jasoncwarner_robert_ancell and htorque that seemed to make a huge diff00:01
jasoncwarner_http://paste.ubuntu.com/694113/ u-g log00:01
jasoncwarner_I still had no mouse movement at the beginning...but that was for a second or two (which I think was X catching up)00:01
mdeslaurThis is weird: bug 85517100:09
ubot2Launchpad bug 855171 in ca-certificates "libnss3.so went missing after upgrade" [Undecided,New] https://launchpad.net/bugs/85517100:09
mdeslaurwhoops, wrong channel00:09
brycehjasoncwarner_, excellent, thanks for confirming that00:36
cyphermoxwould adding GNOME to NotShowIn cover just the Gnome Shell cases, or also break fallback and others? :)01:35
jbichaGNOME is Shell & Fallback01:38
cyphermoxjbicha: yeah, that's the problem probably ;)01:41
cyphermoxgoing to test this now01:41
jbichalike gnome-datetime-panel.desktop shows in Fallback g-c-c but it's OnlyShowIn=Gnome;01:42
jbichaecho $XDG_CURRENT_DESKTOP01:43
cyphermoxoh, I know about all of this, just "saying out loud" that I'm pretty much screwed either way01:45
cyphermoxjbicha: see https://bugs.launchpad.net/ubuntu/+source/network-manager-applet/+bug/852961 :)01:46
ubot2Ubuntu bug 852961 in network-manager-applet "nm-applet is wrongly started with Gnome Shell" [Undecided,New]01:46
cyphermoxI can't add it to NotShowIn without breaking something, and if I don't add it, something's still going to be broken01:47
cyphermoxbut I think Shell > fallback; so I'll add it.01:47
jbichabut I don't have nm-applet in my Shell session, just Fallback or Unity01:48
stgrabercyphermox: shell isn't shipped by default by a derivative...01:48
cyphermoxalso, really?01:48
cyphermoxstgraber: so edubuntu ships fallback?01:48
stgrabercyphermox: fallback is an option at install time in edubuntu01:48
jbichamy notifications area is blank on new start & I have the normal Shell network menu01:48
cyphermoxjbicha: that's weird01:49
cyphermoxwhy would this bug be open then?01:49
jbichaperhaps he's using an old shell version01:49
cyphermoxalright, thanks for clearing that up01:50
jbichaoh wait, he sees notifications in both shell style & an uglier style when the network disconnects for instance01:55
jbichaI can duplicate that01:55
cyphermoxwell then yeah, this is because nm-applet is getting started02:00
cyphermoxfixing this is going to be painful, because we're obviously limited by what's being run02:00
cyphermoxand shell / fallback sharing the same desktop value doesn't help02:00
jbichaI think gnome-shell uses nm-applet, just with a different frontend02:01
cyphermoxjbicha: no02:01
cyphermoxit uses its own frontend in js02:01
jbichaalso the latest gnome-shell (not in the archives yet) doesn't show network notifications for connections/disconnections02:01
jbichaso it's at least visually fixed with that02:01
cyphermoxis that going to be shipped?02:02
jbichajust waiting for GNOME to get around to releasing the new gjs which gnome-shell silently depends on02:02
* cyphermox cries02:03
cyphermoxI just won't touch this and sleep on the issue02:03
jbichathey just assume that you're going to be using the latest gjs & mutter & don't bother updating the configure.ac to reflect that02:03
brodercyphermox: i thought i remembered dcbw saying that nm-applet didn't actually provide ui with shell these days02:20
cyphermoxbroder that's what I remember too02:21
cyphermoxnm-applet shouldn't be running in shell, that's most likely the problem because it currently gets started in /etc/xdg/autostart02:21
brodercyphermox: no, no - i thought he said it runs but only does...something minor and invisible02:21
cyphermoxbroder: maybe02:22
cyphermoxAIUI it's just used for the wireless/mobile dialogs02:23
cyphermoxsomething that might be fixed by the new n-m-applet package with libnm-gtk split out02:23
brodercyphermox: http://paste.ubuntu.com/694185/ - from about a week ago02:23
cyphermoxI'll need to setup something with gnome-shell and mess around with it to see02:24
broderoh, it sounds like maybe the only problem is notifications?02:24
jbichabroder: yes I believe so02:25
cyphermoxsounds like it now02:26
cyphermoxjbicha: I just updated the bug, could you try figuring out what sends the notifcation that looks wrong?02:26
cyphermoxor if they have a different text, just a screenshot of each may be enough :)02:27
