[01:39] hey robert_ancell , how goes the compositor stuff? still having fun? [01:53] jasoncwarner_, yep, still ironing out details [01:53] robert_ancell: progressing the way you envisioned originally? any setbacks? [01:54] jasoncwarner_, just tricky setting everything up right but is working as expected when done right. [01:56] robert_ancell: nice...can't wait to see it :) [01:57] still thinking first cut later this week? [01:57] jasoncwarner_, yep [01:57] cool. thanks, man [02:20] RAOF, any reason why xwayland wouldn't be emitting SIGUSR1? [02:20] RAOF, btw I found my problem, I had a \n on the end of WAYLAND_SOCKET [02:20] None that I can think of? [02:20] Yay! [02:20] You should also be able to listen for the ready signal. [02:20] so X server is starting, but I just get a black screen and lightdm says it's waiting for SIGUSR1 as normal [02:21] But since you already listen for SIGUSR1 that's probably easier. [02:25] robert_ancell: The xwayland server *should* wander through the same codepath that sends SIGUSR1 to the parent process. [02:26] RAOF, I just turned on -retro and don't see the background, so it might well be something else [02:26] Oh, have you called switch_to? It won't display anything until you explicitly switch to the client. [02:27] Not that this should affect it sending SIGUSR1. [02:27] ah, that could be it [02:27] On the basis that explicit is better than implicit :) [02:29] Also that it's hard to manage the transitions between clients if starting a new client automatically displays it. [02:29] RAOF, yeah, that's a problem [02:30] RAOF, can we handle that? [02:30] We do; starting a new client doesn't display it, the display manager needs to explicity switch_to it. [02:30] I'd like them to start rendering to an offscreen buffer, and then I'll switch them over when ready [02:30] Which is why you're not seeing anything with -retro :) [02:31] X is happily rendering away to its offscreen buffer ☺ [02:49] robert_ancell: pong [02:49] desrt, hey, was wondering if quantal should be using d-conf 0.13.0 [02:49] yes [02:50] k [02:50] i have some pretty big changes in the pipe [02:50] will go out with next week's gnome release [02:50] have you been following the branch? [02:50] nope [02:51] http://git.gnome.org/browse/dconf/log/?h=wip/reorg [02:53] RAOF, btw you should add some _exit(EXIT_FAILURE); calls after our exec calls in simple-display-manager - get's confusing if the exec fails otherwise :) [03:28] Good morning [03:40] robert_ancell: is there a reason you're uploading everything to quantal-proposed? [03:41] jbicha, aren't we supposed to? or has the plan changed again [03:42] I think that was just for the Alpha freeze or if stuff is likely to cause issues [04:02] robert_ancell, jbicha: Yeah, quantal-proposed is (still) just for transitions or during freezes; unless something's changed recently there's not yet any auto-migrate system in place. [04:34] RAOF, can you take a quick look at src/seat-weston.c in lp:~robert-ancell/lightdm/weston-compositor and see if anything stands out why I wouldn't be getting the ready signal? [04:35] I can see the greeter running, and the weston log seems to show the ready event being generated [04:35] http://paste.ubuntu.com/1083918 [04:36] and afaict it's doing the same thing your DM does (which I can confirm working here) [04:48] robert_ancell: There doesn't seem to be much of a difference, except that you're passing a stack variable to wl_system_client_add_listener. [04:48] yeah, i tried making it static, no change [04:49] It would almost certainly have crashed if that were the case, yeah. [04:49] RAOF, can I call wl_display_manager_switch_to_client at any time? [04:49] Once the client has started, yes. [04:49] i.e. once I get that signal? [04:50] It's definitely safe any time after you get that signal, yes. [04:50] I'll just check if it's safe sooner, too. [04:50] RAOF, I just connected it up without waiting for the signal and it appears to work... [04:51] but getting some crazy colour effects here [04:52] RAOF, are you busy at the moment? [04:53] I'm wandering through the SRU queue; I can happily interrupt that. [04:53] RAOF, ok, let's get it running on your machine and see if you see the same things [04:53] step 1: Get lp:~robert-ancell/lightdm/weston-compositor [04:53] Done. [04:54] step 2: Build with ./autogen.sh --prefix=`pwd`/install; make ; make install [04:54] step 3: Stop existing lightdm (I think required?) [04:54] Seems likely. [04:55] step 4: get weston.conf from http://paste.ubuntu.com/1083931 [04:55] step 5: sudo ./src/lightdm -c weston.conf [04:55] logs go into install/var/log [04:55] I have to ssh into the machine to stop it with a 'killall lightdm' [04:56] step 5: witness the weirdness [05:00] what I'm seeing is the accumulating over the top of the existing rendering, but it could be a gtk/u-g bug? [05:02] once in a session I'm seeing various rendering bugs (window seems off center, old contents being left behind, flickering etc) [05:04] Aaah. [05:04] I think this might be my bug. [05:04] apart from that, things generally seem to be running [05:05] What appears to be happening is that the *root* window is mad. [05:05] compiz seems to be using more CPU than normal [05:05] madder than normal? [05:05] robert_ancell: more CPU than normal... [05:05] robert_ancell: I didn't know the dial went to 12! [05:06] lifeless, :) [05:06] thats unfair really; but I seem to be a magnet for compiz glitches :) [05:06] robert_ancell: Yeah, that's using llvmpipe. [05:07] Which is also why compiz is rendering oddly. [05:07] RAOF, oh, my 3d has disabled? [05:07] Yus [05:07] Wheee! [05:07] RAOF, did you log into a session ok? [05:07] That root window madness on unity-greeter is FUN! [05:08] it's the new design, don't you know? [05:08] Yeah, logged into a session; that's where I ran glxinfo, to find out that it's llvmpipe. [05:08] I did notice one improvement however - no flicker when logging out and back to the greeter! [05:09] Yes. Also, as I say, a fun burst of colour! [05:09] I like how it flickers with the cursor, and there is a motion blur on the labels [05:10] I also like how the keymap's qerty, rather than dvorak. Makes entering my password that bit much more enjoyable! [05:11] Hah. [05:12] So, that's why there's a little pink band down the right hand side - weston's lying about the screen resolution. It says it's 1360x768; it's really 1366x768 [05:12] oh, that again [05:12] hi folks [05:12] Compiz is officially Not A Fan of llvmpipe :) [05:12] any chance we can have someone look at https://bugs.launchpad.net/ubuntu/+source/gdb/+bug/1006860 ? [05:12] Ubuntu bug 1006860 in gdb "gdb crashes when loading core files (in is_ctor_or_dtor)" [High,Confirmed] [05:12] RAOF: we'll get there [05:13] RAOF, how bad do these issues look> [05:13] ? [05:13] bad = complexity/possibility of fixing [05:13] robert_ancell: Can we get design to sign off on the awesome bloom when returning to unity-greeter from a session? :) [05:14] llvmpipe rather than hardware rendering is probably trivial; set EGL_LOG_LEVEL=info and I'd wager it's stupidly loading egl_gallium+llvmpipe rather than egl_dri2+i965 [05:15] The root window thing... is less familiar to me. This might just be a switch that I need to flick, though. Alternately, I could treat the root window specially. [05:16] The 1360*768 rather thar 1366*768 is something that we'll be fixing in the course of handling RANDRish stuff anyway. [05:17] Fundamentally the problem seems to be that the root window is accumulating drawing, rather than overwriting. [05:17] right [05:17] That shouldn't be terribly hard to fix, once I work out why it's actually happening. [05:18] ok, I'll tidy this stuff up and get it ready for wider testing and work on the screen lock stuff. I think you can safely reproduce everything on your system so I'll await an updated PPA then we can recommend other people start testing when it looks more sane [05:19] RAOF, the other bug I noticed was I couldn't access the auto-hid launcher, not sure what's causing it [05:19] i.e. access by dragging onto it [05:19] moving onto it [05:19] I know! [05:20] That happens because the drag-to-reveal effect requires X getting motion events outside the visible area; weston isn't sending those, because they're nonsensical to clients. [05:20] but pitti's alt+ctrl+t always saves me :) [05:20] still works. [05:21] The launcher will work again once we swap out weston input handling for real xf86-input-*. [05:22] RAOF, oh, and the new API you added for key grabbing - should I use that to connect up