[00:02] DBO: around? [00:09] jbicha: hey [00:13] desrt: good evening [00:13] jbicha: what timezone do you operate on? [00:14] desrt: same as you, would you believe it was in the upper 80s here today? [00:14] crazy weather [00:14] where do you live? [00:14] same here [00:14] nuts [00:15] just south of kenvandine in Columbia, South Carolina [00:15] * desrt knows by the use of "upper 80s" that it's not so close, yet not so far [00:15] ah. cool. [00:15] lol [00:15] so this just happened: [00:15] Get:13 http://ca.archive.ubuntu.com/ubuntu/ precise/universe gnome-shell amd64 3.3.90-0ubuntu1 [331 kB] [00:18] that's good, right? [00:20] jbicha: yes. quite :) [00:20] no more gnome3 ppa required [00:21] so there does appear to be a bit of an issue... [00:22] dunno if you know about it but it seems like there is a conflict with tweak-tool now [00:22] desrt: yes, the new gnome-tweak-tool just finished building [00:22] ah. great. [00:23] ppa-purge was having a heart attack [00:23] I should have done a whole bunch more breaks as some things are actually still broken but apt isn't telling you that [00:24] i really think aptitude is not so clever sometimes when removing a single package would resolve a conflict and instead it proposes to remove 12 others without solving it [00:24] if you want thing to work, keep the gnome3 ppa until tomorrow [00:24] oh =) [00:24] anything in particular? [00:24] cheese won't work, for one [00:24] ah. i don't use that. [00:25] the libcogl5 stuff at http://people.canonical.com/~ubuntu-archive/nbs.html [00:25] so what's the deal with the soname bump? [00:25] i guess it went through afterall? [00:26] the map plugin from eog-plugins won't work until libchamplain gets updated [00:27] desrt: they did 4 soname bumps this cycle and were working on one more this week so I sent them an email & they agreed that it would be a good idea to stop doing that until next cycle :) [00:27] gotcha [00:27] well, in any case, good work [00:28] looks like you really hauled ass today :) [00:28] maybe i can convince tiffany to go back to ubuntu now :p [00:29] what's she use? [00:29] fedora mostly [00:29] i think she's planning to try mint when the next version is done, though [00:29] the dev version of fedora is a painful way to go... [00:30] all of her classmates/students have been switching and she's wondering what the fuss is about :p [00:30] ya. i don't have any great love of fedora as a distro :) [00:37] desrt: did gnome shell 3.2 on ubuntu work on your arm box? [00:37] it's headless [00:38] ah, well clutter's a bit broken on arm now [00:38] anything to do what that earlier define? :) [00:41] jasoncwarner_: good morning [00:42] desrt: I think one is related to that https://launchpadlibrarian.net/96831041/buildlog_ubuntu-precise-armhf.clutter-gst_1.5.4-0ubuntu1_FAILEDTOBUILD.txt.gz [01:26] desrt, you around? Do you remember updating GTK+ 3's 043_ubuntu_menu_proxy.patch for bug 915241? [01:26] Launchpad bug 915241 in gtk+3.0 "Update menubar patches" [Low,Fix released] https://launchpad.net/bugs/915241 [01:27] i remember helping cimi to update the patch during the rally [01:27] why do you ask? [01:28] desrt, the property "ubuntu-local" got dropped, and I don't see anyway for apps to opt-out on a per-menubar manner now. (as glade needs to for embedded menubars) [01:29] curious. [01:29] i recall asking around if anyone was using that property for anything like that [01:29] and i got told no :p [01:29] desrt, if it was just an oversight, I can add it back. But was curious if it was intentional and what the replacement would be [01:29] desrt, :) [01:30] desrt, I think freeciv also uses it [01:30] specifically, i remember asking ted if it was used to supress the panel's menu from itself getting sucked up [01:30] mterry: it removal was intentional [01:30] but clearly it should be added back [01:30] or at least some other mechanism [01:31] desrt, OK. I can't think of a reason not to just add it back (vs another mechanism) [01:31] better check with cimi, of course [01:31] it's his baby [01:31] but i think you're right [01:31] desrt, ok, will do [01:31] GTK+ 2 still has the property === webjadmin_ is now known as JackyAlcine [01:35] night all! [01:35] ta === chaoticuk_ is now known as chaoticuk === chaoticuk_ is now known as chaoticuk [02:51] Bah. No Robert? [02:52] Hm. Now that I say that out loud, I suspect that perhaps a baby might be involved in that lack... [02:52] RAOF: oi :P [02:52] RAOF: and yes. [02:54] lifeless: Sorry, other Robert :) [02:54] I know :). Hey, who is our n-m guru ? [02:56] cyphermox, IIRC. [02:56] Yup. [02:57] Although for some reason I always associate Sweetshark with n-m maintainership. I have no idea why. [03:13] RAOF: Mathieu Trudel-Lapierre [03:13] seems like he does the vast majority of uploads [03:13] desrt: Yeah, he's the guy. [03:19] RAOF: so have you really never heard of this purple snow issue i was mentioning earlier? [03:19] i wonder if you have some time to work through it with me... [03:20] ie: i could try to reproduce it and we could debug [03:20] Can you help me swap in that context? [03:21] i have a t420 [03:21] most of the time it sits in a dock, connected via DVI to two 24 inch panels [03:21] i undock it when i leave my house [03:21] sometimes undocking it causes a purple snow effect on the laptop panel [03:22] i have it right now, for example [03:22] Ok. [03:23] Let me build a test unity-greeter to see if I've discovered the corruption bug, and then I can do some debugging. [03:26] * RAOF *loves* how trouble-free "bzr pull" is in an unclean tree. [03:28] RAOF: http://people.gnome.org/~ryanl/MVI_0037.AVI [03:29] a couple of notes: the problem seems to happen most (exclusively?) when i boot the laptop in the dock with the lid closed (ie: laptop panel not in use during startup) [03:29] docking the laptop again allows me to use the external monitors normally [03:30] the issue only ever impacts the internal screen [03:31] notice also that exactly half of the screen (left side) is stretched out over the entire display [03:31] it's also like this if i switch to the console [03:33] desrt: Ok. So, throwing "drm.debug=0x4" on your kernel command line is likely to be the sweetspot for debugging vs verbosity in dmesg. [03:33] It would seem likely to be a modesetting issue. [03:34] k. lemme reboot and try to reproduce [03:35] * desrt gets on irc from another machine [03:39] arf. too much debugging output -> causes the buffer to ring [03:39] * desrt composes multiple files into one [03:40] RAOF: okay.. here you go http://fpaste.org/yQ59/ [03:40] see the markers in the file for the point at which i docked/undocked [03:41] the general scenario is that i booted the laptop in the dock with the lid closed, logged in, undocked, redocked === m_conley is now known as m_conley_away [03:47] Huh. It seems to be flipping fbc on and off. [03:48] desrt: I don't suppose that booting with i915.i915_enable_fbc=0 fixes it for you? [03:48] allow me to try that :) [03:48] That's what looks most suspicious. [03:50] does not fix it [03:50] Boo. [03:51] BOOT_IMAGE=/boot/vmlinuz-3.2.0-18-generic root=UUID=7796de19-491a-4714-882f-0a758240deb9 ro quiet splash vt.handoff=7 i915.i915_enable_fbc=0 [03:51] that's right? [03:52] you saw the video? [03:53] Yes. [03:53] desrt: http://lists.freedesktop.org/archives/intel-gfx/2012-March/015473.html [03:54] hello hello [03:54] well isn't that a timely post? [03:55] RAOF: do you do drm or am i better speaking with some kernel folks at this point? [03:55] Sarvatt: Nice catch, thanks. [03:55] desrt: Talking to kernel guys wouldn't hurt. If that patch gets applied we can ping kernel guys to get it integrated in precise. [03:56] looks like ajax is on the case [03:56] that gives me confidence [03:56] nah that patch is very much a hack [03:56] and yeah exactly :) [03:57] Let's see if this unity-greeter still lets me log in... [03:58] preferred mode for the panel and assume it's dual-link LVDS if the pixel [03:58] clock is >112MHz, since that's the crossover frequency. [03:58] RAOF: any way i can find out what this is on my panel in order to confirm ajax's idea? [03:58] resolution is 1600x900... [03:59] 1600*900*60 gives 86.4 mhz, but i don't know what the proper calculation for adding in the blanking intervals is [04:02] desrt: Yeah, it's in the modeline. Let me nab it... [04:03] [ 9.395491] [drm:drm_mode_debug_printmodeline], Modeline 72:"1600x900" 60 110000 1600 1664 1706 2010 900 903 906 912 0x48 0xa seems to be your 1600x900 modeline; that suggests a 110MHz clock. [04:03] * RAOF thinks he's reading that right. [04:03] so ajax says the crossover is 112 [04:03] i'm below that [04:03] so i'd guess it should be single channel by his measure [04:03] which seems wrong... [04:04] There might be necessary signalling overhead? [04:04] Bah. Stupid X. [04:05] If I wanted you to keep unity-greeter's window open, I'd have set the mode to RetainPermannent. Oh, wait... [04:05] RAOF: he specifically says "pixel clock" [04:06] "All display modes that use a pixel clock below 165 MHz, and have at most 24 bits per pixel, are required to use single-link mode. All modes that require more than 24 bits per pixel, and/or 165 MHz pixel clock frequency must use dual-link mode." says wikipedia, though [04:08] you're looking at dvi dual link not lvds [04:09] oh. duh. [04:27] Sarvatt, RAOF; thanks for the help. -> bed [06:21] good morning [06:36] Good morning [06:38] Good morning. [06:38] Or, rather, grrrrrrrrrrr! morning. [06:39] Who here has a clue as to what vala does? [06:40] FSVO, yes. [06:42] lifeless: Given http://paste.ubuntu.com/884386/ as a code fragment, why is display being freed *immediately* after being new'd? [06:43] (ie: we have an static class variable display, we set it, the free method is called immediately after the create method) [06:43] that seems surprising [06:43] Also, C apparently makes a rubbish IL. [06:44] Oh, no, sorry. I'm misreading the C code. [06:44] that makes more sense [06:45] Which doesn't help me explain why that fails, but the same thing with display being a local variable succeeds. [06:46] http://paste.ubuntu.com/884388/ is the full method. Remove the static class variable, stick a "var" in front of the first "display", and it works. [06:55] Oh, marvelous. A race condition? [06:58] Mmmm, yes! [06:58] Sweet accidental success! === tsimpson_ is now known as tsimpson [07:23] Oh! *That's* why unity-greeter's think kinda works! [07:26] my osd notifications turned blue?! [07:26] Someone's choking them!! :'( [07:26] BigWhale: Unity, for some reason, thinks "blue" is a representative colour for your background. [07:27] RAOF, must be my mood or something. :> [07:28] I changed desktop background ... now it's ok [08:04] ctr-alt-t does not open a terminal anymore since the update to gnome-shell 3.3.90 [08:04] bbl [08:04] and it's still the shortcut set in gnome-control-center [08:05] xclaesse: maybe mutter regression? jbicha would know more [08:05] ctr-alt-n still works to open nautilus, but IIRC that's a custom shortcut I did [08:09] RAOF: can you jumpon #ubuntu-unity please? [08:09] xclaesse: works on compiz and metacity here [08:13] yeah must be mutter/gnome-shell regression since it was working before today's update [08:32] didrocks, xclaesse, hi, this is due patching g-c-c and g-s-d to use gconf instead of gsettings [08:32] why is it using gconf? [08:32] and iirc the terminal shortcut was removed upstream [08:32] ricotz: well, that doesn't change the default value in mutter [08:32] ricotz: the default should still be ctrl + alt + t === tkamppeter_ is now known as tkamppeter [08:34] I created a custom binding that opens gnome-terminal on ctr-alt-t and it works fine :) [08:36] hey didrocks, bonjour [08:37] pitti: hey Martin! :-) [08:38] good morning everyone [08:38] hey chrisccoulson [08:38] hey chrisccoulson [08:38] hi pitti, didrocks. how are you? [08:39] I'm fine, thanks! hacked on pygobject until late last night, and I learned a lot about all that hairy marshalling [08:39] sounds fun :) [08:39] it is after a couple of hours when stuff is finally starting to work :) [08:40] chrisccoulson: well, release of unity-2d with multimonitor has some unexpected hickupps [08:40] otherwise, I'm fine :) [08:41] didrocks, oh? i guess i could help test that stuff :) [08:41] chrisccoulson: really? nice! :) [08:41] chrisccoulson: so bzr branch lp:~ubuntu-desktop/metacity/ubuntu [08:41] bzr bd it please :) [08:41] and then bzr branch lp:unity-2d [08:41] ok, doing that :) [08:41] bzr bd it as well [08:42] so you set "hide-mode to 1" in gsettings for 2d [08:42] and you should get push to reveal [08:42] (not monitor barriers yet) [08:42] ok, i'll test that [08:42] yeah, i tried unity-2d a few days ago, and i actually missed the lack of monitor barriers :) [08:43] chrisccoulson: it's a pending merge, was supposed to be merged overnight, we are checking why it wasn't (but the main driver is at the dentist :p) [08:43] hey [08:43] salut seb128 [08:43] hi seb128 [08:43] hey didrocks, chrisccoulson how are you? [08:43] seb128, yeah, not too bad thanks [08:44] seb128: I'm fine, thanks, you? [08:44] bonjour seb128 [08:44] seb128: comment vas-tu? [08:44] hey pitti, wie gehts? [08:44] do you guys have https://launchpad.net/~mozillateam/+archive/firefox-next/ in your sources btw? :) [08:44] seb128: gut, danke! having fun with pygobject [08:44] I'm good thanks ;-) still waking up but coffee is helping :p [08:45] chrisccoulson: not ATM, should we? [08:45] chrisccoulson: I thought we'd be uploading firefox betas into precise directly? [08:46] with "we" == "you", I mean :) [08:46] pitti - yeah. we're not uploading beta's to precise now because the next firefox release is after the release of precise [08:46] so i'm back to using the PPA now, which means that will be the first post-release update [08:47] on 24th april ;) [08:49] oh, a new lightdm, robert_ancell was around today? [08:49] seb128, he is probably unable to sleep now ;) [08:50] chrisccoulson, ;-) [08:50] chrisccoulson, when is your second one due btw? [08:50] seb128, the second week of july [08:50] ok [08:50] we will still see you at UDS I guess then? [08:51] seb128, yeah, i'll still be at UDS [08:51] good ;-) [08:52] * pitti runs sudo add-apt-repository ppa:mozilla-team/firefox-next [08:52] precise isn't broken enough, no fun! [08:53] even my Alt key works *grumpf* [08:53] pitti, excellent :) [08:53] chrisccoulson: hm, I get nothing new from the PPA [08:53] pitti, i've got publishing disabled until i've done an upgrade test on all of the other releases [08:53] i'll do that today though :) [08:54] pitti, we got a new clutter, that's some new crack, not enough to make desktop issues though ;-) [09:00] brb, switching to 2d [09:03] didrocks, so, i guess once the barrier stuff lands, i'll get a launcher on my second monitor too? [09:03] chrisccoulson: no, the default will be one monitor only [09:03] chrisccoulson: then, there will be an option for a launcher on each monitor [09:03] ah, ok. is that the case