themes/dont-set-all-bgs/+merge/19723416:10
larsubut it's a better hack than the one we had before16:10
larsuochosi: ^^16:10
larsuand like I said, it might be breaking other stuff as well16:11
seb128larsu, looks fine to me, can you nudge Cimi about it?16:11
seb128I don't know those css stuff enough to approve it16:11
larsuhe doesn't seem to either. I'm mostly looking for people to try it out16:11
seb128ok, I can do that16:12
ochosilarsu: thanks, will take a look at that16:12
larsuochosi: it's not the right fix and you'll still need a patch, but at least it's not as intrusive16:13
desrtseb128: the fallback fallback case is running well here.  i blame the old kernel.16:13
seb128larsu, seems to work for me (at least for infobars), I'm going to keep running it and watching for glitches16:13
desrtand i don't really feel like reproducing my thinking from a year ago16:13
seb128desrt, ok, great, don't bother more about it then16:13
seb128desrt, thanks for checking16:13
larsuochosi: and I'd appreciate some more testing. It might be that it still breaks some stuff, I've only been running it for half a day16:13
seb128desrt, yeah, I forgot we had that discussion before, I wouldn't have pinged you otherwise16:14
ochosilarsu: i'll see what i can do!16:15
ochosilarsu: i can't but wonder why this problem can't be tackled in the overlay-scrollbars directly so themes wouldn't have to jump through extra hoops16:15
seb128ok, dropping offline for another 15 minutes, brb16:17
larsuochosi: ya, we tried. overlay-scrollbars installs a window filter function which (I think) forces the parent window to become native16:17
larsuwhich breaks bg rendering16:17
larsuit would need a refactor of o-s to make this work properly16:17
larsuand since we don't know if or how long we'll keep them, we don't want to spend too much time on it right now16:18
ochosimaybe gtk will get something like that natively at some point...16:18
larsuI highly doubt it16:18
ochosiyeah, i understand, the focus is elsewhere at the moment16:18
Laneyglib autopkgtest failed on i38616:27
Laneyhttps://jenkins.qa.ubuntu.com/view/Trusty/view/AutoPkgTest/job/trusty-adt-glib2.0/21/ARCH=i386,label=adt/consoleFull16:27
LaneyI can't find the failure16:28
Laneytoo much output16:28
Laneyah got it16:28
Laneyno, not got it16:28
Laneyyes, yes I have16:29
LaneyGLib:ERROR:/build/buildd/glib2.0-2.39.1/./tests/testglib.c:590:timer_tests: assertion failed (elapsed <= g_timer_elapsed (timer, NULL)): (5e-06 <= 5e-06) Test glib/testglib.test failed: Child process killed by signal 616:29
desrtcute....16:33
desrti love floating point16:33
seb128xclaesse, I've that, but it's only visual, e.g if I type my password I'm back to my desktop16:37
xclaesseseb128, hm, didn't try typing the password16:38
seb128xclaesse, you should ;-)16:38
seb128but agreed it's a confusing/annoying be16:38
seb128bug16:38
xclaesseI tried ctr-alt-f1 then ctr-alt-delete to reboot but it doesn't work16:38
seb128weird, maybe it's another issue then16:39
seb128going to a vt works for me16:39
seb128do you get any apport report about kernel oops or video gpu error?16:40
xclaesseI'll let you know if typing the pwd works next time it happens16:40
xclaesseseb128, should they appear in /var/crash? nothing related to that there16:40
seb128righ16:40
seb128right16:41
seb128you can also check dmesg/syslog/Xorg.0.log.old I guess16:41
seb128but try first if typing your password works16:41
seb128in which case that's the same issue I'm seeing here16:41
seb128it's a bit weird though, gnome-screensaver didn't change for ages16:41
seb128and I didn't remember it doing that before16:41
seb128could be a compiz/unity issue of course...16:42
seb128bregma, ^ do you know?16:42
