[00:01] robert_ancell and htorque that seemed to make a huge diff [00:01] http://paste.ubuntu.com/694113/ u-g log [00:01] I still had no mouse movement at the beginning...but that was for a second or two (which I think was X catching up) [00:09] This is weird: bug 855171 [00:09] Launchpad bug 855171 in ca-certificates "libnss3.so went missing after upgrade" [Undecided,New] https://launchpad.net/bugs/855171 [00:09] whoops, wrong channel [00:36] jasoncwarner_, excellent, thanks for confirming that [01:33] hmm [01:35] would adding GNOME to NotShowIn cover just the Gnome Shell cases, or also break fallback and others? :) [01:38] GNOME is Shell & Fallback [01:41] jbicha: yeah, that's the problem probably ;) [01:41] going to test this now [01:42] like gnome-datetime-panel.desktop shows in Fallback g-c-c but it's OnlyShowIn=Gnome; [01:43] echo $XDG_CURRENT_DESKTOP [01:45] oh, I know about all of this, just "saying out loud" that I'm pretty much screwed either way [01:46] jbicha: see https://bugs.launchpad.net/ubuntu/+source/network-manager-applet/+bug/852961 :) [01:46] Ubuntu bug 852961 in network-manager-applet "nm-applet is wrongly started with Gnome Shell" [Undecided,New] [01:47] I can't add it to NotShowIn without breaking something, and if I don't add it, something's still going to be broken [01:47] but I think Shell > fallback; so I'll add it. [01:48] but I don't have nm-applet in my Shell session, just Fallback or Unity [01:48] cyphermox: shell isn't shipped by default by a derivative... [01:48] really? [01:48] also, really? [01:48] gah [01:48] stgraber: so edubuntu ships fallback? [01:48] cyphermox: fallback is an option at install time in edubuntu [01:48] mkay [01:48] my notifications area is blank on new start & I have the normal Shell network menu [01:49] jbicha: that's weird [01:49] why would this bug be open then? [01:49] perhaps he's using an old shell version [01:49] maybe [01:50] alright, thanks for clearing that up [01:55] oh wait, he sees notifications in both shell style & an uglier style when the network disconnects for instance [01:55] I can duplicate that [02:00] oh [02:00] well then yeah, this is because nm-applet is getting started [02:00] fixing this is going to be painful, because we're obviously limited by what's being run [02:00] and shell / fallback sharing the same desktop value doesn't help [02:01] I think gnome-shell uses nm-applet, just with a different frontend [02:01] jbicha: no [02:01] it uses its own frontend in js [02:01] also the latest gnome-shell (not in the archives yet) doesn't show network notifications for connections/disconnections [02:01] mkay [02:01] so it's at least visually fixed with that [02:02] urgh [02:02] is that going to be shipped? [02:02] just waiting for GNOME to get around to releasing the new gjs which gnome-shell silently depends on [02:03] * cyphermox cries [02:03] I just won't touch this and sleep on the issue [02:03] they just assume that you're going to be using the latest gjs & mutter & don't bother updating the configure.ac to reflect that [02:20] cyphermox: i thought i remembered dcbw saying that nm-applet didn't actually provide ui with shell these days [02:21] broder that's what I remember too [02:21] nm-applet shouldn't be running in shell, that's most likely the problem because it currently gets started in /etc/xdg/autostart [02:21] cyphermox: no, no - i thought he said it runs but only does...something minor and invisible [02:22] broder: maybe [02:23] AIUI it's just used for the wireless/mobile dialogs [02:23] something that might be fixed by the new n-m-applet package with libnm-gtk split out [02:23] cyphermox: http://paste.ubuntu.com/694185/ - from about a week ago [02:24] right [02:24] I'll need to setup something with gnome-shell and mess around with it to see [02:24] oh, it sounds like maybe the only problem is notifications? [02:25] broder: yes I believe so [02:26] sounds like it now [02:26] jbicha: I just updated the bug, could you try figuring out what sends the notifcation that looks wrong? [02:27] or if they have a different text, just a screenshot of each may be enough :) [02:27] we don't patch how nm-applet does its notifications - how would this not be broken anywhere that's using gnome-shell? [02:28] I expect that Fedora would have the same problem [02:29] err...wait, i was looking at n-m, not nm-applet :) [02:30] broder: actually, I think it was lightly patched, but I don't think it would affect this [02:30] it might be caused by differences in the notification themes, I've seen something getting broken in notifications if I install xubuntu-desktop on top of ubuntu-desktop, notifications start looking different, getting buttons, etc. [02:31] is gnome-shell a notification daemon? i guess...if notify-osd grabbed the dbus name, but gnome-shell short-circuited notifications internally to keep them from hitting dbus... [02:32] when I try to connect to a random encrypted wifi network, I get 2 popups to enter the password, one from gs, one from nm-applet [02:33] yuck [02:33] not sure if the second one is just because I canceled the first one though [02:33] dcbw said that should happen around gnome 3.2 or so - is that what we're at at this point? [02:33] pretty much [02:34] "should" happen? [02:34] we're on 3.1.90 and current is 3.1.92 though [02:35] well...gnome-shell definitely looks like it ships a secret agent [02:36] (look at src/shell-network-agent.{h,c} and js/ui/networkAgent.js) [02:37] cyphermox: getting screenshots is a bit tricky since this is my web connection :) [02:37] I'm going to reboot into Fedora [02:37] ok [02:37] broder: jbicha: I really should be going to bed [02:38] ok, it's not a big issue anyway [02:38] jbicha: does gnome-shell set any useful environment variables so we could patch nm-applet to check those and exit immediately if they're set? [02:38] you want to ping me if you find something ? :) [02:38] you guys seem to be getting more progress than I do [02:38] cyphermox: it looks to me like nm-applet just shouldn't be running under gnome-shell anymore [02:39] I think you could sniff dbus to check if you're in gshell or not [02:40] broder: I kind of got that feeling but I think you still need it for some dialogs at least [02:41] cyphermox: does nm-applet (not nm-connection-editor) do anything besides secret prompts? [02:41] broder: that no, I don't think so [02:42] cyphermox: then just have nm-applet exit(0) if it's running under gnome-shell...however you determine that [02:42] not sure how the more complex wifi connections are done [02:43] well, i've got an oneiric machine, and infrastructure for 802.1x and wpa enterprise , so i can try to do some experimentation at some point [02:44] but from my quick skim of the gnome-shell code, i see code for 802.1x, PPPoE, and modem connections [02:44] only thing i don't see offhand is vpn [02:45] I'll look again in the morning === eeejay is now known as eeejay_is_afk [02:50] Fedora 16 doesn't show notifications for nm-applet so they only have the enter wifi password double dialog part of the bug [02:59] ok [03:00] maybe it's caused by one of those old patches we carry and I keep having to fix [03:02] well I think we want connection notifications in Unity though [03:09] yeah [03:11] jbicha: which still means there is no easy fix, I'll setup gnome-shell tomorrow on my system and try to make sense of what sends what [08:03] mvo: morning - I need you help on some popcon related stuff [08:03] mvo: how is the #stars computed from the popcon rating? [08:04] digging through the S-C code only confused me more :-) [08:04] my guess is that we normalize the score with the "max_popcon_desktop" metadata from the xapian index and multiply by 5... but I can't find that calculation anywhere [08:06] also, I am not sure whether that makes sense at all :-) [08:07] hey [08:08] bonjour seb128 [08:08] hey pitti, how are you? [08:09] seb128: furiously debugging ubiquity problems :) [08:09] quite well, thanks [08:09] just really concerned about the release time now [08:09] pitti, "fun" ;-) [08:09] is it worth testing the installer? [08:09] or is it "known buggy"? [08:09] it crashes if you are offline [08:10] should work fine, just look a bit ugly if you are online and don't use the live session [08:10] so, still worth testing if you want to [08:10] kamstrup: *pfff* ;) we don't use the popcon for the stars, we use the ratings data [08:10] kamstrup: what is the use-case, do you want to show it in unity? [08:10] mvo: lol, that explains my inability to find it :-D [08:10] pitti, let's see, I usually use the 10v as a test box and the wifi doesn't work out of the box [08:11] mvo: the dash allows you to filter by the S-C rating [08:11] mvo: so it's not in the xapian index? [08:11] seb128: right, it'd fail there, unless you have an ethernet cable plugged in [08:11] I did a test install on that one yesterday, with success [08:12] mvo: hit super-a and unfold the "Filter results" expander, and you'll see what I mean [08:12] kamstrup: no, because its so dynamic. we store it currently in a extra file [08:13] kamstrup: but that sounds like that needs some re-thinking [08:13] mvo: a fudge... [08:13] mvo: yeah, probably means we need to pull that feature from the dash... [08:13] oh, everyone back :) [08:13] kamstrup: aha, I was totally unaware of this feature in the dash [08:13] there were only 20 people or so when I woke up, I guess freenode had problems? [08:14] mvo: yeah, it's not exactly a prominent place, hidden away inside the expander [08:14] rodrigo_: yes, it did; should be fixed now, though [08:15] kamstrup: so we currently use a xapian.KeyMaker and throw that as the sorter to the db [08:15] robert_ancell, hey [08:15] robert_ancell, what are you still doing online? ;-) [08:15] seb128, hey, just logged on to bug you... [08:15] mvo: ah, so a KeyMaker that pulls the ratings from the file in order to generate the collation key? [08:15] robert_ancell, oh great, it has been a while I didn't catch you (part due to the fact I was on holidays ;-) [08:15] kamstrup: and the review stats (star rating plus wilson-score based "score" of the ratings) are stored in a pythin pickle [08:15] kamstrup: yeah [08:15] robert_ancell, what's up? [08:16] seb128, can you try / find victims to try lightdm + unity-greeter master today and check they're good? I've been using them today, and I think everything's ok, but being this close to release a second opinion would be good [08:16] mvo: hehe, a pickle, that puts me in a pickle (wow, worst pun ever) [08:16] kamstrup: now of course we can make this a c++ friendlier format for you, thats easy [08:16] kamstrup: lol, indeed [08:16] robert_ancell, I'm happy to test those [08:16] robert_ancell, did you see the translation bug I commented on? [08:16] kamstrup: if this software stuff does not work out, just do standup commedy ;) [08:17] seb128, there's one remaining bug 845549 which I had a solution for but a commenter says it should be done better with the alternatives. So could you follow up on those changes then undo mine if they're better [08:17] Launchpad bug 845549 in lightdm "Do not ship /etc/lightdm/lightdm-gtk-greeter.conf" [High,Fix committed] https://launchpad.net/bugs/845549 [08:17] mvo: out only option is prettymuch json I guess... but I don'tknow if that kills performance... I gues it'll be a very large file [08:17] robert_ancell, ok [08:18] kamstrup: would a libdb format work for you too? [08:18] seb128, yes, I fixed that. it was gettext not being initialized correctly [08:18] robert_ancell, great [08:18] mvo: libdb? you mean bdb, or? [08:18] robert_ancell, ok, I will test both trunks, review open bugs and drop you an email at the end of my day [08:18] seb128, there will still be bugs with PAM messages not being localized, but they're more unlikely to occur [08:18] kamstrup: yeah [08:19] seb128, thanks! [08:19] robert_ancell, yw, great job btw, lightdm really works great for me ;-) [08:19] mvo: do we ship a bdb lib on the cd? [08:19] well the current version does, let's see trunk :p [08:19] seb128, :) [08:20] seb128, There's still more bugs cropping up, but I think we're on top of the most important ones. [08:20] kamstrup: need to check, I can also play a bit with json. alternatively I could provide you a dbus service [08:20] kamstrup: but that does not is speedy enough I guess [08:20] mvo: hmm, we can't depend on a running s-c [08:20] robert_ancell, yeah, I've not seen a lot of "important to fix for oneiric" ones, out of the ones assigned to you [08:21] robert_ancell, I will review the bugs today though just to be sure [08:21] mvo: json is easy, but scares me because I've seen how slow it is in gwibber [08:21] kamstrup: it would be something extra, not part of main s-c, but its probably a bad idea even then [08:21] kamstrup: yeah, exactly [08:21] mvo: it takes >1s 100% CPU churn to parse ~1k messages in gwibber [08:21] kamstrup: s-c is already pretty slow to startup, I don't want to make it worse :/ [08:21] granted, that's some complex json, but still [08:22] mvo: lol, right :-) [08:22] seb128, thanks. there will be some things that will be annoying, but I guess we can put in the release notes "use gdm if this particular feature is important" (assuming it works in GDM) [08:22] mvo: a mmap()able db format should make it faster than a pickle for S-C as well I guess [08:23] kamstrup: let me check the CD to see if dbm might work, we just need to be careful with the layout so that there are no interoperatable issue. but I think that should be ok as the data is really just a mapping of str -> (int, int, int) [08:23] mvo: yeah, sounds fit for a keystore [08:24] robert_ancell, right [08:24] kamstrup: I think for next cycle we need to talk some more and we need to provide you proper APIs [08:26] mvo: I have libdb5.1 providing /usr/lib/i386-linux-gnu/libdb-5.1.so here... i don't know if that's on the cd... [08:26] but i guess so [08:28] jasoncwarner_, so bug 828112 is fixed? Using packages from the standard repos? [08:28] Launchpad bug 828112 in unity-greeter "Password field feedback slow at times" [High,Triaged] https://launchpad.net/bugs/828112 [08:28] kamstrup: both 4.8 and 5.1 are installed for me on the live-cd so we should be fine [08:29] robert_ancell, the updates I mentioned there are not in the standard repo but in the ubuntu-desktop ppa [08:29] seb128, is there a bug I can merge this with? [08:30] robert_ancell: I had to manually create an /etc/X11/xorg.conf file to stop the HDMI probe [08:32] robert_ancell: right, we need a wrapper around gnome-session for the lightdm guest account that we can hook the AA profile to [08:32] pitti, yeah, annoying! [08:32] robert_ancell: but that can be a simple two-liner (she-bang and exec gnome-session "$@"), as in gdm [08:33] jjardon: good morning [08:33] jjardon: out of interest, what's the advantage of i-power talking to gnome-settings-daemon instead of upower directly? [08:33] pitti, right, but there's no concept of sticking one in in lightdm, so we need to work out the right place to do it [08:34] mvo: ok, cool. I've no idea, but I assume we should use 5.1... [08:37] jasoncwarner_, bug #855350 is the bug dholbach opened about the output probbing issue [08:37] Launchpad bug 855350 in xorg "long reprobes slow down boot/login" [Undecided,New] https://launchpad.net/bugs/855350 [08:38] jasoncwarner_, if you want to comment on it [08:42] pitti: gnome-settings-daemon parse all the devices info for me, so I get a list of available devices with all their info only with a dbus call [08:42] jjardon: that makes it slightly dependent on GNOME, though? [08:43] jjardon: I just