for 3d too? [09:10] morning chrisccoulson! Have you had the chance to look at the e-mail I sent you re: Firefox .desktop translations? [09:11] oh, sorry, not yet. been busy ;) [09:13] chrisccoulson, no worries, just wanted to make sure you got it. In short, do you think it will be possible to do an upload of the new translations for FF and TB from that wiki page by NonLangpackTranslationDeadline? [09:13] sorry, I mean .desktop file translations [09:13] dpm, probably not. i want to avoid doing any more uploads in precise if i can, as it makes the next beta -> stable merge tricky [09:14] in general, we just add things to trunk and have them flow in to the distro with the usual updates [09:14] unless it's for a really critical issue [09:15] chrisccoulson, hm :/ do you think for next cycle we could put the .desktop file in LP, so that it can be translated there and there is no manual work involved in adding new translations? [09:15] possibly ;) [09:15] chrisccoulson: it will be the case for 3d as well [09:15] ok [09:16] good morning, people from the future [09:22] hey Ursinha, how are you? [09:23] hey Ursinha [09:24] pitti, I'm fine :) will you have some time later today to take a look at the gravity report thing? I'd love some feedback about how the gravity is reflecting the importance of the bug [09:27] Ursinha: I need to run out for some 20 mins, can do later yes [09:28] pitti, no need to rush, actually it's super early over here, was just assuring I could have a slot today :) [09:38] seb128, did you say dx was doing some work on dconf? [09:39] or, more precicely, who should I assign a libdconf-qt0 bug too? [09:43] rickspencer3, dx yes [09:44] rickspencer3, it's for the unity2d team [09:44] thanks seb128 [09:46] rickspencer3, yw [09:55] wow, gnome-shell has come a long way [09:57] still missing file search, but the overall optics/animations are nice [09:58] but the windows key doesn't open Activities any more? [09:58] desrt: ^ do you know? [09:58] i like gnome-shell too :) [09:58] pitti, throw your cursor to the top left corner [09:58] yes, I recognized that [09:58] chrisccoulson, pitti: you ubuntu haters :p [09:58] but that requires using the mouse [09:58] pitti, I think super should work [09:58] seb128, i'm not a hater. i like unity too [09:58] not here [09:59] pitti, "should", I guess it's a bug [09:59] chrisccoulson, that's ok then ;-) [09:59] although, going back to unity-2d makes me realize how much i prefer the dash in 2d [09:59] unity is still a lot more keyboard and file handling friendly [09:59] chrisccoulson, just because of the iced icons? [09:59] it's nicely coloured with big, crisp icons and smooth animations :) [09:59] seb128, yeah, the icons are a big factor :) [09:59] i still don't get why we deliberately make them blurry in 3d [10:00] chrisccoulson, I hope for you they don't fix it :p [10:00] it just doesn't look right [10:00] heh [10:00] chrisccoulson, I think 3d just respects the design [10:00] so the day 2d catches up on design you will be disappointed [10:00] yeah, i'll need to find a way to revert that [10:00] :) [10:00] ;-) [10:00] pitti, shell is nice looking solid and smooth [10:00] I just hate the activity stuff [10:01] it's too much context change every time you want your launcher or focus something else or open something [10:01] like your screen keep zooming out and in [10:02] yes [10:02] I looked at it around oneiric release time the last time [10:02] I was mostly curious to see how far it got since then [10:02] the multimonitor story is still better in unity :) [10:03] the two biggest things that jumped my eye were that I have access to files (nautilus) again, and the nicely done indicators/animations [10:03] yeah, their indicators are quite nice [10:03] I'm sure they don't call it like that [10:06] hum, upgraded my netbook yesterday [10:06] (from oneiric to precise) and now, bzr can't find bzrlib [10:06] (it's in pyshared though, should be some symlink not beeing done) [10:08] Sweetshark, https://bugs.launchpad.net/ubuntu/+source/libreoffice/+bug/916291/+attachment/2874475/+files/lo-common.dpkg-70000.tgz [10:08] Launchpad bug 916291 in libreoffice "failed to upgrade from Oneiric to Precise: ERROR: Cannot determine language! - exit status 134" [Critical,Triaged] [10:08] chrisccoulson: are you interested in looking at bug 949796? [10:08] Launchpad bug 949796 in lightning-extension "lightning thunderbird extension does not follow system time zone settings" [Low,New] https://launchpad.net/bugs/949796 [10:08] chrisccoulson: if not, I'll just unassign it, as it's universe [10:08] hum, under a prompt, it works with 2.7 which is the default, maybe a bzr issue? [10:09] pitti - hmm, that's not the sort of thing i'm going to have time to look at. and if i looked at lightning, that probably wouldn't be the highest priority (with missing integration in the date/time indicator) [10:10] chrisccoulson: ack, unassigning [10:10] feel free to unassign that [10:10] thanks [10:14] * chrisccoulson fires up lucid VM [10:15] maaaan, this feels old ;) [10:15] my chroots all look the same :) [10:16] well, I do miss stuff like "apt-get download" [10:36] chrisccoulson, went on a cruisade against adobe now? ;-) [10:37] seb128, how come? [10:37] what did i do now? :) [10:37] chrisccoulson, the firefox update wants to remove adobe-* for me, I'm trying to reactivate partners :p [10:37] seb128, oh, you've still got the oneiric partner repo in your sources? [10:37] chrisccoulson, yes [10:38] yeah, that's why :) [10:38] the new firefox breaks the old flash package because it installs stuff in to /usr/lib/firefox [10:38] it's fixed if you use the precise source [10:38] chrisccoulson, ok, it's happier now ;-) [10:40] hi! are we getting the new gnome-tweak-tool to match the FFe on gnome-shell? [10:41] ah. wait. nvm. seems it finally entered the archive. :) [11:03] no robert ancell? [11:06] Riddell, he's on paternity leaves [11:07] Riddell, and it's after his IRC,work around in any case, he usually leave around 8am utc [11:07] seb128: any other experts in lightdm.conf if me and agateau are stumped? [11:08] Riddell, don't ask to ask just ask? [11:08] Riddell, i.e I might be able to answer the question ... or not ;-) [11:08] Riddell, so ask away and we will see ;-) [11:10] well on the kubuntu-active live cd it tries to log into ubuntu.desktop [11:11] user-session=plasma-active being set and autologin-session=plasma-active being set [11:11] dh_auto_configure -- --with-greeter-user=lightdm --with-user-session=ubuntu is in the rules file [11:11] but i don't seem to be able to override that on the live cd even though it works fine on my installed system [11:12] I don't know if that's quite on topic for this channel :) [11:12] Riddell, did you check after boot that lightdm.conf has the value you expect? [11:12] Riddell, is casper or something generating a default config and overwriting yours? [11:16] seb128: I'm editing lightdm.conf after boot [11:16] to this http://starsky.19inch.net/~jr/tmp/lightdm.conf-kubuntu-active [11:17] Riddell, do you have a /usr/share/xsessions/lightdm-autologin.desktop ? [11:18] not sure what --with-greeter-user=lightdm --with-user-session=ubuntu is doing, but you normally need a postinst files calling /usr/lib/lightdm/lightdm-set-defaults [11:18] seb128: no [11:19] Riddell, also autologin-session=UNINPLEMENTED in the source lightdm.conf [11:19] autologin-session= is set by casper but it isn't actually implemented says agateau [11:19] Riddell, set user-session= where