xclaesseseb128, I don't know if it's the same issue, but a collegue has the same problem on F2016:42
xclaessesame symptoms on the same hw16:43
seb128weird16:43
seb128maybe a video driver issue...16:43
xclaesseit's even worse on f20 because it does that when resuming on the dock16:43
xclaessehere it fails only when resuming outside of the dock16:43
seb128your description sounds like my issue16:44
seb128I'm just surprised that going to a vt didn't work16:44
xclaessedunno if f20 has the same kernel/intel drv than 13.1016:44
ochosilarsu: hm, i applied that patch to our xubuntu theme, but it seems i still get black areas in gedit's tab-bar. what other fixes are you expecting?16:45
=== Ursinha is now known as Ursinha-afk
larsuochosi: do you have the new overlay-scrollbars?16:46
ochosilarsu: hm, no, i'm still on saucy here16:46
larsuhttps://code.launchpad.net/~larsu/overlay-scrollbar/fix-for-3.1016:46
larsuochosi: I didn't test that, sorry16:46
ochosilarsu: i assume this isn't packaged anywhere?16:47
larsuochosi: no, but it should be, soon16:47
ochosiok, would be nice if there could be some update for saucy too (or at least a PPA)16:48
xclaesseseb128, just tested, typing pwd doesn't work16:49
seb128xclaesse, :-(16:49
seb128xclaesse, can you ssh to the machine when that happens? or is it down/locked/frozen?16:50
xclaessesshing it is what I'm trying now, need to find another machine :p16:50
* xclaesse is alone at office atm16:51
seb128cyphermox, hey, can we get indicator-power and overlay-scrollbars landing in trusty?16:51
cyphermoxseb128: sure16:52
seb128indicator-power is needed to fix an issue with the new glib (which is already in trusty) (changing the preferences doesn't work with it)16:52
cyphermoxindicator-power is also used by touch though, so we need to take the time to test it16:52
seb128the scrollbar is to prepare gtk 3.1016:52
Laneyoh dear16:52
Laneyjust spilt a bit of tea inside my desktop16:52
cyphermoxoh16:52
seb128Laney, turn it off and clean I guess16:52
xclaesse"please insert a bootable floppy" --> seriously, latest lenovo laptop's bios are still asking for floppy when the disk is not bootable oO16:53
Laneyyeah, checking if any actually went inside16:53
ochosilarsu: one more question, i thought 14.04 will come with gtk3.8 not 3.10?16:54
larsuochosi: looks like we're getting 3.1016:54
ochosioh16:55
ochosii wasn't aware of that16:55
larsuyou don't sound very happy about those news ...16:56
desrt...maybe 3.12 :)17:01
* desrt coughs17:01
seb128desrt, yeah, let's see about that17:02
seb128desrt, I'm sure larsu is looking forward spending another 10 days fixing some issues in unity due to a GTK update ;-)17:02
desrtseb128: after the bumpy ride we've been having with glib and gtk 3.10 i'm surprised you still want to try 3.12 gtk at all :)17:02
desrtbut i guess it is work that we'd have to do anyway17:03
seb128desrt, I should have written "rrrrrrrrrrrrrrrrright, let's see about that"17:03
LaneyI always knew seb128 was a crack pu$ha17:03
seb128;-)17:03
larsudesrt: and it seems to be a slow gtk cycle... compared to the last one17:03
seb128larsu, LOL17:04
larsubut if I have to touch scrollbars again, I'm out17:04
seb128larsu, you want to sign for this one?17:04
* larsu runs far, far away17:04
seb128see17:04
larsu;)17:04
desrtglib is the fun one this cycle (after a slow last cycle)17:04
seb128it's going to take us a good cycle to recover from all the fun we are having17:04
seb128desrt, yeah, making rhythmbox and n-m not run, you are having a good start17:04
desrtseb128: i'm sure we'll get a pair of fixes there on monday or tuesday17:05
larsuyou broke rhythmbox?!17:05
desrtlarsu: soup, more specifically17:05
larsuhow so?17:05
desrtlarsu: soup has a gobject constructor that finalizes the constructed object, during construction17:05