broderwe don't patch how nm-applet does its notifications - how would this not be broken anywhere that's using gnome-shell?02:27
jbichaI expect that Fedora would have the same problem02:28
brodererr...wait, i was looking at n-m, not nm-applet :)02:29
cyphermoxbroder: actually, I think it was lightly patched, but I don't think it would affect this02:30
cyphermoxit 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.02:30
broderis 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...02:31
jbichawhen I try to connect to a random encrypted wifi network, I get 2 popups to enter the password, one from gs, one from nm-applet02:32
jbichanot sure if the second one is just because I canceled the first one though02:33
broderdcbw said that should happen around gnome 3.2 or so - is that what we're at at this point?02:33
cyphermoxpretty much02:33
cyphermox"should" happen?02:34
jbichawe're on 3.1.90 and current is 3.1.92 though02:34
broderwell...gnome-shell definitely looks like it ships a secret agent02:35
broder(look at src/shell-network-agent.{h,c} and js/ui/networkAgent.js)02:36
jbichacyphermox: getting screenshots is a bit tricky since this is my web connection :)02:37
jbichaI'm going to reboot into Fedora02:37
cyphermoxbroder: jbicha: I really should be going to bed02:37
jbichaok, it's not a big issue anyway02:38
broderjbicha: 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
cyphermoxyou want to ping me if you find something ? :)02:38
cyphermoxyou guys seem to be getting more progress than I do02:38
brodercyphermox: it looks to me like nm-applet just shouldn't be running under gnome-shell anymore02:38
jbichaI think you could sniff dbus to check if you're in gshell or not02:39
cyphermoxbroder: I kind of got that feeling but I think you still need it for some dialogs at least02:40
brodercyphermox: does nm-applet (not nm-connection-editor) do anything besides secret prompts?02:41
cyphermoxbroder: that no, I don't think so02:41
brodercyphermox: then just have nm-applet exit(0) if it's running under gnome-shell...however you determine that02:42
cyphermoxnot sure how the more complex wifi connections are done02:42
broderwell, i've got an oneiric machine, and infrastructure for 802.1x and wpa enterprise , so i can try to do some experimentation at some point02:43
broderbut from my quick skim of the gnome-shell code, i see code for 802.1x, PPPoE, and modem connections02:44
broderonly thing i don't see offhand is vpn02:44
cyphermoxI'll look again in the morning02:45
=== eeejay is now known as eeejay_is_afk
jbichaFedora 16 doesn't show notifications for nm-applet so they only have the enter wifi password double dialog part of the bug02:50
cyphermoxmaybe it's caused by one of those old patches we carry and I keep having to fix03:00
jbichawell I think we want connection notifications in Unity though03:02
cyphermoxjbicha: 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 what03:11
kamstrupmvo: morning - I need you help on some popcon related stuff08:03
kamstrupmvo: how is the #stars computed from the popcon rating?08:03
kamstrupdigging through the S-C code only confused me more :-)08:04
kamstrupmy 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 anywhere08:04
kamstrupalso, I am not sure whether that makes sense at all :-)08:06
pittibonjour seb12808:08
seb128hey pitti, how are you?08:08
pittiseb128: furiously debugging ubiquity problems :)08:09
pittiquite well, thanks08:09
pittijust really concerned about the release time now08:09
seb128pitti, "fun" ;-)08:09
seb128is it worth testing the installer?08:09
seb128or is it "known buggy"?08:09
pittiit crashes if you are offline08:09
pittishould work fine, just look a bit ugly if you are online and don't use the live session08:10
pittiso, still worth testing if you want to08:10
mvokamstrup: *pfff* ;) we don't use the popcon for the stars, we use the ratings data08:10
mvokamstrup: what is the use-case, do you want to show it in unity?08:10
kamstrupmvo: lol, that explains my inability to find it :-D08:10
seb128pitti, let's see, I usually use the 10v as a test box and the wifi doesn't work out of the box08:10
kamstrupmvo: the dash allows you to filter by the S-C rating08:11