alt+ctrl+n? [05:22] Yes [05:22] * robert_ancell takes a list [05:25] You should also hook up ctrl+alt+delete to it. [05:25] Oh! [05:26] robert_ancell: I'm pretty sure llvmpipe is your fault; you need to give us users access to /dev/dri/card0 [05:26] RAOF, how do I do that? [05:27] Normally that'd be handled by consolekit, IIRC. [05:27] RAOF, yeah, so below the DM, I'll chase it down [05:28] RAOF, is that no used in the current X? === robert_ancell_ is now known as robert_ancell [05:29] It is used in the current X; when you set up the ConsoleKit session it adds your user to /dev/dri, /dev/sound (whatever that actually is), etc acls. [05:30] You'll notice that if you try and pull up the sound properties under your weston session you'll just get the dummy output; this is the same problem. [05:33] ok yay perfect timing [05:33] closed my laptop lid, opened it, and X is wedged [05:33] shows background only, can't alt-tab or anything, moving the mouse changes it from an insertion bar to a pointer in a rectangular region of the screen. [05:33] RAOF, oh, I think it's because it's not running all those same shell scripts when run manually [05:33] Any hints as to whats naffed ? [05:33] RAOF, ok, shouldn't be too hard [05:35] lifeless: My guess is that the kernel has forgotten about a VLBANK. [05:35] lifeless: If you switch to a VT, killall -9 compiz (a couple of times, until it sticks), and then DISPLAY=:0 metacity, you should get something backe. [05:35] drm_mode_getfb is whinging about an invalid id [05:36] but then I've switched to a VT :) [05:36] Yeah, it does that. [05:36] woohoo [05:36] thanks [05:36] robert_ancell: Heh. Don't leave your session idle too long; I think I may have missed disabling the weston screensaver :) [05:36] so, how do I get unity back? [05:36] Reboot. [05:37] _serious_ ? [05:37] RAOF, yeah, I noticed that! IMO a system screensaver makes more sense anyway... [05:37] Anything that tries to glXSwapBuffers will hang in the kernel, waiting for a vblank interrupt that never comes. [05:37] RAOF, ok, gtg, thanks for all the hard work in the display layer, it's nice to start to see something actually running [05:37] So, now's not a good time to fire up WoW, either :) [05:38] robert_ancell: Yeah, it's sweet to see something sensible actually running :) [05:38] RAOF: ah, I quit WoW :) [05:40] the intro screen is a great test program though [06:14] RAOF: thanks, and how do I avoid that happening in future ? [06:14] good morning [06:14] lifeless: I *think* this is fixed in the latest Quantal kernel? [06:14] ok, so upgrade :P [06:14] I'm still precising it up [06:14] You're running precise? [06:14] Oh! I wasn't aware that this was a problem on precise. [06:14] tada. [06:15] ubuntu-bug linux! [06:15] I had a hang, RAOF told me to ? [06:15] I think the answer might be either “don't have your screen off for a long¹ time”. [06:15] ¹: For suitably undefined values of “long” [06:15] RAOF: like, 30 seconds ? [06:16] For me it was ~hours. [06:17] If you get the system into that state again, you should find that compiz is hung. Tell the kernel guys that GL applications are locking up when they try to swapbuffers. [06:25] RAOF: bug https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1022819 [06:25] Ubuntu bug 1022819 in linux "X hung, mouse moved and changed shape but no graphics updates" [Undecided,New] [06:44] * TheMuso considers nvidia drivers after having another period of no screen rendering when CPU load is a little higher. [06:44] s/a little higher/high/ [06:52] Good Morning Everyone. [07:00] Why Hello There! [07:04] RAOF: So jumpy so early in the morning. :> === ubuntu is now known as Guest65589 === Guest65589 is now known as NathanStilwell === tkamppeter_ is now known as tkamppeter [07:45] pitti, can you have a look at bug 1022785? The user seems to have lost his /etc/init.d/cups file in his old CUPS version and so is not able to upgrade to a new Upstarting CUPS version. Do we need a change on CUPS? [07:45] Launchpad bug 1022785 in cups "package cups 1.5.0-8ubuntu6 [modified: usr/lib/cups/filter/commandtoescpx usr/lib/cups/filter/commandtopclx usr/lib/cups/filter/imagetopdf usr/lib/cups/filter/imagetops usr/lib/cups/filter/imagetoraster usr/lib/cups/filter/pdftoijs usr/lib/cups/filter/pdftoopvp usr/lib/cups/filter/pdftopdf