noticed because I'm currently debugging bug 854717, and it's indicator-power which dbus-launches g-s-d [08:43] Launchpad bug 854717 in ubiquity "Broken panel icons and dialog style during ubiquity-dm and OEM install/final user configuration" [Critical,Confirmed] https://launchpad.net/bugs/854717 [08:43] (a broken g-s-d) [08:43] jjardon: that's not directly i-power's fault, of course, but it made me wonder why it uses g-s-d [08:44] RAOF, Sarvatt - is the eDP slowness likely to affect changing the screen brightness via xrandr? [08:44] i noticed this morning that the new kernel fixes a huge lag i was also seeing when adjusting the brightnes [08:44] s [08:45] unless that's just my imagination ;) === chrisccoulson_ is now known as chrisccoulson [08:45] chrisccoulson_: if xrandr --verbose contains a "BACKLIGHT", then we are using xrandr calls for the brightness keys, yes [08:45] the old gnome-power-manager had a fallback to poke the values into the ACPI files in /proc/, but I guess that's gone these days [08:46] we now rely on the XBACKLIGHT extension, which I believe goes through XRandR [08:46] (I have no idea where this ends up at the kernel level, though) [08:46] pitti - oh, no BACKLIGHT on my machine here :/ [08:46] pitti: well, gsd is a core part of GNOME, I really do not think that you can run a Unity session without it. Also, I'd have to duplicate a lot of code that Its already in gsd, so I chooosed to use gsd directly [08:46] jjardon: I mean for XFCE and the like [08:47] jjardon: I was just wondering if g-s-d's power module provides any functionality which upower doesn't, or whether it just offers a more convenient API [08:47] pitti: XFCE has is own power indicator [08:47] because if g-s-d is adding functionality, I'd rather move that down to upower [08:49] seb128 and didrocks, can I reschedule our talks tonight? have a conflicting music lens discussion at the same time... [08:49] jasoncwarner_: sure [08:49] jasoncwarner_, "no" ;-) [08:51] pitti: gsd uses upower to get all the info about power devices. It parses it and you get all the info with a dbus call. So, yes, It's a much more convenience api and you save a lot of code [08:53] jjardon: fun, I thought it would be a lot more convenient to use e. g. up_client_get_devices() and teh libupower-glib API than fiddling with dbus calls yourself [08:53] jjardon: so if libupower-glib sohuld have an API that you would like to have and is missing, I'm still interested in hearing about it; we can just add it [08:53] hmmm...is mumble not working for anyone else right now? setting up on my recent reinstall and it hangs configuring it...have to force quit it... [08:54] jasoncwarner_: WFM [08:54] jasoncwarner_, works here [08:55] kamstrup: proof of concept lp:~mvo/software-center/review-stats-as-dbm-for-unity-with-best-regards [08:55] jjardon: we just noticed that we can't use indicator-power in ubiquity-dm for that reason (see my last comment on bug 854717) [08:55] Launchpad bug 854717 in ubiquity "Broken panel icons and dialog style during ubiquity-dm and OEM install/final user configuration" [Critical,Confirmed] https://launchpad.net/bugs/854717 [08:55] pitti: sure. have to go now, I'll follow the discussion later ;) [08:55] kamstrup: exports (iii) ratings_average, ratings_total, dampended_rating, the last one is what you want to use as the key, its calculated using a wilson score on the ratings spread [08:56] kamstrup: well, I guess not necssarily, depends on your UI needs, but that is what we use [08:58] kamstrup: I guess it would be nice to export it to a tool for you too so that you can trigger updates on the stats yourself instead of relying on s-c doing it [09:00] mvo: I don't think generally the dash should cause changes to local data [09:00] mvo: it's for lookups and access [09:01] 'k [09:01] mvo: usage logging hapens indirectly via zeitgeist then [09:01] kamstrup: if you need the review spread (10 x 1 star, 4 x 2 stars) etc, just let me know, that can be added as well [09:02] mvo: ok - not for O definitely [09:02] kamstrup: and we might need a bugreport if that should make it into the archive for reference for the release team, uploads are hand-approved currently [09:03] mvo: ok... [09:03] mvo: what format do you want to use for the payload? [09:03] mvo: A GVariant "(iii)" would be extraordinarily handy for me :-) [09:04] dunno how easy they are to serialize in Python though... lemme see [09:05] kamstrup: I already use python struct [09:06] kamstrup: so you should be able to just use something like struct { int , int, int } in your c(++) code [09:06] mvo: oh? you sure? [09:07] kamstrup: well, that is what the python struct module is all about, I haven't written a test C client for this yet, but it should just work(tm) [09:08] kamstrup: let me quickly verify this [09:13] pitti, or someone else, when is the new pygobject supposed to enter the archives? [09:14] davidcalle: 3.0.0? after b2 [09:15] pitti, thanks, I'm worried about it. Looks like it makes python lenses segfault. [09:15] seems like a bug for kamstrup ;-) [09:16] davidcalle: that's why I didn't push it into b2 yet; I tested it with apport, jockey, software-center etc., but not with everything yet; I did notice a regression in gtimelog, but I got that fixed now [09:16] davidcalle: you can test it from the PPA, maybe you can get a stack trace and file a bug? [09:17] pitti, I'm on the PPA. I will file a bug, thanks. === ember` is now known as ember [09:27] davidcalle: you have the ppa url, I can debug the crasher [09:28] kamstrup: add-apt-repository ppa:ubuntu-desktop/ppa [09:28] pitti: thanks [09:28] kamstrup: you'll also get a faster startup from that, for bonus :) [09:30] jasoncwarner_: want to skip our 1-on-1, too? [09:30] kamstrup, thanks :) Was learning how to get stack traces. [09:31] * didrocks reboots [09:35] pitti: I'm good for a bit if you have a couple of minutes...want to catch up on a couple quick things... [09:36] i got mumble working as well (took a reboot) [09:39] waow, that was… weird [09:39] libnss3 package installed, but no /usr/lib/i386-linux-gnu/libnss3.so [09:40] network-manager and nm-applet didn't appreciate :) [09:41] didrocks: see topic in #ubuntu-devel [09:41] geser: argh, ok… thanks for the pointer [09:45] pitti, hi. If an application wants to discourage you from logging out or shutting down while it finishes a task (e.g. burning a CD, defragmenting a disk, installing a package), what's the appropriate way to do that? Is it a ConsoleKit API? (I tried reading the CK docs but didn't understand them) [09:45] mpt - gnome-session [09:45] that has an inhibit API [09:46] (for logging out, anyway) [09:46] chrisccoulson, ah, ? [09:46] didrocks, mhr [09:46] for shutting down, i'm not sure if there is an inhibit API [09:46] didrocks, mhr3 and cyphermox got that bug as well this week [09:47] didrocks, cyphermox tried to figure why the file had been cleaned, not sure they figured it though [09:47] yeah, seems ca-certificates being a little bit more too drastic in his pick :) [09:47] davidcalle: don't stop then :-) [09:47] seb128: what geser told, topic of #ubuntu-devel [09:47] ok [09:47] davidcalle: did you have a bug report? [09:47] ca-certificates eats system libraries, DO NOT UPDATE (bug #855171) [09:47] Launchpad bug 855171 in nss "libnss3.so went missing after upgrade" [Undecided,Confirmed] https://launchpad.net/bugs/855171 [09:47] ;-) [09:47] I got lucky it didn't eat mine [09:48] indeed, well, I just lost 3 minutes… [09:48] fortunatly, I had it still in cache and --reinstall worked [09:48] kamstrup, I'm filing one in a few minutes. [09:48] otherwise, I would have to hunt for an usb key, and that would have took more time :) [09:49] thanks chrisccoulson [09:51] stupid tb, stop marking the selected message as read when entering a mailbox! ;-) [09:52] kamstrup, https://bugs.launchpad.net/unity/+bug/855402 [09:52] Ubuntu bug 855402 in unity "Python lenses segfault with pygobject3" [Undecided,New] [09:53] davidcalle: tanks [09:53] thanks [09:53] seb128: +1 [10:04] chrisccoulson, your g-s-d package is indeed much quicker [10:05] chrisccoulson, I was looking though at making those patches upstreamable, and I think it would be better, for the share-the-gnome-rr-screen one, to have some code in gnome-settings-daemon/gnome-settings-daemon/gnome-settings-screen.