kamstrupmvo: so it's not in the xapian index?08:11
pittiseb128: right, it'd fail there, unless you have an ethernet cable plugged in08:11
pittiI did a test install on that one yesterday, with success08:11
kamstrupmvo: hit super-a and unfold the "Filter results" expander, and you'll see what I mean08:12
mvokamstrup: no, because its so dynamic. we store it currently in a extra file08:12
mvokamstrup: but that sounds like that needs some re-thinking08:13
kamstrupmvo: a fudge...08:13
kamstrupmvo: yeah, probably means we need to pull that feature from the dash...08:13
rodrigo_oh, everyone back :)08:13
mvokamstrup: aha, I was totally unaware of this feature in the dash08:13
rodrigo_there were only 20 people or so when I woke up, I guess freenode had problems?08:13
kamstrupmvo: yeah, it's not exactly a prominent place, hidden away inside the expander08:14
pittirodrigo_: yes, it did; should be fixed now, though08:14
mvokamstrup: so we currently use a xapian.KeyMaker and throw that as the sorter to the db08:15
seb128robert_ancell, hey08:15
seb128robert_ancell, what are you still doing online? ;-)08:15
robert_ancellseb128, hey, just logged on to bug you...08:15
kamstrupmvo: ah, so a KeyMaker that pulls the ratings from the file in order to generate the collation key?08:15
seb128robert_ancell, oh great, it has been a while I didn't catch you (part due to the fact I was on holidays ;-)08:15
mvokamstrup: and the review stats (star rating plus wilson-score based "score" of the ratings) are stored in a pythin pickle08:15
mvokamstrup: yeah08:15
seb128robert_ancell, what's up?08:15
robert_ancellseb128, 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 good08:16
kamstrupmvo: hehe, a pickle, that puts me in a pickle (wow, worst pun ever)08:16
mvokamstrup: now of course we can make this a c++ friendlier format for you, thats easy08:16
mvokamstrup: lol, indeed08:16
seb128robert_ancell, I'm happy to test those08:16
seb128robert_ancell, did you see the translation bug I commented on?08:16
mvokamstrup: if this software stuff does not work out, just do standup commedy ;)08:16
robert_ancellseb128, 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 better08:17
ubot2Launchpad bug 845549 in lightdm "Do not ship /etc/lightdm/lightdm-gtk-greeter.conf" [High,Fix committed] https://launchpad.net/bugs/84554908:17
kamstrupmvo: out only option is prettymuch json I guess... but I don'tknow if that kills performance... I gues it'll be a very large file08:17
seb128robert_ancell, ok08:17
mvokamstrup: would a libdb format work for you too?08:18
robert_ancellseb128, yes, I fixed that.  it was gettext not being initialized correctly08:18
seb128robert_ancell, great08:18
kamstrupmvo: libdb? you mean bdb, or?08:18
seb128robert_ancell, ok, I will test both trunks, review open bugs and drop you an email at the end of my day08:18
robert_ancellseb128, there will still be bugs with PAM messages not being localized, but they're more unlikely to occur08:18
mvokamstrup: yeah08:18
robert_ancellseb128, thanks!08:19
seb128robert_ancell, yw, great job btw, lightdm really works great for me ;-)08:19
kamstrupmvo: do we ship a bdb lib on the cd?08:19
seb128well the current version does, let's see trunk :p08:19
robert_ancellseb128, :)08:19
robert_ancellseb128, There's still more bugs cropping up, but I think we're on top of the most important ones.08:20
mvokamstrup: need to check, I can also play a bit with json. alternatively I could provide you a dbus service08:20
mvokamstrup: but that does not is speedy enough I guess08:20
kamstrupmvo: hmm, we can't depend on a running s-c08:20
seb128robert_ancell, yeah, I've not seen a lot of "important to fix for oneiric" ones, out of the ones assigned to you08:20
seb128robert_ancell, I will review the bugs today though just to be sure08:21
kamstrupmvo: json is easy, but scares me because I've seen how slow it is in gwibber08:21
mvokamstrup: it would be something extra, not part of main s-c, but its probably a bad idea even then08:21
mvokamstrup: yeah, exactly08:21
kamstrupmvo: it takes >1s 100% CPU churn to parse ~1k messages in gwibber08:21
mvokamstrup: s-c is already pretty slow to startup, I don't want to make it worse :/08:21
kamstrupgranted, that's some complex json, but still08:21