larsuwhat does rhythmbox use soup on startup for?17:05
seb128larsu, but putting a new construction restriction with a g_error, without checking if anyone was doing the stupid things out there17:05
larsudesrt: lol17:05
desrtlarsu: soup-service... presumably media sharing17:05
seb128what desrt said17:06
larsuwe have that enabled by default?17:06
seb128the bt is in the daap code17:06
seb128well, I do on my laptop17:06
larsuah, okay17:06
seb128I'm sure others do17:06
larsuI think it's not enabled by default, but many people have it17:06
desrtamusingly, the person who added the assert to glib is danw17:06
desrtand the person who wrote that constructor in soup is.... danw17:06
larsuwho also maintains soup?17:06
larsuya :D17:06
desrtit gets better17:06
desrtthe last person to touch the constructor code in n-m is...... danw17:07
larsuhow do you write a constructor that finalizes the object?17:07
larsuha, brilliant17:07
mhr3seb128, i know how we can get most of the bugs fixed - we just need to ensure they happen on your computer ;)17:07
ochosilarsu: nah, i've prepared our themes already for 3.10. usually gtk3 updates suck because they break theming everywhere (otherwise, what would themers do all day?)17:07
desrtthink GInitable17:07
ochosilarsu: i was just really sure 14.04 was going with 3.8 for some reason17:07
desrtimplemented via g_object_new() returning NULL17:07
larsuochosi: ya, that was the original play I think. Or at least, it was discussed17:08
larsudesrt: ugh17:08
seb128ochosi, you probably remember https://lists.ubuntu.com/archives/ubuntu-desktop/2013-October/004311.html17:08
desrtlarsu: ya...17:08
desrtlarsu: thing is, it's never been valid... it always leaked memory17:08
seb128ochosi, read https://lists.ubuntu.com/archives/ubuntu-desktop/2013-November/004343.html if you are interested in the current plans17:08
desrtbecause we keep a linked-list of in-construction objects in gobject.c17:08
Laneyerr17:09
Laneynow the tests failed on amd6417:09
desrtand if you return NULL from your custom constructor, it never gets removed from that list17:09
Laneyhttp://d-jenkins.ubuntu-ci:8080/view/Trusty/view/AutoPkgTest/job/trusty-adt-glib2.0/22/ARCH=amd64,label=adt/consoleFull17:09
ochosithanks seb12817:09
desrtLaney: is there a way for us normal humans to view these webpages?17:09
Laneyoops I meant to link to the public one17:09
Laneyhttp://d-jenkins.ubuntu-ci:8080/view/Trusty/view/AutoPkgTest/job/trusty-adt-glib2.0/22/ARCH=amd64,label=adt/consoleFull17:10
LaneyWHAT17:10
larsuthat tea is messing with your clipboard!17:10
desrtjust one of those days17:10
Laneyhttps://jenkins.qa.ubuntu.com/view/Trusty/view/AutoPkgTest/job/trusty-adt-glib2.0/22/ARCH=amd64,label=adt/consoleFull17:10
Laneysuspicious one to fail ...17:11
ochosiseb128: nice to read that you're also taking into account the CSD problem for non-gnome desktops. xubuntu folks will happy to hear17:11
seb128ochosi, ;-)17:12
desrtGLib-GObject:ERROR:/build/buildd/glib2.0-2.39.1/./gobject/tests/object.c:134:test_object_constructor_infanticide: stderr of child process (/object/constructor/infanticide/subprocess [17405]) failed to match: *finalized while still in-construction*17:13
desrtseb128: looks like you forgot to revert the testcase that tested those comits you reverted :)17:13
seb128desrt, I didn't17:13
seb128or was that a different commit ?17:13
seb128it passed fine on i38617:14
seb128wth?17:14
Laneywhy doesn't it fail all the time?17:14
Laneyit passed on amd64 previously too17:14
seb128desrt, that's the revert, https://launchpadlibrarian.net/157885850/glib2.0_2.39.1-0ubuntu1_2.39.1-0ubuntu2.diff.gz17:14
desrti wonder if we're running the old set of installed tests against the new glib?17:14