[ch] or something [10:05] rodrigo_, yeah, i planned to do it that way at some point [10:05] that just returns a shared instance of GnomeRRScreen, instead of using the property on the GdkScreen [10:05] chrisccoulson, ok, cool [10:05] yeah, that would probably be better actually [10:05] chrisccoulson, I can do it if you want [10:06] rodrigo_, sure, feel free :) [10:06] i'll push the current branch to bzr in a moment [10:06] about the media keys one, not 100% sure, so better to file a bug for hadess to review it [10:06] rodrigo_, did you see my notes on https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis ? [10:07] chrisccoulson, no, looking [10:08] chrisccoulson, yes, we can have the patches as they are for oneiric, I'll prepare the patch for git master [10:08] thanks :) [10:09] chrisccoulson: firefox sometimes opens tabs on a wrong window here, as if the focus is not on the window where I want it to be. known issue or not? [10:10] tjaalton, no, i've not seen that before. is this using compiz? [10:10] actually, i'm not sure if that would matter [10:10] chrisccoulson: yes, and happens only with fullscreen windows, it seems [10:10] how are you opening the new tab? [10:10] ctrl-t [10:11] that should open in the window that gets the event :/ [10:11] well, all the kbd shortcuts end up in the wrong window, so ctrl-w might close tabs behind my back :) [10:11] i've seen focus issues with compiz like this ;) [10:12] eg, bug 838625 [10:12] chrisccoulson: speaking of which, did you upgrade using the ppa? [10:12] Launchpad bug 838625 in compiz "Unpredictable focusing after switching viewports with the keyboard" [Undecided,Confirmed] https://launchpad.net/bugs/838625 [10:12] didrocks, yeah. i didn't see any issues afterwards [10:12] great :-) [10:14] chrisccoulson: oh, if it's related to the stacking bug then I'll just wait for the updates [10:14] tjaalton: you can try the ubuntu-desktop ppa [10:14] happens with fullscreen windows that are on separate viewports, one on top of the other [10:14] didrocks: ppa:ubuntu-desktop? [10:14] tjaalton: ppa:ubuntu-desktop/ppa indeed :) [10:15] didrocks: thanks, adding [10:15] tjaalton: keep us in touch :) [10:15] i should try the 3d session again to see if the stacking fixes fix the focus issues i was seeing too [10:16] chrisccoulson: oh, you didn't try it? wasn't the purpose of installing the ppa :-) [10:16] it's possible to fix the focus by going to another viewport with a non-fullscreen ffox window [10:16] didrocks, yeah, i didn't notice any stacking issues (like, mouse clicks going to the wrong windows and the dash appearing underneath everything else) [10:17] but i didn't use it for long enough to see if there were other focus issues. i wasn't sure if that was all related to the stacking bugs === vuntz_ is now known as vuntz [10:26] rodrigo_, i've pushed the gsd changes to bzr btw [10:27] remember also that there is a gnome-desktop change [10:27] chrisccoulson, ok [10:28] chrisccoulson, for the gnome-desktop change, I'll let you do the upstreaming :) [10:28] rodrigo_, sure, no problem [10:31] didrocks, can we drop your gwibber upload to the desktop ppa? it's the same version than oneiric and 2 weeks old [10:32] bah, tb [10:32] "Thunderbird is already running, but is not responding. To open a new window, you must first close the existing Thunderbird process, or restart your system." [10:33] seb128: yes [10:33] didrocks, thanks [10:33] yw === tkamppeter__ is now known as tkamppeter [10:44] rodrigo_, gnome bug 659689 [10:44] Gnome bug 659689 in libgnome-desktop "gnome_rr_config_apply_from_filename_with_time does an unnecessary XRR roundtrip if the config file doesn't exist" [Normal,Unconfirmed] http://bugzilla.gnome.org/show_bug.cgi?id=659689 [10:44] chrisccoulson, ok thanks === Pendulum_ is now known as Pendulum [11:02] pitti: after some research, I think the best solution here is wait until the gsd dbus service is ready and then use it from indicator-power, so we do not have conflicts. Working on a patch [11:05] jjardon: i. e. not activate it yourself? [11:06] jjardon: we have an alternative proposal for ubiquity to start g-s-d earlier, wait for it to appear on the bus, and then start its panel [11:06] but of course the two don't exclude each other :) [11:07] pitti: yeah. Use g_bus_watch_name() to start to use the service when It appears === MacSlow is now known as MacSlow|lunch [11:13] gutted - http://www.bbc.co.uk/news/uk-england-somerset-14998964 [11:13] just think of all that wasted cheese [11:15] hum, cheese! ;-) [11:15] well, uk cheese, it's probably not real cheese ;-) [11:15] indeed, it's not cheese [11:16] I tried that in Ireland, "white sheddar" looks like "gruyère" [11:16] not at all the same on pasta :p [11:16] heh [11:18] chrisccoulson, will we get a stable tb before oneiric? [11:18] it's weird to be greeted by a "welcome to thunderbird beta" ;-) [11:19] seb128, yes ;) [11:19] great ;-) [11:52] pitti: just committed this patch. Maybe you can verify that It solves the problem for you? http://paste.ubuntu.com/694409/ [11:52] jjardon: you can link that to bug 854717, BTW [11:52] Launchpad bug 854717 in ubiquity "Broken panel icons and dialog style during ubiquity-dm and OEM install/final user configuration" [Critical,Fix released] https://launchpad.net/bugs/854717 [11:53] jjardon: nice! [11:53] jjardon: once it's built, I can try that with an older ubiquity version, that'll double-check [11:58] pitti: great :). Just added a comment to the bug pointing to the commit [12:13] lunch, bbl [12:20] http://cdimage.ubuntu.com/daily-live/20110921.1/ up for testing [12:21] pitti, great [12:36] seb128, didrocks: https://wiki.ubuntu.com/OneiricOcelot/TechnicalOverview#Ubuntu_Desktop seems current to me; can you think of anything which we introduced after b1 which ought to be described? [12:36] didrocks: oh, oneconf perhaps? [12:37] pitti: yeah, as syncing is now on, I can put a sentence about it [12:37] pitti, nothing I can think offhand but I will tell you if I figure one [12:37] nothing for unity [12:37] * pitti will review our meeting pages [12:38] didrocks: I'll add CJK support for the dash search [12:39] hm, we really need to fix versions.py to take the unapproved queue into account [12:39] pitti, i've updated the etherpad for the todo list, but yes [12:40] looking into it [12:40] seb128: ooh, new g-i? /me grabs [12:40] it should check the queue but maybe it's only the new one, not "unapproved" [12:40] pitti: indeed, adding [12:40] for upload in distro_series.getPackageUploads(status='Unapproved'): [12:40] no, it checks for Unapproved, weird [12:40] doing glib-network as well [12:41] pitti, danke [12:43] * didrocks doesn't like people getting a traceback, and then uploading it only after dist-upgrading… [12:45] didrocks, you hate users, just say it! ;-) [12:45] seb128: you teached me so! :p [12:46] * didrocks looks if apt has the info of when the package was installed [12:46] maybe apport can check the timestamp in /var/crash and when the package was installed [12:46] didrocks, what do you try to prevent? apport should not let you report bugs where the versions listed are outdated [12:47] didrocks: yes, that's been on my TODO list for quite long, I hope I can do that in perky penguin [12:47] seb128: basically, someone got a