usr/lib/cups/filter/pdftops usr/lib/cups/filter/pdftoraster usr/lib/cups/filter/pstopdf usr/lib/cups/filter/rastertoescpx [07:46] tkamppeter: easiest is usually to reinstall the package with --force-confmiss [07:46] tkamppeter: conffile handling is not cups specific, so there is little that cups should do there [07:48] pitti, thanks. I answered to that bug now. === zyga is now known as zyga-afk [08:36] hey desktopers [08:36] bonjour seb128 [08:36] hey pitti, how are you? [08:36] didrocks, it's meeting reminder day you didn't remind me :p [08:36] seb128: I'm great, thanks! happily hacking on libudev [08:36] seb128: I was waiting for you to say hi on #ubunut-desktop first :p [08:37] didrocks, ;-) [09:04] jasoncwarner_, kenvandine, TheMuso, cyphermox, RAOF: you guys need to better filter your emails ;-) [09:05] jasoncwarner_, kenvandine, TheMuso, cyphermox, RAOF: you are the ones who didn't reply to my "Do you all receive,filter team emails from launchpad?" email from last week [09:05] * mlankhorst didn't eitiher :D [09:05] ups [09:05] seb128: oh, I didn't respond to that...oops [09:05] mlankhorst, indeed! [09:05] jasoncwarner_, did you get it? ;-) [09:05] I was thinking about that [09:05] it showed up in launchpad.net/contactviaweb [09:05] it's not possible to have group discussions if you use that feature, is it? [09:06] seb128: I did...and marked it "TODO" and then, didn't do it [09:06] jasoncwarner_, ok, good [09:06] but just hit reply and sent a blank mail back [09:06] enjoy\ [09:07] Laney, no, I think it's a one way street ... maybe an alias or group for the team is better after all, still I would like those emails to not land in null ;-) [09:08] mlankhorst, you should have received an email as well, you can reply to it without using the launchpad ui [09:08] yeah it would be good, but probably not for 'official' use, rather so we don't miss random contacts [09:08] for chat a proper mailing list would be better I suppose [09:08] right [09:08] we already have lists [09:08] I don't think we have so much private in team discussions [09:08] I thought that point was that you don't like maintaining a list in your client? [09:09] right, I found the launchpad feature handy to contact all members of a team [09:09] i.e for meeting reminders [09:09] i.e -> e.g === ubuntu is now known as Guest87398 === Guest87398 is now known as Nathan_S [09:40] didrocks, ping [09:40] didrocks, HIDE [09:40] ;-) [09:40] rickspencer3, bonjour [09:40] didrocks, Nathan_S is trying to install quickly, but apt can't find it [09:40] seb128, haha [09:40] rickspencer3: pong ;) [09:40] rickspencer3: on precise? [09:40] universe is enabled? [09:40] or quantal? [09:40] yeah, universe should be enabled :) [09:41] didrocks, Nathan_S is on precise, and universe is enabled (according to his software-sources_ box [09:41] did he refresh the repository? [09:41] didrocks, he's on a stock 12.04 install running from a USB key [09:41] I guess he could try: [09:41] sudo apt-get update [09:42] Nathan_S, ^ [09:42] rickspencer3: I'm pretty sure the repos has to be refreshed (apt-get update) [09:42] btw, hi Nathan_S ;) [09:43] Hi rick :) its updating as we speak [09:43] thanks guys :) [09:43] didrocks, Nathan_S is going to spend some time this week going through all the quickly tutorials [09:43] hey Nathan_S ;) [09:43] sweet [09:43] it would help if quickly installed, though :) [09:43] can be somewhat useful, yeah [09:43] :) [09:43] update done, searching for quickly :) [09:46] rickspencer3: waow, I never looked at the reviews in software-center [09:46] it still didn't turn up on software center, but i typed in the command line you told me and it is working away [09:46] the Quickly ones are really good :) [09:46] didrocks, are they good? [09:46] nice [09:46] cool! [09:46] Nathan_S: I think you need to look at "technical items" [09:46] * rickspencer3 looks [09:46] when you look for Quickly === zyga-afk is now known as zyga [09:47] technical items are shown and it isn't there :/ but it is installing after typing it in on terminal [09:48] Aha it has appeared on software center! [09:49] hum, weird, I'm pretty sure it's lost in the technical items area if you click on it :) [09:50] Nathan_S, can you please make sure to note that Quickly didn't show up before you