Laneyah, could be17:15
desrtand by old i mean new and by new i mean reverted17:15
xclaesseseb128, just tested, I cannot even ping my laptop after resume17:15
seb128xclaesse, seems like a kernel issue then :/17:15
xclaesseseb128, I see /var/crash/susres.2013-11-29_12:12:51.860919.crash17:15
xclaessedunno if it's related17:15
seb128xclaesse, do you have an oops in dmesg/syslog?17:15
seb128xclaesse, seems likely yes17:15
Laney+ spec=/tmp/adt-run.yxo4nf/dsc0/glib2.0_2.39.1-0ubuntu1.dsc17:15
Laneythat's the previous one17:16
Laneyargh17:16
seb128how is that happening?17:16
=== Ursinha-afk is now known as Ursinha
LaneyJENKIIIIIIIINNNNNNNNSSSSSSSSSSSSS17:16
desrtin general i think it does make sense to run the old set of installed tests against the new library17:17
desrtcan catch ABI break type issues17:17
desrtbut in this case it's obviously a problem since reverting that patch intentionally broke ABI...17:17
LaneyW: Failed to fetch bzip2:/var/lib/apt/lists/partial/ftpmaster.internal_ubuntu_dists_trusty-proposed_main_source_Sources Hash Sum mismatch17:18
Laney E: Some index files failed to download. They have been ignored, or old ones used instead.17:19
Laneyactually W: Failed to fetch bzip2:/var/lib/apt/lists/partial/ftpmaster.internal_ubuntu_dists_trusty_universe_binary-amd64_Packages Hash Sum mismatch is more relevant17:19
xclaesseseb128, I don't see anything special in dmesg, but I think it only contains the msg from last boot17:19
desrtLaney: you're back in your world now and out of mine :)17:19
* Laney stabs17:19
larsuseb128: bad news about the message dialog problem (which I've finally gotten around to)17:19
* desrt goes back to happily hacking on dconf17:19
larsuseb128: my fix doesn't work for many apps (including gedit), because they use custom dialogs17:19
Laneyhmm, I don't know if I believe that though17:21
xclaesseseb128, another really annoying issue is windows changing its virtual desktop every time there is a screen change17:22
Laneyjibel: you around?17:22
seb128larsu, *great*17:22
* seb128 head->desk17:22
seb128loving GTK17:22
larsuseb128: I had a feeling this is how you'd respond17:22
larsuso now, we'd really have to fix apps themselves...17:23
seb128larsu, that's fine, it's almost beer'o'clock17:23
larsuright. I don't plan on doing it now :)17:23
attenteseb128, is ctrl+alt+t working for you under a russian layout?17:23
seb128attente, no, and it was no in 13.04 stock issue either when I tried iirc17:23
attenteseb128, i don't know why, but it seems to be working in 14.0417:24
Laneyjibel: if/when you are - could you take a look at https://jenkins.qa.ubuntu.com/view/Trusty/view/AutoPkgTest/job/trusty-adt-glib2.0/22/ARCH=amd64,label=adt/consoleFull please?17:24
attenteseb128, and there's also no conflict with ctrl+shift+t in gnome-terminal either17:24
seb128attente, that doesn't make sense, we didn't change things there17:24
Laneyjibel: The installed-tests test installs an old libglib2.0-tests which makes the test fail17:25
attenteseb128, it's really weird... everything's working :/17:25
Laneybut the 'build' test got the current correct glib17:25
seb128attente, can you try with another keybinding? ctrl-alt-t is handled by compiz as well17:25
seb128it's a special case17:25
seb128like ctrl-alt-l might be a better one17:25
seb128or assign ctrl-alt-c to gnome-calculator17:25
attenteah. ok. things make sense again17:26
attenteit doesn't really explain why ctrl-shift-t would work though...17:26
attenteoh right... never mind17:26
attentethe key binding wasn't set properly17:26
xclaesseseb128, larsu: why can't you just revert the commit that introduce the regression? or maybe just in ubuntu package...17:27
xclaesse(about the gtk issue, I've read backlog on #gtk+)17:27