kdesession.desktop is your session in /usr/share/xsessions [11:19] didrocks: what does that do? [11:19] Riddell, i.e you need to have a name matching a session in there [11:20] seb128: yeah, user-session= doesn't help either [11:20] Riddell: it's editing lightdm.conf to the needed defaut values (and casper hooks that in as well) [11:20] Riddell, did you put a session which matches a .desktop in /usr/share/xsessions? [11:20] but if your live already have the right lightdm.conf values, its not related [11:20] (just that it should rather be edited by this tool) [11:22] seb128: yep [11:23] tried autologin-session=plsama-active and user-session=plasma-active and I have plasma-active.desktop [11:23] Riddell, can you pastebin the /var/log/lightdm/lightdm.log you get after the CD boot? [11:23] http://starsky.19inch.net/~jr/tmp/lightdm.log [11:23] I can work around it with casper cp plasma-active.desktop ubuntu.desktop but it's not very elegant [11:26] Riddell, weird, I don't know then, either the config is buggy or autologin just doesn't respect the user-session, maybe try asking the xubuntu guys on #ubuntu-devel how they do it? [11:27] Riddell, sorry but I don't know offhand, needs some debugging, maybe the xubuntu team has experience dealing with that and know what is happening? [11:27] xubuntu or lubuntu [11:27] yeah this is why i was asking for robert, go to the expert when in doubt :) [11:27] ;-) === greyback is now known as greyback|lunch [11:37] yay martinpitt@gnome.org [11:37] pitti, \o/ [11:41] uploaded to keyservers; if someone feels like gpg-signing that new ID using caff (and verifying the address), I'd be grateful [11:45] pitti, ok, will do [11:45] pitti, "caff"? [11:46] from the signing-party package [11:46] seb128: merci! [11:46] pitti, yw [11:46] seb128: or keybuk's signkey.pl, or whatever you use to sign keys with [11:46] I usually use caff or signkey to verify the mail addresses [11:47] pitti, I'm bad at signing key, I usually google and copy the commands I find on a debian wiki or mailing list :p [11:47] i. e. you send signatures to these mail addresses instead of uploading them, so that the recipient has to prove that he has access to these addresses [11:47] I never remember how I'm supposed to do it [11:47] seb128: caff is for you :) [11:47] pitti, thanks for the hint ;-) [11:47] it tells/asks you everything [11:48] * pitti throws hands into air upon libgnomekbd/gnome-screensaver/libxklavier [11:48] pitti: what did you do to deserve that eliteness? [11:49] Riddell: I became a committer some 1.5 years ago, a foundation member a month ago, a project maintainer about a month ago, and just now release tarball upload rights [11:49] fonudation membership goes with @gnome.org [11:50] I get sucked in pretty deeply :) [11:51] pitti, oh, it's only a start, before you know if vuntz will dump gnome-panel maintainship on you or something ;-) [11:51] pygobject keeps me more than busy, thanks :) [11:53] seb128: well, if I get it, I'll change it to take at most 10x10 pixels, and provide quick keyboard access to _everything_! [11:53] vim style, of course [11:53] ;-) [11:54] hjkl change between windows, :new gedit opens an application, d/p do copy&paste, etc. [11:54] who needs a desktop when he can get a vim! [11:55] pitti: oh you'd change it to be the HUD? :) [11:55] Riddell: heh, good point -- that's actually fairly close, in theory at least :) [11:58] pitti: HUD is in now? if I get a live image I can test it without setup? [11:59] yes, for a while already === MacSlow is now known as MacSlow|lunch === greyback|lunch is now known as greyback [12:35] ok, I need some food I'm starting [12:35] bbiab [12:56] * seb128 shakes fist at pitti [12:56] pitti, you are 1 bug ahead again on the desktop bugs fixed list [13:06] amateurs… :) [13:07] seb128: good morning [13:07] jbicha, hey jbicha, how are you? [13:08] good, looks like I've traded complaints about the power icon not working for keyboard shortcuts not working [13:08] didrocks, well at least I still kick your ass on launchpad karma score ;-) [13:08] jbicha, well, that was expected with the revert of the gsettings keybindings no? [13:09] yes, and the Ctrl+Alt+T one should be easy to fix [13:09] what do you think of g-s-d's 90_set_gmenus_xsettings.patch ? [13:09] I hope for you the power icon works again !!! ;) [13:10] seb128: yeah, I'm really wondering why [13:10] something is a bit unfair! [13:10] :) [13:10] jbicha, I can fix that today (90 xsettings) [13:10] seb128: disabling that patch makes the new gmenu work on Unity sort of (the menu actually shows twice) [13:11] jbicha, yeah, that's not called "work", it would turn off appmenus stripping from unity [13:11] jbicha, like menus would be locally displayed [13:12] jbicha, but I know how to fix it, I will do that in a bit [13:13] seb128: well it's better to show the menus twice than not at all (or only in HUD), but an actual fix would be even better [13:14] jbicha, you speak about gmenus right? [13:15] jbicha, that patch is not specific to gmenus, it tells what to do with any menu, without it gedit, nautilus etc would have dual menus as well [13:15] jbicha, but gmenus support is still on the appmenu's list for precise [13:15] so that should be hopefully fixed for beta2 [13:18] seb128: oh that's why I've been seeing so many menus all over the place! :) [13:18] ;-) [13:20] in fact I might call desrt for help on the gsd stuff when he's online === charles_ is now known as charles === MacSlow|lunch is now known as MacSlow [13:52] desrt: hm, did you see the followup comment in bug 918607? [13:52] Launchpad bug 918607 in pygobject-2 "Trying to register gtype 'GMountMountFlags' as enum when in fact it is of type 'GFlags'" [Medium,Fix released] https://launchpad.net/bugs/918607 [13:57] as of yesterday, the n-m indicator doesn't show me lists of APs? [13:57] and alt-f10 no longer maps to full-screen app [13:57] (though the keyboard shortcuts mgr says it does) [13:57] cyphermox: ^ could be a regression from the new nm yesterday [13:57] alt+f10 is unity [13:57] -> #ubuntu-unity for it [13:57] i don't see it in myunity [13:57] hallyn: oh, that is super+up now [13:57] oh [13:57] we stopped taking unity keybinding complains here ;-) [13:58] and super+cursor down for restoring [13:58] hallyn: can you screenshot so I can see? [13:58] pitti: interesting - the dash overlay with kbd shortcuts hasn't been updated :) [13:58] cyphermox: of n-m? [13:58] hallyn: as in, perhaps file a bug, make sure there's a screenshot for the menu, yes [13:58] lemme try... getting fancy here [13:58] oh ffs. [13:59] now it's showing it [13:59] hallyn: it is here [13:59] or at least the module used for your wifi [13:59] huh [13:59] but i've had two laptops not show it [13:59] maybe it just takesa while... [13:59] hallyn: alt+f10 opens the menu, super up/down for window managemetn [13:59] hallyn: that would suck anyway, though [13:59] actually super-down worked, but super-up not :) [13:59] cyphermox: i'll try to log in/out and get a screenshot [13:59] ok [13:59] * cyphermox tries to upgrade another computer to see [14:03] well this just isn't fair [14:03] when i wanted to check if i was using the bad AP... [14:03] oh well, thanks anyway :) [14:03] hallyn: I blame didrocks [14:04] :) [14:05] I wonder what would make scan results be slow though [14:05] (assuming that's what happened) [14:06] well, it may only happen when i'm on this bad AP when it's bad... maybe tha tmakes it hang [14:06] i just rebooted and the lsit came up [14:06] bad