[01:52] JackYu: ping [01:54] hapyyaron, pong [06:58] hey robru, still around to update me on the Unity8/Upstart app launch things? [06:59] didrocks, haha. so mirv filed a bug about upstart, and unity8 never landed due to merge conflicts in trunk. [07:00] didrocks, also i had a power outage here and it resulted in corruption on my boot drive, so I've spent most of the day reinstalling and recovering my system :-/ [07:01] robru: argh, I hope you can recover :/ [07:01] robru: what's this merge conflicts in trunk? This is for the branch we are waiting on landing? [07:01] didrocks, almost there... [07:02] robru: also, having a link to this branch we wait on for releasing would be appreciated ;) [07:02] didrocks, yeah, sil told me the MP that we should watch, and I saw it get approved, and then jenkins failed to merge it due to flaky tests, so I went for a manual merge and there were conflicts, it wouldn't merge after that. I told them on the MP to rebase on trunk [07:02] making sense [07:02] didrocks, https://code.launchpad.net/~aacid/unity8/broken_collapse/+merge/197844 [07:03] robru: excellent thanks! [07:03] robru: btw, do not add any other landing please until we promote an image [07:03] especially risky ones like ubuntu-ui-toolkit [07:03] didrocks, they told me it fixed test failures ;-) [07:03] we want to be able to go back to a good state before putting cracks in [07:03] robru: yeah, fixes and regressions :p [07:03] (potential at least) [07:04] apart if they only touch codes in tests/ [07:04] didrocks, ok, please let t1mp know that you veto'd it, since I already said i would do it [07:04] robru: will do [07:04] yeah, look at trunk [07:04] quite a lot of changes [07:05] let's hope we can promote an image today [07:21] didrocks, ahhhhh, I need the VPN details. is there a wiki for that somewhere? [07:34] robru: hum, I think I sent an email about that 3 weeks ago :) [07:35] robru: sent [07:35] robru: I hope you recovered your emails [07:36] didrocks, emails? those are all in gmail! [07:36] the cloud is great ;-) [07:36] didrocks, yeah, personal files were well backed up, but I lost the system partition, so things like vpn config tokens are gone [07:40] robru: oh, mine in my ~ dir [07:41] thanks for network-manager [07:41] to* [07:41] didrocks, where does it hide? i was using network manager... [07:42] robru: IIRC, it's in gsettings [07:43] didrocks, how do I extract gsettings from a not-running user directory? ~/.local where? [07:43] robru: hum, forget about it? I don't find them in a dconf dump [07:45] didrocks, anyways, it's midnight. i'll get back on the VPN tomorrow [07:45] yep, sounds better :) === penghuan_ is now known as penghuan === mvo_ is now known as mvo === pitti_ is now known as pitti [09:02] good morning desktopers [09:02] happy friday! [09:03] hey! [09:03] xnox: w o w ! [09:03] Laney, hey, how are you? [09:03] bonjour seb128, hey Laney [09:03] pitti, salut, ça va ? [09:03] hey seb128 pitti [09:03] ça va bien ! [09:03] pretty decent, glad that it's friday :-) [09:03] Laney, congrats on defeating webkitgtk [09:04] crazy eh? [09:04] ;-) [09:04] just have to test build every future upload on 4 arches from now on [09:04] I'm surprised you didn't go "I never want to upload webkit ever again" [09:05] there is that [09:05] ;-) [09:06] Laney: I don't even think we need it much longer. The web seems to be a passing fad [09:06] port all to oxide [09:06] then deprecate webkitgtk and let attente maintain it [09:06] you guys are just mean to attente! [09:06] attente already had his troll friday last night [09:06] oh, wait, it's friday [09:06] ok then [09:06] haha [09:07] I thought you redeclared them to learn-something-Fridays? [09:07] I guess we all learned how much we like trolling... [09:07] lol, indeed [09:07] larsu, btw, gtk 3.10.6 in the ppa, seems to work fine there [09:08] I'm pondering uploading to trusty today or give it the W.E and upload on monday [09:09] give it the weekend, please :) [09:09] oh, invisible cursor [09:09] k [09:09] haven't had that bug in a couple of weeks :P [09:09] * larsu doesn't want to wake up to angry emails on Saturday [09:09] Laney, :-( [09:09] Laney, when did you last update/restart xorg? is that happening at boot/login? [09:09] i just turned it on today [09:09] and after starting the session, don't know if it was there on the greeter [09:09] is that current