seb128xclaesse, I'm thinking about it17:27
seb128it would have been nice to fix it properly/upstream though17:27
xclaesseif the fix only fix something in anaconda, and make everything else regress, that really seems a good candidate for revert17:28
xclaesseat least for ubuntu which is not impacted at all17:28
seb128right17:28
desrtseb128, Laney, xclaesse: is there some kind of 'best practice' for upstreams that want accounts created for their software?17:28
seb128well, the previous behaviour was sort of a hack17:28
seb128the thing is that half the universe rely on that wrapping at 64017:28
desrtor is it more or less something that the packager is supposed to recognise and take care of?17:28
seb128desrt, "accounts"?17:28
desrtya... like gdm for example17:29
desrtit needs a gdm user/group created17:29
seb128oh17:29
seb128you mean system accounts17:29
desrtya17:29
seb128like gdm running under its own user?17:29
seb128that's something done in the postinst etc17:29
desrtyes.  precisely.17:29
seb128so no, just file a bug request is in the distro bug tracker17:30
seb128or put a note in NEWS and hope the packager see it17:30
desrtis there a best practice for the upstream project to indicate that it wants distro packages to create such an account?17:30
desrthrmph.17:30
desrtwalters: how does gnome-continuous deal with this, if you don't mind?17:30
desrtseb128: i guess you can't fail the ./configure because the account doesn't exist... because the account would only be created in the postinst....17:31
* desrt wonders about 'make install' rules that need to install files with correct ownership17:32
seb128desrt, right17:32
desrttricky business, this is :)17:33
seb128desrt, same issue, make install is often ran as anuser17:33
seb128desrt, yeah, we change permissions in the postinst when needed as well17:33
desrtseems like there ought to be a better way....17:34
* desrt wants /var/lib/dconf for the phone stuff, but wants it owned by dconf:dconf for the future when system dbs are written there17:34
desrtif the way you normally deal with that is chown from a postinst script anyway then maybe i should just wait until when the account is actually needed, and then the (existing) dir can be chowned at that point?17:36
seb128right17:36
* desrt takes a look at some of the prior art here17:36
waltersdesrt: i have a hardcoded list https://git.gnome.org/browse/gnome-ostree-integration/tree/src/lib-passwd17:41
walterswhich works fine because the set of software is also a hardcoded list17:42
Laneyseb128: lp:~laney/ubuntu-system-settings/graph-dashed-axes17:48
Laneynot MPing it yet because I think I need to adjust the drawing of the actual line to not go over the axes17:49
Laneybut you could branch off that to do your labels stuff17:49
* Laney is off, have a good weekend all17:52
desrtwalters: might break when i add an account for dconf...18:12
desrtseb128: so....18:15
desrtif [ -d /var/lib/gdm ]; then chown gdm:gdm /var/lib/gdm chown -R gdm:gdm /var/lib/gdm/.gconf* chmod 0750 /var/lib/gdm18:15
desrtfi18:15
desrtie: we only chown/chmod in the case that we create the directory for the first time18:16
desrtwhich i think is a pretty sane thing to do18:16
desrtbecause you don't want to override the sysadmin's preference if he set it to something else and you're just upgrading the existing package18:16
desrtseb128: sort of suggests that i should ask you to make the account even if i'm not using it yet, so that you can create the directory with the correct perms18:17
seb128desrt, right, we can also do trick like "if upgrade from version <; then chown"18:17
desrtthat's possible?18:17
seb128yes18:18
desrtneat18:18
seb128that's how we do migrations18:18
desrtmaybe i shouldn't worry, then18:18