AP? [14:07] yeah (but the only one that works in this part of the house). only sometimes bad [14:07] hallyn: do you have a recent wireless card what could be having N issues, or some very relatively old one that uses ipw2200? [14:09] cyphermox: why me? :p [14:09] because you hadn't said anything, that was suspicious :) [14:10] never ever ;) [14:10] * didrocks whistle [14:24] dpm, there? [14:24] seb128, there, but on a call [14:25] dpm, in fact it might be a pitti's question [14:26] * desrt builds a kernel [14:26] ignore that [14:26] desrt, !!! [14:26] desrt, hey [14:26] uh oh [14:26] what's up? :) [14:26] dpm, pitti, that's ok, french translations are behind [14:26] hey desrt [14:26] good morning [14:26] desrt, a couple of easy questions for you if you have a minute [14:26] seb128: yes, known; langpack export was failing for a while, shoudl be fixed now [14:26] desrt, a hud bug also for you [14:26] micahg: ping [14:27] uh oh hud bug [14:27] better tell me that one first :) [14:27] desrt, is hudindicatorsource.c hud_indicator_source_name_appeared() where you build the system indicator names to display? [14:27] desrt, i.e .user_visible_name = N_("Device"), doesn't show translated for me [14:28] desrt, I guess you forget i18n init calls somewhere? [14:28] curious!! [14:28] yes. you're probably right. [14:28] like the binddomain? [14:28] okay. that's an easy bug :) [14:28] desrt, yes ;-) I was just trying to figure why those don't show translated [14:28] ya. you're absolutely correct. [14:29] RESOLVED: NOFRENCHALLOWED [14:29] desrt, you are at the end of the chain, I started by launchpad and langpacks export :p [14:29] lol [14:29] * seb128 distro patch, no en_CA allowed [14:29] don't worry. the americans know how to spell "Device" correctly, as far as i'm concerned [14:30] desrt, second one will interest you, I'm looking at fixed g-s-d for shell menus [14:30] i'll keep it :) [14:30] desrt, http://git.gnome.org/browse/gnome-settings-daemon/commit/?id=4c8902a8f10a78ad67b1443ef9039dda526238e8 is the commit [14:30] say, is .Xdefaults supposed to get loaded at login by xrdb these days? [14:30] desrt, do you see an issue duplicating the logic? [14:30] bc i'm having to do it by hand [14:30] desrt, i.e have g_bus_watch_name on shell and a second on unity with different callbacks? [14:31] hallyn, no, GNOME3 dropped support for those antic stuff [14:31] g-s-d stopped loading them [14:31] seb128: could use the same callbacks and a counter, i suppose [14:31] desrt, no, I need to set different values under unity [14:32] oh. that's right. [14:32] seems pretty reasonable. [14:32] desrt, but watching for both should be no issue right? [14:32] desrt, ok, good, thanks [14:32] desrt, happy kernel building ;-) [14:32] you can't just duplicate it straight, though [14:32] desrt, still trying to figure the trim stuff? [14:33] seb128: no. video driver bug. [14:33] now that i know about fstrim i'm much happier, actually [14:33] ok [14:33] a half year worth of daily jhbuilding on my SSD left it in pretty bad shape :) [14:33] desrt, can't duplicate? wdym there? [14:34] it's a lot faster now [14:34] seb128: so the code does this: notify_have_shell (user_data, TRUE); [14:34] desrt, I should look at how much mine is fragmented [14:34] it won't be that simple... [14:34] seb128: it's very easy. as root just do fstrim -v / [14:34] and also for /home if it is separate [14:34] desrt, ok [14:34] it's not to do with fragmentation but rather the number of free blocks that have not been returned to the SSD [14:35] (the more free blocks the SSD has, the better job of wear-leveling it can do) [14:35] desrt, can I do that while the disk is mounted? [14:35] yup [14:35] it's fast, too [14:35] cool [14:35] running... [14:35] (well... relatively fast) [14:35] desrt, anyway, back to gsd [14:35] so what do poor dudes like me do where fstrim doesn't work? [14:35] right [14:35] so each callback does this: [14:35] notify_have_shell (user_data, TRUE); [14:35] I basically wanted to add notify_have_unity() and the 2 _unity() on off [14:36] which is not quite right... [14:36] because if unity is running -> both set to TRUE [14:36] gnome-shell started -> one set to TRUE (and was already TRUE) [14:36] unity stops running -> both set to FALSE [14:36] but gnome-shell is still running... [14:36] i know this sounds pretty theoretical, but it's actually the case quite often [14:36] desrt, /: 14438768640 bytes were trimmed [14:37] since it's not unity, but rather unity-panel-service, and i often do run that under gnome-shell [14:37] seb128: 14gig. not bad. [14:37] how much free space does 'df' say? [14:37] /dev/sda1 72G 58G 10G 86% / [14:37] hah [14:37] so basically all of your 'free' space was stuck in the filesystem [14:37] and the SSD thought you had none at all [14:37] including the 'reserved' area, even [14:39] after doing that i noticed a bit of a performance increase [14:39] i think maybe we should have this in a cronjob or so.... [14:39] I will look at my next login to see if there is any difference on charts [14:40] it mostly affects the performance of write-heavy loads [14:40] (i think?) [14:40] desrt, so the g-s-d stuff is a bit tricky, I'm tempted to do an hack then [14:40] seb128: you just need to keep an extra state variable around [14:40] desrt, like call the same "update xsettings" callback whenever unity or shell come or leave [14:41] seb128: want me to write the patch? [14:41] desrt, if you want to do it in an upstream friendly way that would be great [14:41] i.e try to get it in GNOME 3.4 maybe [14:42] desrt, the extra state will not cover case where you log into shell, start unity and unload shell then though? [14:42] seb128: so i'm actually not thrilled about what gnome is doing [14:42] i.e you want to do a "revert to before the previous change"? [14:42] the current code is a bit of a hack anyway [14:42] but i'm not sure it's appropriate to change it at this point [14:43] desrt, I'm fine having it changed upstream next cycle [14:43] seb128: basically, there needs to be a 'shell_running' bool and a 'unity_running' bool [14:43] i'll update the distro patch :) [14:43] desrt, thanks [14:43] * desrt is waiting for a kernel to build anyway :p [14:44] seb128: so you help me with quilt? [14:44] desrt, sure? [14:45] i want to edit the existing patch... [14:45] apt-get source gnome-settings-daemon [14:45] cd gnome-settings-daemon-... [14:45] ya. i did all the easy stuff so far :) [14:45] quilt pop 90_ [14:45] edit [14:45] quilt refresh [14:45] give me 90_... [14:45] great. thanks. [14:45] if you change new files you need to quilt add them before starting changing them [14:45] can i do a test build in between? [14:45] or will quilt become unhappy? [14:45] will only be the one file... [14:46] desrt, quilt doesn't care, as long as it can apply,unapply the patch on build,clean [14:46] desrt, if you build inline, ie ./configure etc it will not care at all [14:46] nod. [14:46] it will just apply your changes to the patch when you "quilt refresh" [14:46] i'll just get the patch right the first time :) [14:46] your changes to that specific file [14:47] quilt feels like one of those really cool things that i should learn [14:47] so no need to bother will all the build generated files or anything [14:47] but i've never had much reason [14:47] desrt, the way you should do it is to build, quilt pop 90.... then [14:47] hack, make, cp try [14:47] until happy [14:47] quilt refresh then [14:47] and send me the updated patch [14:48] ya. working