trusty? [09:10] ish [09:10] no ish [09:10] I think I dist-upgraded yesterday [09:10] not today [09:10] Laney, https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/1238410 [09:10] Launchpad bug 1238410 in X.Org X server "Inconsistent cursor visibility with cursor plugin enabled" [Medium,In progress] [09:10] Laney, it was supposed to be fixed with xorg's update on wednesday [09:10] Laney, can you make sure if you have that version or not? [09:10] xorg-server? [09:10] see comment 24 [09:10] on that bug [09:11] ok /me looks [09:12] ah [09:12] Installed: 2:1.14.3-5ubuntu1 [09:12] Candidate: 2:1.14.4-1ubuntu1 [09:13] tedg: what do we need to do in xfce4-indicator-plugin to handle upstart indicator activation? [09:13] just system("init ..."); ? [09:13] Laney, I prefer that ;-) [09:13] ali1234, Yeah, did you see the commit to unity-greeter? [09:13] tedg: yes [09:13] indeed [09:13] ali1234, I think that's the best way. [09:14] i guess i need to use something that doesn't wait for the process to finish, ie fork() [09:14] ali1234, Though, I'd really recommend switch XFCE over to Upstart User Sessions :-) [09:14] xubuntu uses upstart [09:15] System session, probably not user session. [09:15] it uses upstart for the user session [09:15] Really? [09:15] yes really. we had to get it fixed [09:15] Cool. Then you shouldn't need to do anything. [09:15] well, ok then, your new indicators are broken :( [09:15] Oh, you need to emit that event. [09:15] half the time they don't load up [09:16] So just a "initctl emit-event indicator-service-start" [09:16] hmm. [09:16] So just a "initctl emit-event indicator-services-start" [09:16] (notice typo in first one) [09:18] assuming you meant initctl emit [09:18] is that command supposed to hang forever? [09:18] Yes, sorry it's early :-) [09:18] Shouldn't be forever... [09:18] oh i see... i already had run the command that starts the job direct [09:18] You can use --no-wait if you want it exit right away. [09:19] ok. so what should emit this event? [09:19] the indicator plugin? or something earlier in the session startup? [09:19] Laney, why did you put a block on webkitgtk again? (did I ask before?) [09:19] just to test it [09:19] I'll remove it shortly [09:19] ok [09:20] then start on the g-o-a transition [09:20] \o/ [09:22] ali1234, I'd say the indicator plugin, then you can emit the end as well to manage the full lifecycle if it gets added or removed. [09:24] tedg: ok, thanks. === ochosi_ is now known as ochsoi === ochsoi is now known as ochosi [09:26] oh come on [09:27] Laney, don't tell me that webkit is buggy [09:27] no [09:27] my session won't open [09:27] won't open? [09:27] do you need help debugging? [09:27] greeter goes away, hangs [09:27] well, not hang, but no compiz [09:28] sec, checking logs [09:28] no compiz or no unity? [09:28] both [09:28] hum [09:33] I've got various things like this in logs: (gnome-settings-daemon:3897): Gdk-WARNING **: gnome-settings-daemon: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. [09:33] silly question, but is your hard drive full? [09:34] /dev/sda5 120G 102G 13G 90% / === czajkows1i is now known as czajkowski [09:37] Laney, what do you have in your gnome-session.log ? [09:37] Laney, those seems like xorg is closing/have an issue [09:37] a load of those [09:38] does it work to start a guest session? [09:39] no, it's the same [09:40] using nouveau? [09:40] yeah [09:40] well, you can try to ruling out lightdm by going to a vt and do startx : --1 or something [09:51] hey ho ... my trusty desktop system just came to stall (no mouse/keyboard input for 20+ seconds) with oneconf-service and x*apt-something looping at 100% at the same time [09:52] maybe there is an unfortunate timed cron job that kicks both at the same time? [09:53] it is just not storm safe [09:53] check the ropes ;) [09:53] asac, I'm not sure they are croned, they might just react to apt index changes, didrocks probably knows since he wrote oneconf [09:54] ogra_, do you still get s-c to hit that Xerror? [09:54] seb128, dunno, will test when i'm at the machine [09:54] (and report back) [09:54] ogra_, thanks (could use debug infos from somebody having the issue) [09:54] yup [09:59] they aren't cronned but trigger by dpkg [09:59] well for oneconf at least [10:04] Laney, did you figure it out? [10:05] no :( [10:05] is startx working? [10:10] It waits for ages and eventually I