crash, then apt-get upgrade, then run apport to report the bug [12:48] retracing will fail for those usually, but yeah it creates noise [12:48] seb128: apport will tell "I have the latest and greatest" and upload :) [12:48] seb128: not on python! [12:48] pitti: that would be excellent ;) [12:48] the ones I would really like to block is those happening at session closing and showing at next login ;-) [12:48] didrocks: this can be done cheaply with comparing the time stamp of the executable file at crash and at GUI time [12:48] seb128: indeed [12:48] seb128: bug 460932, but need to sit down for a while to see how we can prevent it [12:48] Launchpad bug 460932 in apport "Do not catch crashes which happen at logout" [Wishlist,Triaged] https://launchpad.net/bugs/460932 [12:48] pitti: hum? yeah, that's a cheap and nice idea :) [12:49] didrocks: faster than md5sum'ing etc. [12:49] pitti: let me put that on my TODO [12:49] didrocks: bug 132904 [12:49] Launchpad bug 132904 in apport "Ignore crashes for running programs whose package got updated underneath" [Low,Triaged] https://launchpad.net/bugs/132904 [12:49] didrocks: I'll assign it to me and tag it pet-bug [12:49] didrocks: one of my goals for next cycle is to fix pet-bugs again :) [12:50] pitti: oh, a really recent bug :) [12:50] pitti: great! :-) [12:50] pitti, one way would be to not spawn dialogs for things not from the current boot [12:50] i.e anything older than dmesg stamps [12:50] hah, it is already === MacSlow|lunch is now known as MacSlow [12:50] bug 853951 is _really_ beginning to irk me [12:50] Launchpad bug 853951 in unity "Workspace switcher makes windows lose decorations" [Undecided,New] https://launchpad.net/bugs/853951 [12:51] smspillaz, ^ [12:51] mdeslaur, never got this one [12:51] seb128: hmm, that's worth thinking through indeed [12:51] mdeslaur: urgh? doesn't happen here. [12:51] mdeslaur: didrocks it's a known issue, I'm going to work on it when I get a chance [12:51] didrocks, smspillaz, seb128: thanks [12:51] I wonder what the "test drive" button in usb-creator does [12:51] mdeslaur: smspillaz: targetting it then [12:51] it's on the dialog after the iso recording [12:54] didrocks: thanks for the wiki update; I did some cleanups, and signing off now [12:55] pitti: great, thanks :) === m_conley_away is now known as m_conley [13:16] i've just been trying to debug another issue with the volume notifications, and this time i don't think it is the fault of g-s-d [13:17] according to pulseaudio indicator-sound-service responds to each volume change by setting the volume again [13:17] and it ends up cancelling the mute when g-s-d initially sets it [13:17] chrisccoulson: lol [13:18] ie, g-s-d sets muted and displays it on screen, but indicator-sound-service immediately unmutes it again [13:18] so the next time you press the volume down media-key, you get the same "muted" icon without the throbbing effect [13:18] time to report an indicator-sound bug ;) [13:19] that's been bugging me for weeks now [13:31] vuntz, can you review a gnome-desktop patch from chrisccoulson? [13:32] * pitti grabs gtk+3.0 [13:35] nice, i'm glad i narrowed down bug 855557, that's been annoying me for ages :) [13:35] Launchpad bug 855557 in indicator-sound "indicator-sound-service interferes with gnome-settings-daemon" [Undecided,New] https://launchpad.net/bugs/855557 [13:35] chrisccoulson, bounce it to dx, or ted, ronoc is on leave [13:35] hum, ok who understand launchpadlib? [13:35] ubuntu_series = launchpad.distributions['ubuntu'].getSeries(name_or_version='oneiric') [13:35] print ubuntu_series.getPackageUploads() [13:35] that gives me [13:35] [13:36] what I can do from it? [13:36] version tries to iterate through it with a "for upload in ubuntu_series.getPackageUploads():" [13:36] but that seems to not work [13:36] iteration usually works [13:37] what doesn't here? [13:37] ubuntu_series = launchpad.distributions['ubuntu'].getSeries(name_or_version='oneiric') [13:37] for upload in ubuntu_series.getPackageUploads(): [13:37] print "unapproved:", upload [13:37] [13:37] -> empty output [13:37] pitti, ^ [13:37] it seems to hang [13:38] seb128: erm, just getPackageUploads()? [13:38] isn't that going to give you a bazillion results? [13:38] getPackageUploads(status='Unapproved') [13:38] pitti, ubuntu_series.getPackageUploads(status='Unapproved'): [13:38] rather [13:38] when I dropped the status to see if that was the issue [13:39] >>> up.total_size [13:39] 30 [13:39] that seems fine [13:39] and iteration works here [13:39] ok [13:39] ubuntu_series = launchpad.distributions['ubuntu'].getSeries(name_or_version='oneiric') [13:39] for upload in ubuntu_series.getPackageUploads(status='Unapproved'): [13:39] print "unapproved:", upload [13:39] -> empty [13:39] for me [13:39] like it returns without printing anything [13:40] >>> for u in up: [13:40] ... print u.display_name [13:40] ... [13:40] pygobject [13:40] gobject-introspection [13:40] eucalyptus [13:40] [...] [13:40] and just printing u gives me "https://api.launchpad.net/1.0/ubuntu/oneiric/+upload/3198332" [...] [13:41] mueuh [13:41] ooh [13:41] pitti, http://pastebin.ubuntu.com/694454/ [13:41] is what I'm trying to run [13:41] seb128: it seems it doesn't work with login_anonymously() [13:41] WTH [13:41] darn, seems we need to login authenticated again [13:41] why?! [13:42] i. e. login_with(), and supply a cookie the first time [13:42] I don't know [13:42] I don't want to start having to maintain a cookie for version [13:42] I guess I should ask on #launchpad, seems a bug [13:43] pitti, well in any case that's why the versions queue use is broken [13:43] ah, ok [13:43] since that's what it does to list items in the queue [13:43] seb128: we could just leave the cookie on ~/.launchpadlib/ on people, and give it a readonly priv? [13:43] pitti, wfm [13:44] but we need an account to use? [13:44] or do you want to use the retracer one? [13:44] I'm happy for it to use my creds, as long as it's readonly and public [13:44] pitti, want to set it up? [13:44] seb128: sure [13:44] thanks! [13:51] seb128: done, next cron job should get it [13:51] in 15 ins [13:51] pitti, danke! ;-) [14:04] seb128: would I need an FFE for the libnm-gtk split (what is needed for g-c-c) ? [14:04] pitti, ^ [14:04] I'm about to write one for NM 0.9.1.90 too. [14:04] cyphermox: they split a source pakcage? [14:04] yeah [14:04] seems fine [14:04] cyphermox: any difference in the binary packages? [14:05] wait no [14:05] they split the binary package [14:05] i. e. source only split, or does it create new library packages? [14:05] new library [14:05] is there an easier way to specify configure option to dh rules than using "override_dh_auto_configure:"? [14:05] that'll mean a transition with potentially many rebuilds, so we should have an FFE [14:05] i.e should I [14:05] seb128: I don't think there is [14:05] override_dh_auto_configure: [14:05] dh_auto_configure --disable-icon-update [14:05] pitti: nothing was using any of it before :) [14:05] dh_auto_configure -- --disable... [14:05] rather [14:05] pitti, ok, thanks [14:06] pitti: it's been split to make wifi/mobile dialogs available to other apps instead of relying on nm-applet [14:06] cyphermox: seems fine [14:18] pitti: my ffe for NM is https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/855592 [14:18] Ubuntu bug 855592 in network-manager "FFE: NetworkManager 0.9.1.90" [Undecided,New] [14:21] cyphermox: updated [14:21] thanks [14:24] seb128: http://people.canonical.com/~platform/desktop/versions.html> darn, no difference [14:24] pitti, "Last updated: Wednesday September 21 2011 13:50:06 +0000" [14:24] pitti, it was before your change [14:24] oh, versions.py stopped running, weird [14:25] * pitti runs in foreground, perhaps it crashed [14:25] maybe it's waiting for lplib