jumped through hoops? [09:50] it's good to know that the web site does now work on a stock install! [09:51] I thought that too, but originally at the bottom it said "show 66 technical items" and now it says "show all 357 technical items" [09:51] and quickly is showing up without showing technical items [09:52] rickspencer3, Yep, its already in the blog [09:52] if your search exactly matches the name of a technical item then it shows up without having to check the box [09:57] mvo_, hey, daily aptdaemon SRU verification ping, 25 days and counting ;-) [10:19] * pitti sighs at desktop-devel-list [10:22] please add me to the desktop team ping ;-) [10:39] Laney, hey [10:39] Laney, done! congrats ;-) [10:41] awesome, cheers! [10:45] Laney, yw, I still have g-s-d to review on my list for today but feel free to merge that in the official vcs with a note in the changelog of what is needed before upload [10:45] Laney, we already have 3.5 ongoing work in the vcs anyway so it just makes sense to keep working there [10:48] pitti, lot of ddl discussion around those ibus changes... [11:15] seb128: yeah, but with a rather harsh tone [11:16] seb128: meh, indeed === MacSlow is now known as MacSlow|lunch [11:30] seb128: Like jasoncwarner_, I got it, went “Ok, I need to read & respond” and then didn't. [11:30] if it's taking less than 5 minutes, do it now :) [11:30] RAOF, ok, good ;-) [11:31] mvo_, hey, ping, daily aptdaemon testing reminder [11:31] didrocks: Yeah, but... mail! :) [11:31] * seb128 hides from mvo_ [11:31] pitti, never had to deal with Bastien before? ;-) [11:31] RAOF: heh, I do that generally to not being depressed seeing the number of answers I have to do every day :) [11:32] there is 3 approches to email, one is to the brave and deal with them as they come, the other options are to just ignore them all or just delete them as they come ;-) [11:33] * RAOF is a fan of “have an elaborate series of filters, dropping everything thats ‘important’ in a single folder and farming everything else off into their own special boxes that can be perused when the mood strikes. [11:34] RAOF, right, the point of that email was to check if people have the right pipes to have those emails landing in a visible box ;-) [11:34] seb128: I did, but it still stings me when discussions go bad like that [11:34] pitti, yeah, ddl can be depressing like that :-( [11:35] pitti, a number of people unsubscribed from ddl due to those discussions, not sure how useful the list is still... [11:35] going to /dev/null for me because of that :/ [11:36] well so far treating everything as launchpad as spam was only wrong once [11:36] ;) [11:37] s/as/from/ [11:57] happy tuesday [12:14] (seems nobody want to get into desrt's trap today ;)) [12:14] didrocks: hey! [12:14] hey desrt! I'm alreay far far away :) [12:18] didrocks, I got the guy in /ignore, you never know [12:19] seb128: i wanted to talk to you [12:19] seb128: i should be bumping the soname of libdconf, right? [12:20] there have been API removals and changes [12:20] didrocks, btw if he ever comes to ask a question that we should reply to please can you answer ;-) [12:20] hm. you're good at this game :) [12:21] :-p [12:21] desrt, hey ;-) how are you? [12:21] feeling a bit better now that you're talking to me :p [12:21] * seb128 hugs desrt [12:21] seb128: ah see, you felt into the trap! [12:21] * desrt has no work to assign today [12:21] promise? [12:21] yes :) [12:22] * didrocks hugs desrt then :) [12:22] "oh btw…" [12:22] didrocks: you know what my standing request is. no need to repeat it :) [12:22] heh, indeed [12:22] blame seb128, he gave me even more work this morning :p [12:23] seb128: so like... i guess there are almost nobody outside of the dconf package itself using libdconf [12:23] desrt, did you manage to get dconf ng to stop whipping user keys? ;-) [12:23] seb128: ya :) [12:23] seb128: but here's the deal.... [12:23] i think you have libdconf and libdconf-dbus-1 in the same package [12:23] oh. you don't. [12:23] desrt, libdconf ... no, I know of some users of the command line utility, not of the lib [12:23] okay. this is easy then [12:24] so does it matter if i increase the soname? [12:24] no [12:24] ok [12:24] may as well keep it at zero then [12:24] less packaging shuffle that way :) [12:24] oh, you