get a timeout locking .Xauthority [10:10] even if I rm it beforehand [10:16] tedg: the indicator services don't start up fast enough and i just get an empty panel [10:22] ali1234, Hmm, but they do start? [10:22] yes [10:22] ali1234, Do you then populate as they come up? [10:22] no [10:23] ali1234, Are you using the libindicator code? I thought it handled that. [10:23] libindicator3, yes [10:24] it doesn't handle it [10:25] i'm building with a long delay to see what happens [10:25] K [10:33] tedg: i see what is happening [10:34] when it runs "initctl emit indicator-service-start" it hangs, because the service is already running for some reason [11:00] tedg: libappindicator is falling back to systray for no good reason [11:05] larsu: http://ubuntuone.com/1GS1IppvwNFus6HMOZEXeA ← see the white bars on the toolbar - is that a gtk / theme problem? [11:08] http://bazaar.launchpad.net/~ted/indicator-application/upstart-job/revision/245 [11:10] tedg: you just broke appindicators for gnome-panel and xubuntu [11:13] ali1234, That's just application indicators, but I expected those folks to use the XDG autostart file. Happy to change it around. [11:14] xdg-autostart doesn't work for appindicator-service [11:14] it immediately exits if nothing is using it [11:14] actually, i don't see how it can work with upstart either [11:14] at the very least there's a nasty race condition [11:17] my logs are full of "service respawning" spam because of this shutdown-if-nothing-is-using-me functionality [11:17] this is probably why they only load up half of the time too [11:18] also the indicator-services-end signal doesn't appear to do anything [11:20] Yeah, probably removing those timeouts makes sense now that they're managed by the session management. [11:20] sometimes they get killed because of respawning too fast [11:21] i'm not sure if xubuntu is using upstart or xdg-autostart to start these services, but all i know is that half of them are already loaded when i log in, sending the startup event hangs initctl, and appindicator-service is impossible to start or use in any way [11:21] Laney: looks like it, yes. Which version of epiphany is that? [11:22] larsu: it's epiphany-browser 3.8.2-4ubuntu1 from trusty-proposed [11:22] ah okay, that's why I'm not seeing that yet [11:22] just about to unblock webkit, so you'll get it soon :P [11:23] yay [11:23] I'll have a look into that, then. Thanks for pointing it out [11:23] yw [11:23] Laney, you really use web from browsing? [11:23] or did you just test webkit? [11:23] EPARSE [11:23] web = epiphany [11:23] was just to test webkit [11:23] s/from/for/ [11:23] oh, right [11:24] do you need extra testers for webkit? [11:24] did you figure out your login issues btw? [11:24] yeah, was a non-executable file which broke an upstart job that blocked the session login [11:25] shrug [11:25] that seems like unreliability [11:26] the design of upstart allows any job to block any other job [11:26] so you'll get things like this, unfortunately [11:26] I'm sure there could be better troubleshooting though [11:27] fair enough, and the real issue seems like your -x file [11:27] yes, I uploaded a fix for that [11:27] you could have like upstart watching for failed job starts or something [11:28] or unity not coming up, and dumping its state to errors [11:28] right [11:30] tedg: via use of INDICATOR_ALLOW_NO_WATCHERS env var, i was able to make the service work properly [11:30] feel free to bash on webkit but I'm going to unblock it anyway; can fix stuff in release :P [11:30] also, src:webkit can be removed now [11:30] ali1234, Great, that makes sense. Makes more sense to do it that way. [11:30] well, once it migrates [11:32] and luckily this timeout is implemented in libindicator3, so it only has to be removed in one place for all of them :) [11:35] uhm ... so after todays upgrade evolution goes constantly to 100% CPu usage :( [11:46] seb128: FYI, bug 1258458 is a dupe, fixed already two weeks ago [11:46] Launchpad bug 1252305 in apport (Ubuntu) "duplicate for #1258458 dpkg-divert not found" [Critical,Fix released] https://launchpad.net/bugs/1252305 [11:47] people ought to upgrade their trusty a bit more often :) [11:57] pitti, thanks, we should also block reports from outdated systems (same way apport is doing it) [11:57] ogra_, weird, we didn't change a lot yesterday [11:57] ogra_, what did you get in