seb128but the snippet you copied would do it every time no?18:18
seb128if [ directory exists ] ...18:18
desrtoh uh ya.. i read that backwards18:18
desrtwhat a strange rule18:18
seb128well, if postinsts hit an error your upgrade/install fails18:19
seb128so you need to check that something exists before trying to chown it18:19
desrti'm guessing gdm creates that dir in 'make install' and it's in an .install file18:19
seb128otherwise if somebody delete the dir it screws the package installation18:19
desrtor maybe adduser makes it18:19
desrtokay.  so if we don't care about this sort of stuff then i _really_ don't care :)18:20
seb128well, better we safe than sorry18:20
* desrt proceeds naively18:20
seb128;-)18:20
seb128we->be18:20
seb128but yeah, don't worry about that18:20
seb128Laney, ok, I'm not going to work on that before next week but good to know ;-)18:21
desrtLaney: how's the u-s-d/u-c-c stuff going?18:24
seb128desrt, robert_ancell is working on it, he has a vcs for both, he sent me an email this morning on u-c-c being working18:27
* seb128 needs to test that next week18:27
desrtthat's going to be fun.... how do you think you manage the transition there?18:27
seb128what transition?18:27
desrtstraight replaces/conflits stuff?18:27
seb128robert_ancell is doing it in a way where they don't conflict18:28
desrtseb128: like... what if we see that gnome-shell is installed and unity is not, at the time of dist-upgrade?18:28
seb128which I think is crazy18:28
desrtseb128: i mean how do we decide which of (new g-c-c)/(u-c-c) to install at time of upgrade?18:28
seb128unity is going to depends on u-c-c18:28
desrtthat'll do it18:28
seb128which is going to bring it for sure18:28
seb128how we clean out g-c-c is to be seen18:28
seb128we have hooks in update-manager18:28
seb128worth thing it's a leftover on disk18:29
desrtright.  makes sense as a dist-upgrade task18:29
seb128it shouldn't show in unity, it's going to be OnlyShowIn=GNOME18:29
desrtif someone does it via apt manually then they get extra packages.  no big deal.18:29
seb128right18:29
desrtseb128: i'd rather suggest NotShowIn=Unity18:30
seb128yeah, it's friday evening, I was not thinking details18:30
desrtcool.   glad to hear it's coming along nicely, in any case18:30
seb128just that we are going to tweak visibilities18:30
seb128yeah18:30
seb128we "just" need to solve the libgnome-desktop issue18:30
desrtbeing able to keep up to date with g-s-d/g-c-c will be really nice... a good reason for having latest gtk18:30
seb128I'm pondering just having a code copy of the functions of the old version in u-c-c18:31
desrtwhat's the problem here?18:31
larsudesrt just won't stop trying :)18:31
seb128they changed the xrandr code18:31
larsu"a good reason for having latest gtk"18:31
desrtoh.  right.18:31
seb128rather than talking xrandr18:31
seb128it's talking gnome-shell dbus18:31
seb128ideally we would have compatible dbus interfaces in unity18:31
seb128but I don't think that's going to happen this cycle18:31
desrthard to argue about this being unfriendly to non-gnome desktops when it's in a library called libgnome-desktop :p18:31
seb128hehe18:32
seb128well, anyway, that's fine, just copy a couple of .c in u-c-c18:32
desrtwe'll eventually need such an interface no matter what18:32
seb128until we get a similar interface in unity18:32
desrtsince xrandr is going away18:32
seb128right18:32
desrtmight be a good topic for the desktop summit18:32
seb128I just don't want to block anyone on that to happen18:32
seb128oh, desktop summit18:32
seb128did I mention that I want a common dbus interface for session stuff18:33
desrtinhibit and such?18:33
seb128like close the session18:33
larsuhm, interesting18:33
seb128or "show me the logout dialog"18:33