mean it this way [12:24] hum, you need to bump it :) [12:24] I though you were asking if an soname increment would create work or issues [12:25] please increment the number if you break abi [12:25] so, libdconf (old) can talk to newer libdconf-dbus? [12:25] you never know, and it's not a big issue [12:26] didrocks: they are unrelated [12:26] libdconf is implemented with GDBus and libdconf-dbus-1 is implemented with libdbus-1 [12:26] oh [12:26] they share some code in the implementation... but it's private symbols [12:26] so my question is: libdconf (old) can talk to the newer service? [12:26] yes. it can. [12:27] although i'm not happy about it :) [12:27] it seems a bit weird to have 2 libs to be honest ;-) [12:27] well, I know that for your proudness, keeping it to a 0 soname was a goal :p [12:27] and the reverse is not true [12:27] seb128: there are 3, in fact [12:28] seb128: if you count the GSettingsBackend GIOModule as a library [12:28] seb128: in any case, talk to the powers that be... apparently it was unacceptable to use GDBus in unity-2d === Ursinha` is now known as Ursinha [12:28] desrt, wth [12:29] desrt, I'm sure unity-2d uses glib ;-) [12:29] it does [12:29] but it uses QtDBus [12:29] which uses libdbus-1 [12:29] oh, I see... [12:29] so in order to share 1 dbus connection, dconf had to use libdbus-1 as well [12:30] well, anyway, this problem will be sooner or later behind us :) [12:30] (unfortunatly later…) [12:30] i don't think so [12:30] i imagine that since we have QML as our developer platform dconf-qt will continue to be used [12:31] right, but it's not a real issue in practice [12:31] the dbus-1 code is actually in a bit of a bad state in dconf [12:31] desrt, to go back to the topic, please bump the soname, it doesn't cost much and it's theorically clean and avoid any issue we might not have though about [12:31] and by bad state i mean 'exactly the same state as it was before' [12:31] * desrt has completely avoided touching it at all for fear of breaking it and no real way of testing it [12:32] seb128: ah. okay. [12:32] seb128: my main concern was that i thought that both libs were in the same package, which is not the case [12:32] because the dbus-1 would have 0 and the client would be 1 [12:32] that's not an issue either [12:32] so what would you call the package? [12:32] we have some packages like that [12:33] we would have increased the name to 1 [12:33] it's less clean, but in practice what matters is to have a rename when the abi is made incompatible [12:33] and then later when i break the libdconfdbus-1 API and bump its soname to 1? [12:33] you rename it again from 1 to... 1? [12:33] when those happen usually debian just put each in their binary then [12:33] which is already the case here... [12:33] ya. that's what i figured you'd have to do [12:33] yup [12:33] so no problem :D [12:34] indeed ;-) [12:34] btw: i remember that vuntz did something... [12:34] perhaps you remember it too :) [12:34] vuntz did something? really? :) [12:34] when? ;) [12:34] a long time ago :) [12:34] heh [12:34] gnome-panel has a libdconf depend, i think [12:34] * desrt checks if that is still the case [12:35] desrt, he was not using the command line utility? [12:35] yup. [12:35] nope. [12:35] desrt, the profiles are handled using dconf [12:35] but I though that was the binary [12:36] ./gnome-panel/panel-layout.c: panel_dconf_write_sync (key, variant, NULL); [12:36] looks quite like an API call to me [12:36] oh. wait. [12:36] ahah [12:36] indeed === MacSlow|lunch is now known as MacSlow [12:37] using dconf_client_write() [12:37] no... not so lucky [12:37] yup [12:37] this will need to be changed [12:37] well, good that the soname change ;-) [12:37] you just need to track vuntz to update his code [12:37] * desrt guesses it will be faster to do the changes himself [12:38] they are quite minor [12:38] in this case [12:38] you are insulting vuntz there! ;-) [12:39] * desrt wonders what the shelf life of gnome-panel is [12:39] speaking of shelf-life of soon-to-be-abandoned projects... [12:39] didrocks: are you basing your migration stuff on the migration stuff in gconf? [12:40] desrt: indeed, adding tests and so on [12:40] desrt: when I'll have time to start it :) [12:40] do you plan to push upstream? [12:40]