now. [14:48] gimme a sec :) [14:50] quilt edit and quilt shell are good to know [14:59] gsd test build now competing with kernel build :) [15:00] seb128: what's up with this: /home/desrt/kitchen/gnome-settings-daemon-3.3.91/plugins/xsettings/gsd-xsettings-manager.c:526: warning: undefined reference to `XOpenDisplay' [15:00] whole lotta errors like this in the build log, but it doesn't seem to fail the build [15:00] desrt, yeah, I've the feeling --as-needed doesn't work as it should but doko says it's working and I didn't find the time to get enough details to prove he's right or open a bug [15:01] i think the problem is that this is a module that assumes it can open the symbols of the program that it is loaded into and the linker doesn't know about that [15:01] so it gives the warnings [15:16] ogra_: Hi Oliver, see that you are piloting today. Can you please help me get two simple MPs merged? [15:16] * https://code.launchpad.net/~gunnarhj/lightdm/guest-session-skel/+merge/97652 [15:16] * https://code.launchpad.net/~gunnarhj/gnome-settings-daemon/patch43/+merge/91210 [15:16] * didrocks goes to take the train (I'm off tomorrow), will try to connect from it using my 3G connexion. Anyway, will work offline on g-c-c in a less-pingy world hopefully ;) [15:19] I guess didrocks quit because he knew I was about to ask him something [15:20] jbicha: pong [15:26] rickspencer3, yeah, I would probably have closed my IRC as well if I knew [15:26] rickspencer3, too late now I guess :p [15:26] fair enough :) [15:27] jbicha, you dupped gnome-screenshot on version, I already added it this morning [15:27] jbicha, I already did the "list individual gnome-utils" component [15:28] jbicha, oh, and I didn't list the viewer since we don't have this source (yet) [15:31] seb128: ok, I had to merge my changes with yours, you can take off the font-viewer if we're not going to bother with it this cycle [15:32] jbicha, let it in, I think we should upload it [15:32] jbicha, do you know if it can be synced or if it needs a merge? [15:32] jbicha, I would appreciate if you could look at it, I can sponsor if needed [15:33] seb128: well at that rate, we might as well do gnome-dictionary too as the last one, right? [15:34] jbicha, that would be good [15:34] jbicha, pitti, others: btw I did some versions bug fixing, improvements today [15:34] micahg: did you still want to update the universe cogl rdepends? [15:34] it should be easy to change between tracking unstable or stable GNOME series now [15:34] or to include,exclude sources from the stable,unstable tracking [15:35] oh, I also made the list of sources blocked on a specific serie or version printed at the bottom of the page [15:35] I'm piloting Monday, so I can sponsor stuff for you then or maybe work on rebuilds over the weekend [15:36] seb128: that last one is nice so that we don't have to keep explaining why we're at certain versions [15:36] right [15:37] is https://bugs.launchpad.net/ubuntu/+source/compiz/+bug/956108 known? i.e. crash of compiz during the upgrade [15:37] Launchpad bug 956108 in compiz "crash during oneiric->precise upgrade leaving screen black while the upgrade was running" [Undecided,New] [15:38] seb128: looks like i have a working patch now [15:39] seb128: http://fpaste.org/LiJM/ [15:41] mvo, no [15:41] desrt, thanks [15:41] woo. and just in time for my new kernel to be done [15:41] desrt, I'm one as well [15:42] seb128: you are a working patch? :) [15:42] desrt, adding those to hud-service.c makes my hud french [15:42] setlocale (LC_ALL, ""); [15:42] bindtextdomain ("indicator-appmenu", "/usr/share/locale"); [15:42] textdomain ("indicator-appmenu"); [15:42] desrt, I'*ve* one :p [15:42] oh. hah [15:42] desrt, well I changed the macros for fixed value because for some reason the -DGETTEXT etc are not set [15:42] we usually don't do it like that [15:42] oh. [15:42] desrt, I will debug that now [15:42] ya... probably better to fix that, then [15:43] want me to do so? [15:43] desrt, I just wanted to see if that was enough to solve it [15:43] desrt, no, that's fine, get going on the xul stuff rather [15:43] :p [15:43] ;-) [15:43] now i can't blame you in my excuses to olli :) [15:44] ;-) [15:44] ricotz: oh, we broke Cinnamon because it doesn't work with the development version of clutter/cogl yet? forking gnome-shell & mutter never seemed like a great idea to me [15:45] fortunately it's not in Ubuntu yet so NotMyProblem either :) [15:45] jbicha, hi, that isnt bothering me though ;) [15:46] (looking at there branches it seems quite a mess) [15:49] I'm amazed that Fedora let muffin in, maybe they'll get cinnamon too [15:50] That was a massive effort. [15:50] To fork gnome-shell AND mutter. [15:50] It's open tech, gotta users the option. [15:51] *gotta give users [15:52] I think they're doing it wrong, Shell supports extensions & if something isn't possible yet, write a patch [15:53] ok, time for some exerice [15:53] exercice [15:53] bbiab [15:58] GunnarHj, is there a bug for the guest-session/skel issue ? [15:59] ogra_: No, is it necessary? [15:59] well, would have been nice :) [15:59] ogra_: Ok, then I'll file one. [16:00] can you add it to the changelog too ? [16:00] i approved the other one, but would like to leave the uploading in sebs hands rather [16:01] jbicha: you're right, but eh. 'tis the ebb and flow of f/oss, I guess. [16:06] seb128, "exercise"? don't you mean "beer"? [16:17] ogra_: Yes I'll modify the changelog as well. Back soon. [16:31] pitti: FYI I updated ubiquity to ignore None values and revert to append() so that it works now and we can upload later today [16:31] stgraber: ah, good [16:31] stgraber: I got the patch in for making insert_with_valuesv() working [16:32] stgraber: but still need a review for the "atomic append" patch [16:32] pitti: cool, I'm sure atomic append will make quite a few people happy ;) [16:33] not sure how much testing our other python apps had with accessibility on, I'm a bit surprised only ubiquity was affected [16:38] stgraber: TBH, I'd rather drop the override API [16:39] it's diverging from the C API [16:39] and we can now use the C API [16:41] hmm, i wish i had more disk space [16:44] * ogra_ hands chrisccoulson a spare Sd card he has lying around [16:44] here, 4G for you [16:46] ogra_: I created bug 956152 and modified the changelog accordingly. [16:46] Launchpad bug 956152 in lightdm "A more flexible use of the /etc/guest-session/skel hook desirable" [Undecided,In progress] https://launchpad.net/bugs/956152 [16:46] GunnarHj, great, will merge and upload soon [16:46] ogra_: Nice, thanks! === lifeless_ is now known as lifeless [17:11] Ohh :( gnome-shell segfaulted :( https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/956212 [17:11] Launchpad bug 956212 in gnome-shell "gnome-shell crashed with SIGSEGV" [Undecided,New] [17:29] chrisccoulson, shusshhh! that's a secret ;-) [17:29] heh, even i did some exercise! [17:32] g'ah, random hanging tests again [17:32] this is getting seriously annoying [17:33] chrisccoulson, lager? ;-) [17:36] sigh :( [17:36] so, the test passed yesterday, and fails today [17:36] no code changes [17:36] fantastic [17:36] its a wednesday test ! [17:36] nothing like random errors [17:58] mterry, question for you, what should I set SetBackgroundFile to in accountsservice if I've no picture background? [17:58] seb128, empty string [17:59] mterry, [17:59] variant = g_dbus_proxy_call_sync (user, [17:59] "SetBackgroundFile", [17:59] g_variant_new ("(s)", ""), [17:59] mterry, basiclly? [17:59] sure [17:59] mterry, thanks [17:59] mterry, fixing a small bug in your gsd patch [17:59] :-/ thanks! [17:59] did I get that case wrong? [17:59] mterry, if somebody picks a solid color it tries to set it to NULL [18:00] which g_variant complains about since that's not a string [18:00] Ah... thought I tried that. Guess not! [18:02] mterry, I will just use background ? background : "" [18:02] seb128, cool === Trevinho is now known as Trevinho|afk [18:05] good night everyone! [18:05] 'night pitti === JackyAlcine_ is now known as jalcine [19:43] chrisccoulson, there? [19:43] seb128, i'm always here ;) [19:43] desrt, ^ [19:44] actually, i'm no longer here :P [19:44] chrisccoulson, desrt is unsure on how firefox works, he's still to make hud work with firefox menus and we have an issue [19:44] ah, ok [19:44] still->trying [19:44] i guess i could have a look at it [19:44] i've not used the hud yet [19:45] chrisccoulson, I think desrt will have specific questions [19:45] chrisccoulson: when i give the 'opened' dbusmenu event, what does that mean? [19:45] does that mean that the menus will be kept up-to-date until i send the 'closed' event [19:45] or does it mean that the menus are only brought up-to-date at that instant and may go out of date? [19:46] to put it more concretely: [19:46] if somehow a bookmark was added while i had the bookmarks menu open [19:46] would i see the change or would i have to close/reopen in order to see the change? [19:46] desrt, the menus stop updating once the opened event is received, and we also use that to dispatch a popupshown event on the menu (which some other parts of firefox might be listening too) [19:46] yeah, you'd need to close and reopen the menu again. i did it that way to stop the history menu jumping around [19:46] ouch [19:47] is that causing a problem then? [19:47] no [19:47] it just sucks :) [19:47] i guess it's the same issue as the 'about to show' thing in Qt [19:47] seb128: so the patch i just pushed could very much have an impact, then [19:48] seb128: i was leaking HudQuery, which means that after the first query, the menus were being told that they were permanently open [19:48] desrt, no luck [19:48] which means that for dynamic menus (like bookmarks) they were probably never being refreshed [19:48] desrt, well the way I test is that I close firefox, open it, tab and type the name of a bookmarl [19:48] bookmark [19:49] tab? [19:49] desrt, i actually don't get anything from firefox in the hud atm [19:49] which doesn't work until I open the menu once [19:49] desrt, tab is the hud key [19:49] even stuff which isn't generated dynamically [19:49] seb128: so... alt? [19:49] desrt, yes, sorry, it's late ;-) [19:49] seb128: with the hud running all along, right? [19:49] chrisccoulson, you need lp:~desrt/indicator-appmenu/hud-rewrite-wip [19:50] ah, ok [19:50] thanks [19:50] seb128: let's get a little bit more deterministic [19:50] desrt, I install the update, kill the hud, start firefox, alt, type [19:50] open a terminal [19:50] killall hud-service [19:50] run hud-service [19:50] start firefox [19:50] desrt, READ [19:50] alt [19:50] ;-) [19:50] i did. it's not deterministic enough :) [19:51] well you got the same first 4 steps [19:51] so please keep going ;-) [19:51] WEIRD [19:51] if i let it activate (the way you're doing it) then i see the bug, indeed [19:51] oh [19:51] your step have "run the hud" before "start firefox [19:52] no, that doesn't make a difference [19:52] okay. i can reproduce it now. [19:52] I had the results listed, I restarted firefox and it's buggy again [19:52] desrt, what did you change? [19:52] i was testing firefox already open -> switch to it [19:53] i have a theory [19:54] when you do the query it causes all of the existing submenus to be opened [19:54] but if opening those submenus results in more submenus being shown (as a direct result) then those new submenus are not also shown [19:54] desrt, no [19:55] actually, this is getting a bit ugly.... [19:56] i'm iterating a hashtable of the items in order to show them all [19:56] and if dbusmenu could possibly call back into my code as a result of that, it could be modifying the hashtable while iterating [19:57] desrt, you are right I think, I've added a bookmark not in a subfolder and it's listed [19:57] this means dbusmenu is calling back into me during the 'opened' event :( [19:58] would it make sense to iterate over a temporary copy? [19:58] yes. that's what i'm thinking [20:04] so. the good news is that dbusmenu actually isn't so insane as to call back into me [20:04] the bad news is that i need another explanation :) [20:04] :( [20:04] i'm sure it will be forthcoming soon [20:06] erm [20:06] dbusmenu_menuitem_send_about_to_show [20:06] what is the difference between this and OPENED/CLOSED events? [20:09] heh, good question! [20:10] so if i send about-to-show along with OPENED then it works fine [20:10] i think the intention is that one is sent before opening the menu, and the other is sent afterwards. but it doesn't actually work like that anywhere [20:10] that's... frustrating [20:11] actually, firefox will probably even behave the same if you just send the opened event [20:15] you appear to be correct [20:16] the bad news is that i can no longer reproduce the problem :p [20:20] hmmm, fingers crossed that none of the tests hang this time: https://launchpad.net/~mozillateam/+archive/firefox-next/+build/3289569 [20:20] if they do, then i'm going to have to drink myself to sleep this evening :) [20:21] dbusmenu is like one race condition wrapped in another :/ [20:22] chrisccoulson, just don't sleep and hack it's easier! [20:22] ;-) [20:27] figured it out [20:28] yay :) [20:28] ah [20:29] fantastic . my build has passed the stage that it hung at [20:29] must have been a SEU ;) [20:29] chrisccoulson, SEU? [20:30] single event upset [20:30] is that a word? ;-) [20:30] well, work -> expression [20:30] yeah. it's a term we were quite familiar with in aerospace [20:30] seb128: okay. please try again. [20:30] although i seriously doubt my problem was caused by SEU. it was more of a joke ;) [20:31] the issue was with my code that automatically issues OPENED events for new submenus that appear while a query is in effect [20:31] it was working for new items, but dbusmenu also has the ability for an item to originally be created as a normal item and suddenly become a submenu [20:31] and in that case, OPENED wasn't being properly sent [20:33] it's actually extremely tricky -- the act of starting the query causes more menu items to pop into existance which then themselves must be added to the query [20:33] this sort of use pattern would never exist outside of something like the hud [20:34] desrt, oh, are we actually changing a normal menu item in to a submenu in firefox? [20:34] chrisccoulson: seems to be the case [20:34] i'm pretty sure i made sure that we didn't do that :( [20:34] DBUSMENU_MENUITEM_PROP_CHILD_DISPLAY is initially unset and then later it becomes set [20:34] it could be a side-effect of the way dbusmenu sends the data across the bus though [20:35] hmmm [20:36] we set it manually when we create the menu in firefox, even before we add any items to it, to try and avoid doing that: http://bazaar.launchpad.net/~extension-hackers/globalmenu-extension/trunk/view/head:/extensions/globalmenu/components/src/uGlobalMenu.cpp#L383 [20:36] desrt, no luck [20:36] tf... [20:36] hum [20:36] seb128: are you getting lots of g_warning messages from libdbusmenu