desrtya.  we hope to talk about this18:33
desrttrying to get bastien to come this year....18:33
seb128or rather the "reboot" dialog18:33
larsuthat could make unity stop exposing an interface in the org.gnome namespace18:33
seb128right18:33
desrtthis stuff ought to be in org.freedesktop to the greatest extent possible18:33
seb128it could also make e.g software-properties stop calling gnome-session-quit18:34
desrteven the resolution stuff.... i can't imagine mir would be that different from wayland18:34
desrtso we may as well just copy gnome designs there18:34
seb128right18:34
desrtand rename to org.freedesktop...18:34
seb128otherwise it's going to be the same as the session stuff18:34
larsulast time we talked about this, there was some controversy around how to end a session18:34
larsulike, should apps be able to inhibit18:34
desrtlarsu: KILLALL18:34
seb128we are going to claim being org.gnome.shell or something18:34
seb128and some people are going to be angry at us18:34
desrtlarsu: there is no controvery.  there is only one right way :)18:34
larsudesrt: obviously you were on the one side of the controversy18:35
desrtseb128: ya... that would just be ridiculous18:35
desrtlarsu: apps can register blocks to logout _with user visible reason only_18:35
desrtthis will cause dialogs to be shown when the user picks 'logout'18:35
larsuhm? How would that look in practice?18:35
larsuah the shell shows a dialog with a string that the apps giveit?18:36
desrtbeyond that, if the logout is going to proceed _nothing_ can stop it18:36
desrtlarsu: yes.  that's why we have inhibit 'reasons'18:36
larsuew have those now?18:36
larsu*we18:36
desrtbut a very simple desktop environment could also implement it by poking each of those apps asking them to resolve the issue18:36
desrt gtk_application_inhibit             (GtkApplication *application, GtkWindow *window, GtkApplicationInhibitFlags flags, const gchar *reason);18:37
larsuright - but does gnome shell support this?18:37
desrtyes18:37
desrtwhen you logout the dialog shows something like:18:37
larsumind = blown18:37
desrtCan't log out because:18:37
desrt - gedit18:37
desrt   2 unsaved documents18:37
desrt( Do it anyway )  ( Cancel )18:38
larsudoes unity do this as well?18:38
desrtthis is the really correct way18:38
* larsu tries18:38
desrtbut it's also acceptable, i think, to have the app in question pop up the dialog for itself18:38
desrtparticularly if there is only one18:38
larsuno, it doesn't :-/18:39
larsuya, that's what unity does18:39
desrtwhat is absolutely unacceptable, however, is for apps to have "tell me when you're about to logout and i may or may not stop the process and/or may do some cleanup..."18:39
larsuapparently only for one app at a time, thought18:39
larsu*though18:39
larsudesrt: ya, that's ridiculous18:39
seb128unity doesn't do anything18:39
seb128gnome-session does that18:39
desrtlarsu: it's currently possible.  it needs to go away.18:39
larsuseb128: right, I meant unity-as-the-desktop18:39
desrtcleanup needs to go in SIGTERM handlers18:40
desrtand if someone wants to _stop_ the logout, they need to know ahead of time that they will do so, and they must give a user-visible reason18:40
larsuand most of the time, no cleanup should be necessary18:40
desrtwell18:40
desrtapps may want to sync their state18:40
desrtsomething like bijiben for example18:40
desrtanyway.... this really ought to be on the desktop summit agenda18:41
desrtit's shameful that this isn't a freedesktop-level spec yet18:41
larsulet's write one :)18:41
seb128ok, calling it a week, have a nice w.e everyone!18:56
=== m_conley_away is now known as m_conley
=== m_conley is now known as m_conley_away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!