your round of upgrades? [11:57] glib perhaps ? [11:58] Laney: if you put INDICATOR_ALLOW_NO_WATCHERS=yes in /etc/environment this may provide a temporary workaround for missing indicators [11:58] though the changes dont look like they could cause issues [11:58] I doubt it [11:58] that's only adding some python debug stuff [11:58] right [11:58] did you get the new webkit? [11:58] ali1234: cheers [11:58] is it in the archive ? then i should have gotten it [11:59] are you going to do a MP to remove the check? [11:59] with the new update manager it is really hard to say what i got [11:59] Laney: yes [11:59] look in /var/log/dpkg.log [11:59] the list of stuff is completely useless [11:59] webkit migrated about 2 minutes before you mentioned the issue [12:00] so i doubt you have that, unless you're running proposed [12:00] * ogra_ wonders why we show the list at all ... it doesnt really help anymore they way it is [12:00] ogra_, what list? [12:01] seb128, update-manager [12:01] why is it not useful? [12:02] seb128: yeah, except that whoopsie circumvents that check [12:02] pitti, I wonder if that's a good idea... [12:02] because it uses weird cryptic package descriptions i would have to decypher first before knowing what i get [12:02] ogra_, we need to put better descriptions in our control files then... [12:03] well... we're asking him for the name of the software [12:03] the description doesnt say something like "webkit foo bar baz" .... but says "very fast html rendering engine" [12:04] thats absolutely not helpful without the exact package name ... [12:04] (i'm making up the texts, just an example) [12:04] you are too technical, that tools is aiming at being user friendly [12:04] well [12:05] my mom wouldnt bother ... [12:05] i wouldnt show that list at all [12:05] the intend is to list end user applications in the main list [12:05] with their icon [12:05] that works quite well [12:06] the technical items are under base OS [12:06] for the tree apps that get updated along the 200 other packages, yes [12:06] *three [12:06] so just show apps with .desktop files or so [12:07] i think what we show now is confusing [12:09] btw, no webkit in dpkg.log [12:09] k [12:09] and you can reproduce the bug after restarting evo? [12:13] ogra_, http://ubuntuone.com/32sd5Xmv7NIgIry7ESlXir [12:13] ogra_, that looks a fine list to me... [12:14] ogra_, the security view is a bit weird, on a normal upgrade you would have the technical items under "Ubuntu base" [12:15] but the top of the list is nice, archive manager, desktop sharing, firefox web browser, etc [12:15] those are quite useful descriptions/items [12:17] seb128: https://launchpad.net/~bjoern-michaelsen/+archive/libreoffice-quantaltest-20120601/+build/5306216 :( [12:18] seb128: I updated the mdds version to the latest upstream, still happens on the trusty toolchain. I gotta check updating to the latest LO41 is of help otherwise it means digging deeper. [12:18] Sweetshark, can we just build with gcc-4.7 or something? [12:19] seb128: hmm, good point. [12:19] seb128: Ill try. [12:19] Sweetshark, thanks [12:23] seb128, yes, the top looks fine ... further down not so much ... i would exclude libs and all [12:25] how do i build libindicator from bzr into a package? http://paste.ubuntu.com/6529519/ [12:32] bah, webkit [12:33] dbg = 350M, installed 1.5G === alan_g is now known as alan_g|lunch [12:40] qengho, hey [12:40] qengho, do you have any news about the autopkgtest blocking chromium in trusty-proposed? [12:58] ali1234: how did you invoke the build? [12:58] "debuild" [12:58] use bzr bd from bzr-builddeb [12:59] do i have to commit changes for that to work? [12:59] no, but you have to bzr add new files [13:00] noted, thanks === alan_g|lunch is now known as alan_g [14:05] hihi, peeps [14:20] desrt, hey, happy friday! [14:21] Laney, so, I need to do a webkit upload, anything you wanted added there? [14:24] holy crap. it's friday! [14:25] yeah, a good friday so far [14:25] :) [14:25] Laney even managed to finally get webkit to build on all archs, including arm* [14:25] building webkit on arm? [14:26] I'm in the night of Friday... [14:26] JackYu, you shouldn't be on the computer, that's not what friday nights are for! [14:27] ...says seb, who will be here 8 hours from now [14:27] seb128, that's right. I will go to bed soon:) [14:28] desrt: morning! Should g_settings_get() disallow