kamstrupmvo: lol, right :-)08:22
robert_ancellseb128, 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
kamstrupmvo: a mmap()able db format should make it faster than a pickle for S-C as well I guess08:22
mvokamstrup: 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
kamstrupmvo: yeah, sounds fit for a keystore08:23
seb128robert_ancell, right08:24
mvokamstrup: I think for next cycle we need to talk some more and we need to provide you proper APIs08:24
kamstrupmvo: 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
kamstrupbut i guess so08:26
robert_ancelljasoncwarner_, so bug 828112 is fixed?  Using packages from the standard repos?08:28
ubot2Launchpad bug 828112 in unity-greeter "Password field feedback slow at times" [High,Triaged] https://launchpad.net/bugs/82811208:28
mvokamstrup: both 4.8 and 5.1 are installed for me on the live-cd so we should be fine08:28
seb128robert_ancell, the updates I mentioned there are not in the standard repo but in the ubuntu-desktop ppa08:29
robert_ancellseb128, is there a bug I can merge this with?08:29
jasoncwarner_robert_ancell: I had to manually create an /etc/X11/xorg.conf file to stop the HDMI probe08:30
pittirobert_ancell: right, we need a wrapper around gnome-session for the lightdm guest account that we can hook the AA profile to08:32
robert_ancellpitti, yeah, annoying!08:32
pittirobert_ancell: but that can be a simple two-liner (she-bang and exec gnome-session "$@"), as in gdm08:32
pittijjardon: good morning08:33
pittijjardon: out of interest, what's the advantage of i-power talking to gnome-settings-daemon instead of upower directly?08:33
robert_ancellpitti, right, but there's no concept of sticking one in in lightdm, so we need to work out the right place to do it08:33
kamstrupmvo: ok, cool. I've no idea, but I assume we should use 5.1...08:34
seb128jasoncwarner_, bug #855350 is the bug dholbach opened about the output probbing issue08:37
ubot2Launchpad bug 855350 in xorg "long reprobes slow down boot/login" [Undecided,New] https://launchpad.net/bugs/85535008:37
seb128jasoncwarner_, if you want to comment on it08:38
jjardonpitti: 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 call08:42
pittijjardon: that makes it slightly dependent on GNOME, though?08:42
pittijjardon: I just noticed because I'm currently debugging bug 854717, and it's indicator-power which dbus-launches g-s-d08:43
ubot2Launchpad 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/85471708:43
pitti(a broken g-s-d)08:43
pittijjardon: that's not directly i-power's fault, of course, but it made me wonder why it uses g-s-d08:43
chrisccoulson_RAOF, Sarvatt - is the eDP slowness likely to affect changing the screen brightness via xrandr?08:44
chrisccoulson_i noticed this morning that the new kernel fixes a huge lag i was also seeing when adjusting the brightnes08:44
chrisccoulson_unless that's just my imagination ;)08:45
=== chrisccoulson_ is now known as chrisccoulson
pittichrisccoulson_: if xrandr --verbose contains a "BACKLIGHT", then we are using xrandr calls for the brightness keys, yes08:45
pittithe old gnome-power-manager had a fallback to poke the values into the ACPI files in /proc/, but I guess that's gone these days08:45
pittiwe now rely on the XBACKLIGHT extension, which I believe goes through XRandR08:46
pitti(I have no idea where this ends up at the kernel level, though)08:46
chrisccoulsonpitti - oh, no BACKLIGHT on my machine here :/08:46
jjardonpitti: 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 directly08:46
pittijjardon: I mean for XFCE and the like08:46
pittijjardon: 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 API08:47
jjardonpitti: XFCE has is own power indicator08:47
pittibecause if g-s-d is adding functionality, I'd rather move that down to upower08:47
jasoncwarner_seb128 and didrocks, can I reschedule our talks tonight? have a conflicting music lens discussion at the same time...08:49
didrocksjasoncwarner_: sure08:49
seb128jasoncwarner_, "no" ;-)08:49
jjardonpitti: 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 code08:51
pittijjardon: 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 yourself08:53
pittijjardon: 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 it08:53