authentification? [14:25] or something went wrong with the cookie [14:25] no, I did that already, and checked [14:25] there is a reason I prefer anonymous ;-) [14:25] ok [14:25] temporarily disabling cronjob (in case I forget) [14:26] pitti: can we go back to the old, non-broken, python-gojbect? :) [14:26] dobey: what is broken? [14:27] seb128: running in foreground now; lp login etc. went fine [14:27] pitti: requiring people to use introspection, when all the APIs they need to use are not introspected, and some of the APIs they need, don't have static bindings [14:28] dobey: how was the latter any different? [14:28] e. g. Soup didn't have static bindings, so you couldn't use it before either [14:28] pitti: back when we could mix static bindings and gi bindings, we could actually have software that works reliably [14:28] that actually never worked reliably [14:28] pitti: right, but we could mix static/gi with old gobject [14:28] it was a source of weird crsahes [14:30] i'd rather have a million bugs depending on upstream getting their act together, than having them tell me i'm not allowed to write applications using APIs I need to use, because they haven't been bound in GI :-/ [14:32] and my particular points of pain right now are Unity and twisted [14:32] not Soup [14:32] dobey: you can still do it by importing all GI bits first, and then the static ones [14:32] but you get to keep both halves when it breaks [14:32] $ python -c 'import gobject; from gi.repository import Gtk' [14:33] -> ImportError('When using gi.repository blabla') [14:33] $ python -c 'from gi.repository import Gtk; import gobject' [14:33] unfortunately not possible [14:33] -> no complaint [14:33] but even this simple one already breaks in glib [14:33] (Warning: g_boxed_type_register_static:..) [14:35] $ python -c 'from gi.repository import Soup; import twisted' [14:35] dobey: ^ that particular combination might work? [14:36] ah, twisted itself doesn't use gobject, that was just a particular submodule [14:37] chrisccoulson, is ureadahead working for you? is it supposed to do something on a ssd? on my chart it doesn't stop any other loading and the bar has no io color, seems weird [14:37] dobey: is there a particular library you need? building a gir is not exactly rocket science, so we can help upstream to build one [14:37] pitti: from twisted.internet.glib2reactor import install [14:37] seb128, yeah, it seems to work ok here [14:37] pitti: the problem isn't that the .gir/typelib is missing; the problem is that glib/gio/gobject stuff isn't totally bound yet [14:37] chrisccoulson, hum ok, I'm testing on the 10v with an install I just did an hour ago [14:38] seb128, actually, maybe not - https://launchpadlibrarian.net/80459240/farnsworth-oneiric-20110920-16.png [14:38] that doesn't make a lot of sense ;) [14:38] your changes in the ppa won 1s on the boot there [14:38] nice :) [14:38] seb128, is that a fresh install? [14:38] pitti: or well, i'm trying to port the glib2/gtk2 reactor to gi/gtk3 APIs with gir, but alas, lack of upstream bindings [14:38] chrisccoulson, well "fresh", rebooted 10 times with some minutes in session [14:39] chrisccoulson, i.e it should have done the ureadahead caching [14:39] there's still the case where it regresses if you've ever changed your monitor settings, which sucks [14:39] (ie, it adds back another reprobe) [14:39] which there isn't really any need for :) [14:39] it's "down" to 26.8s on the 10v [14:41] seb128, what was it in lucid? [14:42] https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis [14:42] 12s... [14:42] still a way to go then ;) [14:42] but it was 37s 1 week ago on oneiric [14:42] so 10.2s in a week is already something ;-) [14:42] :) [14:43] it's not only us [14:43] gnome-session used to start 6s into the boot, it starts at 9s now [14:43] ah, so it's not all our fault then ;) [14:44] I need to debug nautilus [14:44] it was taking 3 seconds of cpu, it's taking like 10s now [14:44] and I've moved all the .so out, so it's not the u1 integration or something else being stupid [14:44] ouch, that's quite a bit [14:45] the other offender is compiz,unity [14:45] yeah, i think those are a lost cause for oneiric ;) [14:45] also we use to run the wm directly [14:45] not delay it to after g-s-d [14:45] i.e on http://people.canonical.com/~pitti/bootcharts/daniel-lucid-20100305-1.png [14:46] lot of small things taking 1s as well [14:46] the helper to detect if we should run 3d or not [14:46] seb128: hm, who knows -- http://people.canonical.com/~platform/desktop/versions.html worked now [14:46] heisenbug [14:46] zg taking over a second cpu [14:46] pitti, looks better ;-) [14:47] seb128: the helper souldn't take any time now, (at least, you should pay the price before gnome-session is there) [14:47] didrocks, it's autologin [14:47] seb128: even with autologin [14:47] seb128: indeed, much more useful again \o/ [14:47] seb128: it's run *before* gnome-session [14:47] didrocks, so it's buggy, it takes 2s of blue bar [14:47] so, we'll all be discussing boot speed at UDS then ? :) [14:48] didrocks, not on my charts [14:48] seb128: ah, it marks the unapproved ones with a little "Q" [14:48] pitti, right [14:49] seb128: can you upload your chart? [14:49] didrocks, yeah, I'm copying them over [14:50] if you are on autologin [14:50] you should see it running twice [14:50] one before gnome-session, paying the price [14:50] and a short time after gnome-session (just checking a file) [14:51] seb128: ls /tmp/unity_support_test* [14:51] didrocks: oh, I was going to ask you about this -- what creates /tmp/unity_support_test.0 ? [14:51] didrocks: it's a predictable file name, so whatever creates it must be *very* careful to not make this a symlink attack vector [14:52] pitti: unity_support_test creates it :) [14:52] (/usr/lib/nux/unity_support_test) [14:52] pitti: hum, it runs under the user or lightdm name, indeed [14:52] didrocks: does it need to be in /tmp/ ? [14:52] pitti: it needs to be somewhere where it's removed at startup [14:52] didrocks, doh, ok, my fault [14:53] pitti: basically, that's the speed improvment I made last week i discussed on the team report [14:53] didrocks, there is a gnome-fallback... on the chart, I though it was the fallback 3d session upstream thing [14:53] pitti: when unity-greeter is running, the support tool test is ran and drop that file depending on the result [14:53] but it seems it's gnome-fallback-mount-helper from g-s-d [14:53] seb128: ahah! :-) [14:54] didrocks, pitti, chrisccoulson: http://people.canonical.com/~seb128/oneiric-Inspiron-1011-oneiric-20110921-6.png [14:54] pitti: then the scripts in /etc/X11/Xsession.d/ before logging try to pick it [14:54] 10v beta2 chart with the desktop team ppa [14:54] seb128: oh, wow, on mine it was 2 s faster even without the PPA: http://people.canonical.com/~pitti/bootcharts/daniel-oneiric-20110914-disable-indicator-session.png [14:55] seb128: however, the desktop PPA should make little difference [14:55] why does it call modprobre and blkid in serie? [14:55] as both CPUs are fully used all the time, the extra wait time doesn't matter [14:55] pitti, it makes a ~1s difference there [14:56] nice [14:56] pitti: I'm happy with any better solution, basically, the test tool should be able to cache its result and we need to have a fast check even if we restart it upon reboot [14:57] didrocks: so it can't go into ~, and it needs to be predictable [14:57] pitti: knowing that it can be run from multiple users on the machcine (and even lightdm) [14:57] didrocks: and needs cleaning up on boot? [14:57] pitti: indeed, if you install a new driver or whatever [14:57] didrocks: then the only locations are /tmp/ and /run/lock/, both of which are world-writeable [14:57] I picked /tmp for that reason [14:57] didrocks: so we indeed