non-copying format strings? It calls g_variant_unref() after all [14:28] or is that a non-issue with all backends? [14:28] larsu: it's a very complicated question, to which the answer is "yes" [14:28] I find the question quite simple. Patch welcome? [14:29] yes please [14:29] the question is complicated by a few things [14:29] okay. This will break at least evince [14:29] because all backends do in fact support this [14:29] but only 99% of the time [14:29] hm? [14:29] if the gvdb gets replaced and reopened then your string could disappear under you if you unreffed the gvariant [14:30] which happens on writes, doesn't it? [14:30] yup [14:30] which don't happen 99% of the time [14:30] also: you have to do another read on the client side as well [14:30] because that's when it checks for the gvdb needing to be reopened [14:30] in theory it's possible to ensure that this isn't happening in your process [14:30] g_settings_get (settings, "&s", &str); g_settings_set (settings, whatever, bar); /* str is dangling */ [14:30] in practice... threads, etc... [14:30] correct? ^ [14:31] no. in this case str would still be OK [14:31] why? Because you still have the old mmap even though the file link is gone` [14:31] yes [14:31] you have still mmaped the unlinked database [14:31] interesting [14:31] so the pointer is still good [14:31] as i said -- it's complicated [14:31] but i agree with your original idea to limit it [14:31] but best avoided... [14:31] yup [14:32] I mean, you can always get 0-copy with g_settings_get_value() [14:32] the other complicating factor: the API for checking format strings for copy-only was added long after g_settings_get() [14:32] precisely. [14:32] I know, I remember when you added it [14:32] also: if you're storing something for which you care about zero-copy in gsettings then you're doing it wrong [14:32] I came across someone doing it today, which is why I started wondering [14:33] ya... so this is an API break [14:33] i recommend doing two patches... first one should throw a critical but not return [14:33] didn't bother you in gmenumodel... [14:33] (same api break) [14:33] seb128: what do you want to upload it for? [14:33] but no [14:33] gmenumodel was very new when i made this break [14:33] gsettings is very old [14:33] fair enough [14:34] I'll do a critical first [14:34] with a note that this won't work much longer :D [14:34] party [14:34] party? [14:35] instead of saying 'awesome' or some other word [14:35] you and your Canadian slang! [14:35] * desrt discovers the hard way that bijiben's owncloud notes storage backend is ... unreliable [14:36] "oh... look at all of these text files in my homedir containing useful information... i should turn those into bijiben notes" [14:36] "okay... now i can delete the originals..." [14:36] it does that?! [14:36] larsu: no. i did that. [14:36] ah. lol. [14:36] I mean, sorry for your data [14:36] "hm. let's open this again.... hey.... why is this empty?" [14:37] larsu: don't be sorry for me. mterry is on my side. [14:37] there's a new tomboy? [14:37] Laney: ish? [14:37] Laney: yes, but it destroys your data [14:37] the local storage backend is working nicely, fwiw [14:37] but uh... owncloud... ya [14:37] don't use that :) [14:38] canonical destroyed my tomboy notes too, don't worry [14:38] Laney: it's actually kinda nice. it looks attractive and is integrated with gnome-shell search [14:38] the reason I went back to editing notes with vi ^^ [14:38] (full text) [14:38] Laney, I've been getting debug infos and pinging people about that software-center Xerror that tops e.u.c since raring and Company has a tentative fix [14:38] ok, well I'd rather you get confirmation in a ppa [14:39] anyone having seen stuff like this (on trusty): http://pastebin.com/x4qyVXsD ? [14:39] Laney, are you able to reproduce? [14:39] no [14:39] Laney, I was going to get confirmation through e.u.c stats [14:39] we know a german man who can though [14:39] at least judging by the frequency he complains about it [14:39] desrt: the faux ruled paper is a bit annoying [14:39] ok; let me throw it in a ppa then [14:39] larsu: i actually like that [14:40] i like that fix [14:40] me too ;-) [14:40] there should be a new point release along shortly so we might not need to upload explicitly for it anwyay [14:40] * desrt likes dejadup for pulling his ass from the fire once again [14:40]