jasoncwarner_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...08:53
pittijasoncwarner_: WFM08:54
seb128jasoncwarner_, works here08:54
mvokamstrup: proof of concept lp:~mvo/software-center/review-stats-as-dbm-for-unity-with-best-regards08:55
pittijjardon: 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
ubot2Launchpad 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/85471708:55
jjardonpitti: sure. have to go now, I'll follow the discussion later ;)08:55
mvokamstrup: 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 spread08:55
mvokamstrup: well, I guess not necssarily, depends on your UI needs, but that is what we use08:56
mvokamstrup: 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 it08:58
kamstrupmvo: I don't think generally the dash should cause changes to local data09:00
kamstrupmvo: it's for lookups and access09:00
kamstrupmvo: usage logging hapens indirectly via zeitgeist then09:01
mvokamstrup: if you need the review spread (10 x 1 star, 4 x 2 stars) etc, just let me know, that can be added as well09:01
kamstrupmvo: ok - not for O definitely09:02
mvokamstrup: and we might need a bugreport if that should make it into the archive for reference for the release team, uploads are hand-approved currently09:02
kamstrupmvo: ok...09:03
kamstrupmvo: what format do you want to use for the payload?09:03
kamstrupmvo: A GVariant "(iii)" would be extraordinarily handy for me :-)09:03
kamstrupdunno how easy they are to serialize in Python though... lemme see09:04
mvokamstrup: I already use python struct09:05
mvokamstrup: so you should be able to just use something like struct { int , int, int } in your c(++) code09:06
kamstrupmvo: oh? you sure?09:06
mvokamstrup: 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)09:07
mvokamstrup: let me quickly verify this09:08
davidcallepitti, or someone else, when is the new pygobject supposed to enter the archives?09:13
pittidavidcalle: 3.0.0? after b209:14
davidcallepitti, thanks, I'm worried about it. Looks like it makes python lenses segfault.09:15
seb128seems like a bug for kamstrup ;-)09:15
pittidavidcalle: 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 now09:16
pittidavidcalle: you can test it from the PPA, maybe you can get a stack trace and file a bug?09:16
davidcallepitti, I'm on the PPA. I will file a bug, thanks.09:17
=== ember` is now known as ember
kamstrupdavidcalle: you have the ppa url, I can debug the crasher09:27
pittikamstrup: add-apt-repository ppa:ubuntu-desktop/ppa09:28
kamstruppitti: thanks09:28
pittikamstrup: you'll also get a faster startup from that, for bonus :)09:28
pittijasoncwarner_: want to skip our 1-on-1, too?09:30
davidcallekamstrup, thanks :) Was learning how to get stack traces.09:30
* didrocks reboots09:31
jasoncwarner_pitti: I'm good for a bit if you have a couple of minutes...want to catch up on a couple quick things...09:35
jasoncwarner_i got mumble working as well (took a reboot)09:36
didrockswaow, that was… weird09:39
didrockslibnss3 package installed, but no /usr/lib/i386-linux-gnu/libnss3.so09:39
didrocksnetwork-manager and nm-applet didn't appreciate :)09:40
geserdidrocks: see topic in #ubuntu-devel09:41
didrocksgeser: argh, ok… thanks for the pointer09:41
mptpitti, 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
chrisccoulsonmpt - gnome-session09:45
chrisccoulsonthat has an inhibit API09:45
chrisccoulson(for logging out, anyway)09:46
mptchrisccoulson, ah, <http://live.gnome.org/SessionManagement/GnomeSession#A8._QueryEndSession>?09:46
seb128didrocks, mhr09:46
chrisccoulsonfor shutting down, i'm not sure if there is an inhibit API09:46
seb128didrocks, mhr3 and cyphermox got that bug as well this week09:46
seb128didrocks, cyphermox tried to figure why the file had been cleaned, not sure they figured it though09:47
didrocksyeah, seems ca-certificates being a little bit more too drastic in his pick :)09:47
kamstrupdavidcalle: don't stop then :-)09:47
didrocksseb128: what geser told, topic of #ubuntu-devel09:47
kamstrupdavidcalle: did you have a bug report?09:47
didrocksca-certificates eats system libraries, DO NOT UPDATE (bug #855171)09:47
ubot2Launchpad bug 855171 in nss "libnss3.so went missing after upgrade" [Undecided,Confirmed] https://launchpad.net/bugs/85517109:47