need to be careful; I'll have a look at the code, I guess it's in lp:unity? [14:58] pitti: no, it's lp:nux [14:58] pitti: my code is pretty silly, but yeah, I will leave a second look :) [14:58] didrocks: I can just run that tool without harm, right? [14:58] didrocks: what's .0, the $DISPLAY? [14:58] pitti: yeah, it won't hurt you, I promess! [14:58] pitti: no, it's the result (0 for ok, 1 for ko) [14:59] ah [14:59] right, it's not user specific [14:59] nice - http://chevrel.org/temp/Capture-trailer_400p.png :) [14:59] pitti: http://bazaar.launchpad.net/~unity-team/nux/trunk/view/head:/tools/unity_support_test.c [14:59] didrocks: dang, it's vulnerable [15:00] $ ln -s pwned /tmp/unity_support_test.0 [15:00] $ /usr/lib/nux/unity_support_test [15:00] $ ls /tmp/pwned [15:00] /tmp/pwned [15:00] didrocks: I'll create a branch [15:00] pitti: thanks, can push it in next nux release [15:00] pitti: I just fopen(resultfilename, "w+") and then fclose it [15:01] didrocks: right, you need to use the Unix open() with O_CREAT|O_EXCL [15:01] the latter is the important bit here, to atomically guard against symlink attacks [15:02] pitti: ok, not really good at security on this, will look at some documentation with your branch :) [15:02] ok, basically, the idea is to fail is the file is already there? [15:03] didrocks: yes [15:03] didrocks: the attach scenario is that you can overwrite/kill any file which is owned by the user who calls support_test [15:03] didrocks, how busy are you? feel free to grab some GNOME updates this week when you have a slot ;-) (hint, gnome-session for example) ;-) [15:03] i. e. lightdm or you [15:03] didrocks: you can e. g. point it to your keyring, or the lightdm configuration, etc. [15:03] didrocks: I'll create a branch/MP [15:03] pitti: oh ok, so some kind of "nullify" attack [15:04] pitti: but to detect if the file exists, I just test resultfile = fopen("/tmp/unity_support_test.0", "r"); (and fclose it if it's there, nothing to read and such). This has no security implication, isn't it? [15:04] seb128: will do [15:04] didrocks, thanks [15:05] seb128: still some compiz churn, but will be ok afterwards, I hope [15:05] didrocks, no hurry, I doubt we will unfreeze before friday morning [15:05] didrocks, but feel free to claim on the pad for later ;-) [15:06] yeah, I'm afraid of that too, I'll try have some iso testing as well tomorrow morning [15:06] didrocks, pitti: is there a possibility of someone malicious creating the file and breaking the GUI on a multiuser system? [15:07] mdeslaur: well, define "breaking" [15:08] didrocks: ie: saying the hardware supports 3d, when in fact it doesn't, so desktop sessions don't come up? [15:08] mdeslaur: the support test tool will say "go ahead", then unity needs to fallback if it can't load, which is a whishlist bug opened [15:09] seb128: I'm maybe blind but I don't see the unity_support_tool being run on your bootchart? [15:09] didrocks, doh, ok, my fault [15:09] didrocks, there is a gnome-fallback... on the chart, I though it was the fallback 3d session upstream thing [15:09] but it seems it's gnome-fallback-mount-helper from g-s-d [15:09] didrocks, ^ [15:10] seb128: yeah saw that, but still, we should it running [15:10] the file should be in a non-user writable location, and maybe the timestamp could be checked to see if it's stale or something [15:10] oh [15:10] didrocks, well in the session it's normal, if it just checks a file it's too short to register on the chart [15:10] mdeslaur: the file is created by the tool runned by the user though [15:10] seb128: but that means that the initial check is fast as well? [15:11] didrocks: I thought you said the tool is run by unity-greeter? [15:11] mdeslaur: ok, so the tool *can* be run by the unity-greeter [15:11] mdeslaur: there are two cases: [15:12] - no autologin and unity-greeter -> the tool is run there, while the user is typing his password, the detection is done and cache the info in /tmp [15:12] didrocks, I don't really know, need to check, it seems weird indeed [15:12] then, Xsession.d (under the new user) check if the file is there, and run the session [15:13] didrocks, it shows on my laptop manual login chart: http://people.canonical.com/~seb128/seb-e6410-oneiric-20110920-9.png [15:13] mdeslaur: if the user is using something else or autologin, the Xsession.d won't find the file and will run the tool then [15:13] seb128: hum, interesting [15:15] didrocks, it's running, the proc_ps.log has it [15:16] didrocks, it's like it was exiting immediatly [15:16] didrocks, maybe lightdm close and destroy the process directly in autologin [15:16] didrocks, I think lightdm does clean behind him [15:16] didrocks: ok...hrm... [15:17] didrocks, lightdm is too light, the helper doesn't have time to run ;-) [15:18] seb128: hum, try logging in slower please :) [15:18] seb128: oh, what you can do [15:18] didrocks, it's *auto*login :p [15:18] seb128: reboot, start unity-greeter [15:18] ah, you mean, in the autologin case, not the bootchart :) [15:18] right [15:18] so, in the autologin case, lightdm doesn't start it [15:18] didrocks, it does [15:19] it shows with one tick in the bootchart log [15:19] seb128: unity-greeter is run? [15:19] no [15:19] it's autologin [15:19] so, it's not lightdm [15:19] no greeter [15:19] well, it's lighdm in same way [15:19] but not *the* lightdm :) [15:19] it's run by /etc/X11/Xsession.d/50_check_unity_support [15:19] so, it should do the full check and drop the file [15:21] well it drops the file [15:23] didrocks, it seems fast, it takes 0.1s (user time) (0.04s system time) to run on a session after flush caches [15:23] ie echo 3 > /proc/sys/vm/drop_caches [15:24] didrocks, so maybe it's fast enough to not be picked by the bootchart, the graphic filter out "noise" [15:24] open("/tmp/unity_support_test.0", O_WRONLY|O_CREAT|O_EXCL, 0666) = -1 EEXIST (File exists) [15:24] yep [15:28] seb128: weird that it's slower when the greeter is there though [15:28] didrocks, it's not the same box, the greeter login is my laptop not the 10v [15:29] could be that the intel probing is faster on the 10v [15:29] quite some people have slow probing issue on modern intel hardware it seems [15:29] seb128: ok, so it seems that the cache (for your laptop) is still useful as winning 0.5s maybe? [15:30] didrocks, right [15:33] desrt, when binding a GSettings key to a GtkAdjustment value that changes quick, is gsettings clever enough to just not write for every change? [15:34] didrocks, mdeslaur: https://code.launchpad.net/~pitti/nux/symlink-attack-fix/+merge/76434 [15:39] pitti: looks ok, thanks [15:40] so, good night everyone! [15:40] +1 as well then :) [15:40] bye pitti [15:40] pitti: good night! [15:40] pitti: merging, thanks a lot! good night [15:44] Laney, rodrigo_: do you hand with tomboy upstream on IRC or something? could you ask if they have any plan to fix https://bugzilla.gnome.org/show_bug.cgi?id=657180 for 1.8? [15:44] Gnome bug 657180 in General "tomboy blocks shutdown" [Normal,Unconfirmed] [15:44] rodrigo_: no. it's not. [15:45] seb128, I am on #tomboy all day yes, but don't participate much [15:45] rodrigo_: it's a difficult problem to solve [15:45] but yes, sure, asking now [15:46] rodrigo_, thanks [15:46] desrt, right, I'm seeing evolution has a gconf-bridge object that does the binding thing that gsettings does [15:46] desrt, they use a timeout, which not sure it's the best way [15:46] rodrigo_: a good approach may be to introduce a flag at the binding API layer [15:46] RATE_LIMIT or so [15:47] yeah, like _DELAYED? [15:47] right [15:47] or maybe apply the rate limiting by default [15:47] and have a flag to turn it off [15:47] it seems like a problem that's unique to property bindings [15:47] and solving the problem at lower layers is substantially more complicated [15:47] rodrigo_: i welcome a bug [15:47] desrt, I'll file one, yes [15:51] desrt, https://bugzilla.gnome.org/show_bug.cgi?id=659738 [15:51] Gnome bug 659738 in gsettings "Need a way to tell GSettings to delay writes for rapidly changing properties" [Normal,Unconfirmed] [15:51] desrt, to be honest, I haven't seen any bug coming from that, so maybe it's not an issue as it was with GConf? [15:52] rodrigo_: it's a performance hit [15:52] ok [15:52] particularly on ext4 with spinning disks [15:52] fsync() take a decade there [15:52] and dconf does fsync() on each write [15:52] (because it has to) === tkamppeter_ is now known as tkamppeter [15:52] so 100 changes -> 100 fsync() round trips. each takes like 40ms [15:53] even though it's async, your disk ends up active for a few seconds after you stop [15:53] that's bad [15:53] yeah [15:53] tedg, hey [15:53] tedg, can you make sure bug #854292 and bug #855557 are assigned and tracked for Oneiric? [15:53] Launchpad bug 854292 in indicator-session "indicator-session disappear after running update-manager" [High,Confirmed] https://launchpad.net/bugs/854292 [15:53] Launchpad bug 855557 in indicator-sound "indicator-sound-service interferes with gnome-settings-daemon" [High,New] https://launchpad.net/bugs/855557 [15:53] seb128: so the mime database stuff in glib... [15:54] seb128: the latest version of the control center no longer uses this API, so it shouldn't be a problem (as i understand it) [15:54] desrt, well as commented on the bug the user bug is still the same [15:54] desrt, open the info capplet and see .local/share/applications/mimeinfo.list being created with random values [15:55] desrt, halfline and alex confirmed it's behaving weird also for them on their fedora with 3.1.92 yesterday so it's not Debian,Ubuntu specific [15:55] desrt, but the bug might be somewhere else... [15:56] seb128: odd. [15:56] seb128: i'll look again, in any case [15:56] thanks [15:56] since lxde and xfce are impacted as well [15:56] well, even for GNOME is seems a not-so-nice bug [15:56] open the control center and get gwibber to open photos, gedit to open calendars and another web browser than the one you used ;-) [15:59] hm, is it just me or does anyone lese find it confusing that "Guest session" is written in the same text / color as ubuntu 11.10 and yet you can click on it but you can't click on ubuntu 11.10 [16:02] seb128: so i do as you say, and i get a file [16:02] but the contents of the file are pretty reasonable [16:02] jpegs open with eog, videos with totem, oggs with rhythmbox, etc. [16:05] desrt, http://pastebin.ubuntu.com/693740/ [16:06] that's the file I get [16:06] desrt, well, it pick handlers for the said type, so if you have i.e only one email client, one web browser, etc you are likely to be ok [16:07] * desrt sort of wonders why merely opening the panel and doing nothing at all is enough to cause a file to be generated [16:07] yeah, that's a bug as well in my opinion, bastien seemed to disagree it's one though [16:07] reminds me that i have a patch to commit to g-c-c [16:07] same, I don't see why it creates "added associations" [16:18] out for a bit, bbl [16:31] have a good night everyone [17:00] rodrigo_, seb128 - what is the outlook on bug 832603? Am stumbling into it on a fresh update. [17:00] Launchpad bug 832603 in gnome-settings-daemon "gnome-settings-daemon crashed with SIGSEGV in g_simple_async_result_complete()" [High,Triaged] https://launchpad.net/bugs/832603 [17:01] * skaet does fresh update from natty, and booting oneiric, this is first thing seen. [17:01] skaet, rodrigo seems to not be online, I don't personally know but I will check with him when he's there [17:01] seb128, thank you. Anything you think he might want gathered from the machine before I reboot? [17:01] skaet, is g-s-d down, like no theme, etc or do you just get the apport prompt but things work normally? [17:01] skaet, add your .xsession-errors to the bug if you can please [17:02] skaet, oh and could you check the timestamp of the .crash and see if it correspond to the login? [17:02] just to make sure it's not a segfault from the old version on logout before you restarted [17:03] g-s-d down, no theme. Its fresh install update from clean installed and updated natty. First boot triggered. [17:03] ok, so please add the .xsession-errors that should be enough [17:03] seb128, will do. Thanks [17:42] re [17:42] ok, that was "interesting" [17:43] g-s-d was not applied any icon theme, theme, etc [17:43] turn out that "don"t run unity-greeter --test-mode" with your user :p it will desactivate all the g-s-d plugins which are not actived on the greeter === eeejay_is_afk is now known as eeejay [18:41] seb128: can you please have someone look at bug 845374 [18:43] micahg, will ping kenvandine tomorrow [18:44] but it got only 3 duplicates in 2 weeks [18:44] seems like a low priority bug, we get lot of bugs that get that in a day [18:46] hrmm [18:46] seb128: really? maybe people aren't using gwibber then, feel free to lower priority [18:47] when did GtkBox switch to not having children expanded by default with _add()? [18:48] micahg, that bug says you don't need to use gwibber [18:48] micahg, gwibber works fine there [18:48] micahg, but anyway, I will get kenvandine to look at it [18:49] * micahg must be subscribe to weird people with long links then... [18:49] micahg, do you use ecryptfs? [18:50] seb128: yes [18:50] micahg, well maybe it's due ecryptfs filenames limitation [18:50] users tend to not run ecryptfs ;-) [18:50] seb128: orly, maybe I should have our ecryptfs guy take a look (tyhicks) [18:50] well it's still a bug, but it wouldn't be the first thing to be hit by it [18:51] evolution has issues due to it as well [18:51] the limitation is due to some patents iirc, so not something ecryptfs will fix [18:51] seb128: that's not quite true [18:51] seb128: actually for 12.04, tyhicks and jjohansen are support to fix this [18:51] ok, I should better not speak about what I don't know :p [18:51] the filename in the bug isn't long enough to be a ecryptfs issue [18:51] seb128: jjohansen had a functional, though non-perfect solution [18:52] kirkland, sorry, I though somebody told me that a year ago [18:52] seb128: there was a slight concern about that, but we cleared it [18:52] great to read ;-) [18:52] mdeslaur, oh ok [18:52] seb128: :-) [18:52] well as said, I will ask kenvandine to have a look when he's back tomorrow [18:52] seb128: thanks [18:52] it might just be a good old standard gwibber bug ;-) [19:09] brb [20:33] tedg: hey [20:33] desrt, Howdy [20:33] tedg: did you want to come to montreal? [20:34] desrt, I have passed a request up the chain, but I haven't gotten a response. Need to ping. [20:34] tedg: i think olli mentioned that it might be useful for you to be there [20:35] desrt, Yeah, and I had sent him justification. But, I never got back a yes/no from him. He's not online now. [20:35] tedg: he asked me if i thought it would be worthwhile. i said yes.... so here's hoping =) === eeejay is now known as eeejay_is_afk [21:15] desrt: btw, thanks for the invite (got it from jason), I'll definitely be there === m_conley is now known as m_conley_away [21:19] cyphermox: what do you think of the proposed fix for bug 852961 ? [21:19] Launchpad bug 852961 in network-manager-applet "nm-applet is wrongly started with Gnome Shell" [Undecided,In progress] https://launchpad.net/bugs/852961 [21:21] jbicha: what should it be set to? [21:25] jbicha: hmm.. maybe that's not such a bad idea, I can start to fix one of mpt's bug reports with this [21:41] bbl, dinner === eeejay_is_afk is now known as eeejay [22:43] cyphermox: nice :) [22:43] cyphermox: be sure to add yourself to the wiki! [22:44] cyphermox: (and ask the others) https://live.gnome.org/Montreal2011/Participants