meanwhile? [20:36] desrt, where? [20:36] from the hud process [20:37] desrt, none [20:38] are you allowing the hud to be activated? [20:38] or starting it manually? [20:38] it's weird [20:38] I tried both [20:38] it seems to work every second time in fact [20:38] I got 2 working case on 5 tries [20:38] it's a race :) [20:38] there's another buglet in the hud that could be confusing you [20:38] i realised the other day that it exists [20:38] let me fix it [20:38] desrt, well there I ran the hud on a command line to see warning [20:39] the window source doesn't pick up on the focus window that's existing when the hud is first started [20:39] I start firefox, alt, type a bookmark name [20:39] so if you don't switch away/back then it won't see it [20:39] restart firefox, alt, type [20:39] sometimes it works, sometime not [20:39] i.e I don't restart the hud between tries [20:39] it works every second try [20:40] i think it may have to do with the amount of time between your keystrokes when you do the search [20:40] ie: if the menus have time to be delivered meanwhile or not [20:40] could be [20:40] I'm starting and typing directly [20:41] let me try with a few seconds pause [20:41] desrt, oh, i just found a case where we can change from a submenu to a normal menuitem and then back again [20:41] chrisccoulson: nice :) [20:41] dbusmenu does that automatically for you if you remove all of a menus children [20:41] desrt, yeah, ok, it's not working when I try like in the second after start [20:41] which is a bit strange :/ [20:41] desrt, so yeah, for normal users it works ;-) [20:42] NOT GOOD ENOUGH! [20:42] ;-) [20:42] seb128: you've not seen dbusmenu warnings? [20:42] * desrt gets them if he types his search too quickly [20:43] LIBDBUSMENU-GLIB-WARNING **: Generating properties error for: 888 [20:43] LIBDBUSMENU-GLIB-WARNING **: Error getting properties on a new menuitem: Error getting properties for ID [20:43] I got some of those now [20:43] "some" being an understatement maybe [20:43] ya... same story here [20:44] * desrt makes a super-deeply-nested bookmark menu for testing [20:47] i hate you, firefox [20:48] 1) create deeply nested bookmark for testing [20:48] 2) see it show up at the second level under "Recently bookmarked" [20:48] 3) delete it from there [20:48] 4) note that it has been completely deleted [20:53] that will probably fix the menuitems changing type: http://bazaar.launchpad.net/~extension-hackers/globalmenu-extension/trunk/revision/281 [20:54] that'll help prevent some thrashing in the hud [20:54] but it should still be able to deal with it in theory [20:54] i'm still not 100% sure that's where it happens, but that's the only place i can think of where it might happen [20:55] :) [20:56] fwiw, it's the common case that items are created without being submenus and only become submenus later [20:57] at least as i see it on my side [21:01] hmmm :/ [21:01] perhaps there's not much i can do to avoid that [21:03] awesome [21:03] so nothing is wrong with the hud or firefox [21:03] it's unity [21:04] dbus output says that the hud is property emitting the QueryChanged signal when the new results come from firefox [21:04] unity is failing to show them [21:04] chrisccoulson: thanks for your help :) === JackyAlcine is now known as jalcine [21:33] kenvandine: thanks for fixing that annoying gwibber crash [21:39] seb128: found the issue... [21:39] desrt, great [21:40] the front end is not responding to the query-updated signals [21:40] unity bug then? [21:40] so as firefox populates the menu items dynamically, the hud-service picks them up and emits change signals for the query, but they never show in the UI [21:40] ya. i think so [21:40] ok, open a bug for gord I guess ;-) [21:41] ya... i'm just making sure that i'm adhering to the protocol properly [21:41] ie: maybe i emit the change signal in the wrong way [21:41] right, time to undock and move to the lounge. this is where my laptop locks up [21:41] chrisccoulson: best of luck [21:42] seb128: also discovered something quite odd [21:43] seb128: you were really right about the every-second-time thing [21:43] it seems like every second time the hud pops open an empty query gets sent to the service which causes the menu items to be pre-populated (causing the next non-empty query to be successful) [21:43] not sure what the story is there, but it's interesting :) === chrisccoulson_ is now known as chrisccoulson [21:48] desrt, ok ;-) [21:48] chrisccoulson, your laptop still freeze on undock? [21:49] seb128, yeah. it did it at the start of the cycle, and then it was resolved at some point [21:49] but now it started again [21:49] everything is running, and i can ssh in to it, but it's just that the display goes blank and i can never get it to come back on again [21:49] even if i restart X [21:50] happens on resume from suspend occasionally too [21:50] i guess i should report a kernel bug [21:50] gord: hey [21:50] gord: please take a peek at https://bugs.launchpad.net/ubuntu/+source/unity/+bug/956480 when you get a chance [21:50] Launchpad bug 956480 in unity "hud searches don't update properly" [Undecided,New] [22:00] desrt, not on a good machine to check atm, but iirc we listen to the UpdatedQuery but only act if the query we receive is the same as returned by StartQuery checked with g_variant_equal - could there be something wrong there? [22:04] gord: so i added some debugging prints to determine exactly that [22:04] gord: my debug prints never ... print [22:04] so i'm wondering if somehow the logic that setups the signal watch is broken [22:07] could be that my understanding of how to install a patched unity is also incorrect :) [22:08] its code that we use throughout the rest of unity, so i would be surprised [22:08] http://bazaar.launchpad.net/~unity-team/unity/trunk/view/head:/UnityCore/GLibDBusProxy.cpp if you want to take a look [22:08] i'm going to make a testing source for the hud [22:09] hum [22:09] the only way to trigger it now is on some weird conditions with firefox [22:09] this will make it a lot easier to see the issue [22:09] looks like a signal in unity doesn't get fired [22:10] http://bazaar.launchpad.net/~unity-team/unity/trunk/view/head:/UnityCore/Hud.cpp#L151 there should be a signal emit about there [22:10] ah. so found the issue already? :) [22:10] i'm going to work on my debug source in any case so that you have a good basis for testing your fixes [22:11] maybe, hopefully :) will wait for the morning before trying it [22:11] okay [22:11] i'll be done the debug source by then [22:24] ok, I'm off [22:24] bye all! [22:24] ta [22:35] gord: hey [22:35] gord: i just pushed a branch with the debug code [22:35] gord: grab this: https://code.launchpad.net/~desrt/indicator-appmenu/hud-rewrite-wip [22:35] then run it like so: HUD_DEBUG_SOURCE=1 G_DBUS_DEBUG=message ./hud-service [22:36] after you do a search you should see a once-per-second update of the time item being sent as a signal on the bus [22:36] unity never refreshes, though [22:43] RAOF: good morning [22:43] desrt: Good morning. [22:43] RAOF: the patch on the xorg list fixes the issue on my laptop [22:44] Your purple snow? [22:44] yup [22:44] Nifty. [22:44] we've a bug for it here: https://bugs.launchpad.net/bugs/956046 [22:44] Launchpad bug 956046 in linux "intel driver fails to use dual-link LVDS if lid is down at boot" [Undecided,Confirmed] [22:47] Cool. [22:50] Hm. How do you get byobu-tmux to send an F5 keypress to the client, rather than eating it? F12-F5 doesn't seem to do it.