seb128I got lucky it didn't eat mine09:47
didrocksindeed, well, I just lost 3 minutes…09:48
didrocksfortunatly, I had it still in cache and --reinstall worked09:48
davidcallekamstrup, I'm filing one in a few minutes.09:48
didrocksotherwise, I would have to hunt for an usb key, and that would have took more time :)09:48
mptthanks chrisccoulson09:49
seb128stupid tb, stop marking the selected message as read when entering a mailbox! ;-)09:51
davidcallekamstrup, https://bugs.launchpad.net/unity/+bug/85540209:52
ubot2Ubuntu bug 855402 in unity "Python lenses segfault with pygobject3" [Undecided,New]09:52
kamstrupdavidcalle: tanks09:53
didrocksseb128: +109:53
rodrigo_chrisccoulson, your g-s-d package is indeed much quicker10:04
rodrigo_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 something10:05
chrisccoulsonrodrigo_, yeah, i planned to do it that way at some point10:05
rodrigo_that just returns a shared instance of GnomeRRScreen, instead of using the property on the GdkScreen10:05
rodrigo_chrisccoulson, ok, cool10:05
chrisccoulsonyeah, that would probably be better actually10:05
rodrigo_chrisccoulson, I can do it if you want10:05
chrisccoulsonrodrigo_, sure, feel free :)10:06
chrisccoulsoni'll push the current branch to bzr in a moment10:06
rodrigo_about the media keys one, not 100% sure, so better to file a bug for hadess to review it10:06
chrisccoulsonrodrigo_, did you see my notes on https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis ?10:06
rodrigo_chrisccoulson, no, looking10:07
rodrigo_chrisccoulson, yes, we can have the patches as they are for oneiric, I'll prepare the patch for git master10:08
chrisccoulsonthanks :)10:08
tjaaltonchrisccoulson: 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?10:09
chrisccoulsontjaalton, no, i've not seen that before. is this using compiz?10:10
chrisccoulsonactually, i'm not sure if that would matter10:10
tjaaltonchrisccoulson: yes, and happens only with fullscreen windows, it seems10:10
chrisccoulsonhow are you opening the new tab?10:10
chrisccoulsonthat should open in the window that gets the event :/10:11
tjaaltonwell, all the kbd shortcuts end up in the wrong window, so ctrl-w might close tabs behind my back :)10:11
chrisccoulsoni've seen focus issues with compiz like this ;)10:11
chrisccoulsoneg, bug 83862510:12
didrockschrisccoulson: speaking of which, did you upgrade using the ppa?10:12
ubot2Launchpad bug 838625 in compiz "Unpredictable focusing after switching viewports with the keyboard" [Undecided,Confirmed] https://launchpad.net/bugs/83862510:12
chrisccoulsondidrocks, yeah. i didn't see any issues afterwards10:12
didrocksgreat :-)10:12
tjaaltonchrisccoulson: oh, if it's related to the stacking bug then I'll just wait for the updates10:14
didrockstjaalton: you can try the ubuntu-desktop ppa10:14
tjaaltonhappens with fullscreen windows that are on separate viewports, one on top of the other10:14
tjaaltondidrocks: ppa:ubuntu-desktop?10:14
didrockstjaalton: ppa:ubuntu-desktop/ppa indeed :)10:14
tjaaltondidrocks: thanks, adding10:15
didrockstjaalton: keep us in touch :)10:15
chrisccoulsoni should try the 3d session again to see if the stacking fixes fix the focus issues i was seeing too10:15
didrockschrisccoulson: oh, you didn't try it? wasn't the purpose of installing the ppa :-)10:16
tjaaltonit's possible to fix the focus by going to another viewport with a non-fullscreen ffox window10:16
chrisccoulsondidrocks, yeah, i didn't notice any stacking issues (like, mouse clicks going to the wrong windows and the dash appearing underneath everything else)10:16
chrisccoulsonbut 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 bugs10:17
=== vuntz_ is now known as vuntz
chrisccoulsonrodrigo_, i've pushed the gsd changes to bzr btw10:26
chrisccoulsonremember also that there is a gnome-desktop change10:27
rodrigo_chrisccoulson, ok10:27
rodrigo_chrisccoulson, for the gnome-desktop change, I'll let you do the upstreaming :)10:28
chrisccoulsonrodrigo_, sure, no problem10:28
seb128didrocks, can we drop your gwibber upload to the desktop ppa? it's the same version than oneiric and 2 weeks old10:31
seb128bah, tb10:32
seb128"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."10:32
didrocksseb128: yes10:33
seb128didrocks, thanks10:33
=== tkamppeter__ is now known as tkamppeter
chrisccoulsonrodrigo_, gnome bug 65968910:44
ubot2Gnome 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=65968910:44
rodrigo_chrisccoulson, ok thanks10:44
=== Pendulum_ is now known as Pendulum
jjardonpitti: 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 patch11:02
pittijjardon: i. e. not activate it yourself?11:05
pittijjardon: 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 panel11:06
pittibut of course the two don't exclude each other :)11:06
jjardonpitti: yeah. Use g_bus_watch_name() to start to use the service when It appears11:07
=== MacSlow is now known as MacSlow|lunch
chrisccoulsongutted - http://www.bbc.co.uk/news/uk-england-somerset-1499896411:13
chrisccoulsonjust think of all that wasted cheese11:13
seb128hum, cheese! ;-)11:15
seb128well, uk cheese, it's probably not real cheese ;-)11:15
didrocksindeed, it's not cheese11:15
didrocksI tried that in Ireland, "white sheddar" looks like "gruyère"11:16
didrocksnot at all the same on pasta :p11:16
seb128chrisccoulson, will we get a stable tb before oneiric?11:18
seb128it's weird to be greeted by a "welcome to thunderbird beta" ;-)11:18
chrisccoulsonseb128, yes ;)11:19
seb128great ;-)11:19
jjardonpitti: just committed this patch. Maybe you can verify that It solves the problem for you? http://paste.ubuntu.com/694409/11:52
pittijjardon: you can link that to bug 854717, BTW11:52
ubot2Launchpad 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/85471711:52
pittijjardon: nice!11:53
pittijjardon: once it's built, I can try that with an older ubiquity version, that'll double-check11:53
jjardonpitti: great :). Just added a comment to the bug pointing to the commit11:58
rodrigo_lunch, bbl12:13
pittihttp://cdimage.ubuntu.com/daily-live/20110921.1/ up for testing12:20
seb128pitti, great12:21
pittiseb128, 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
pittididrocks: oh, oneconf perhaps?12:36
didrockspitti: yeah, as syncing is now on, I can put a sentence about it12:37
seb128pitti, nothing I can think offhand but I will tell you if I figure one12:37
didrocksnothing for unity12:37
* pitti will review our meeting pages12:37
pittididrocks: I'll add CJK support for the dash search12:38
pittihm, we really need to fix versions.py to take the unapproved queue into account12:39
seb128pitti, i've updated the etherpad for the todo list, but yes12:39
seb128looking into it12:40
pittiseb128: ooh, new g-i? /me grabs12:40
seb128it should check the queue but maybe it's only the new one, not "unapproved"12:40
didrockspitti: indeed, adding12:40
seb128for upload in distro_series.getPackageUploads(status='Unapproved'):12:40
seb128no, it checks for Unapproved, weird12:40
pittidoing glib-network as well12:40
seb128pitti, danke12:41
* didrocks doesn't like people getting a traceback, and then uploading it only after dist-upgrading…12:43
seb128didrocks, you hate users, just say it! ;-)12:45
didrocksseb128: you teached me so! :p12:45
* didrocks looks if apt has the info of when the package was installed12:46
didrocksmaybe apport can check the timestamp in /var/crash and when the package was installed12:46
seb128didrocks, what do you try to prevent? apport should not let you report bugs where the versions listed are outdated12:46
pittididrocks: yes, that's been on my TODO list for quite long, I  hope I can do that in perky penguin12:47
didrocksseb128: basically, someone got a crash, then apt-get upgrade, then run apport to report the bug12:47
seb128retracing will fail for those usually, but yeah it creates noise12:48
didrocksseb128: apport will tell "I have the latest and greatest" and upload :)12:48
didrocksseb128: not on python!12:48
didrockspitti: that would be excellent ;)12:48
seb128the ones I would really like to block is those happening at session closing and showing at next login ;-)12:48
pittididrocks: this can be done cheaply with comparing the time stamp of the executable file at crash and at GUI time12:48
didrocksseb128: indeed12:48
pittiseb128: bug 460932, but need to sit down for a while to see how we can prevent it12:48
ubot2Launchpad bug 460932 in apport "Do not catch crashes which happen at logout" [Wishlist,Triaged] https://launchpad.net/bugs/46093212:48
didrockspitti: hum? yeah, that's a cheap and nice idea :)12:48
pittididrocks: faster than md5sum'ing etc.12:49
didrockspitti: let me put that on my TODO12:49
pittididrocks: bug 13290412:49
ubot2Launchpad bug 132904 in apport "Ignore crashes for running programs whose package got updated underneath" [Low,Triaged] https://launchpad.net/bugs/13290412:49
pittididrocks: I'll assign it to me and tag it pet-bug12:49
pittididrocks: one of my goals for next cycle is to fix pet-bugs again :)12:49
didrockspitti: oh, a really recent bug :)12:50
didrockspitti: great! :-)12:50
seb128pitti, one way would be to not spawn dialogs for things not from the current boot12:50
seb128i.e anything older than dmesg stamps12:50
pittihah, it is already12:50
=== MacSlow|lunch is now known as MacSlow
mdeslaurbug 853951 is _really_ beginning to irk me12:50
ubot2Launchpad bug 853951 in unity "Workspace switcher makes windows lose decorations" [Undecided,New] https://launchpad.net/bugs/85395112:50
seb128smspillaz, ^12:51
seb128mdeslaur, never got this one12:51
pittiseb128: hmm, that's worth thinking through indeed12:51
didrocksmdeslaur: urgh? doesn't happen here.12:51
smspillazmdeslaur: didrocks it's a known issue, I'm going to work on it when I get a chance12:51
mdeslaurdidrocks, smspillaz, seb128: thanks12:51
seb128I wonder what the "test drive" button in usb-creator does12:51
didrocksmdeslaur: smspillaz: targetting it then12:51
seb128it's on the dialog after the iso recording12:51
pittididrocks: thanks for the wiki update; I did some cleanups, and signing off now12:54
didrockspitti: great, thanks :)12:55
=== m_conley_away is now known as m_conley
chrisccoulsoni'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-d13:16
chrisccoulsonaccording to pulseaudio indicator-sound-service responds to each volume change by setting the volume again13:17
chrisccoulsonand it ends up cancelling the mute when g-s-d initially sets it13:17
mdeslaurchrisccoulson: lol13:17
chrisccoulsonie, g-s-d sets muted and displays it on screen, but indicator-sound-service immediately unmutes it again13:18
chrisccoulsonso the next time you press the volume down media-key, you get the same "muted" icon without the throbbing effect13:18
chrisccoulsontime to report an indicator-sound bug ;)13:18
chrisccoulsonthat's been bugging me for weeks now13:19
seb128vuntz, can you review a gnome-desktop patch from chrisccoulson?13:31
* pitti grabs gtk+3.013:32
chrisccoulsonnice, i'm glad i narrowed down bug 855557, that's been annoying me for ages :)13:35
ubot2Launchpad bug 855557 in indicator-sound "indicator-sound-service interferes with gnome-settings-daemon" [Undecided,New] https://launchpad.net/bugs/85555713:35
seb128chrisccoulson, bounce it to dx, or ted, ronoc is on leave13:35
seb128hum, ok who understand launchpadlib?13:35
seb128ubuntu_series = launchpad.distributions['ubuntu'].getSeries(name_or_version='oneiric')13:35
seb128print ubuntu_series.getPackageUploads()13:35
seb128that gives me13:35
seb128<lazr.restfulclient.resource.Collection object at 0x929fb4c>13:35
seb128what I can do from it?13:36
seb128version tries to iterate through it with a "for upload in ubuntu_series.getPackageUploads():"13:36
seb128but that seems to not work13:36
pittiiteration usually works13:36
pittiwhat doesn't here?13:37
seb128ubuntu_series = launchpad.distributions['ubuntu'].getSeries(name_or_version='oneiric')13:37
seb128for upload in ubuntu_series.getPackageUploads():13:37
seb128    print "unapproved:", upload13:37
seb128 13:37
seb128-> empty output13:37
seb128pitti, ^13:37
seb128it seems to hang13:37
pittiseb128: erm, just getPackageUploads()?13:38
pittiisn't that going to give you a bazillion results?13:38
seb128pitti, ubuntu_series.getPackageUploads(status='Unapproved'):13:38
seb128when I dropped the status to see if that was the issue13:38
pitti>>> up.total_size13:39
pittithat seems fine13:39
pittiand iteration works here13:39
seb128ubuntu_series = launchpad.distributions['ubuntu'].getSeries(name_or_version='oneiric')13:39
seb128for upload in ubuntu_series.getPackageUploads(status='Unapproved'):13:39
seb128    print "unapproved:", upload13:39
seb128-> empty13:39
seb128for me13:39
seb128like it returns without printing anything13:39
pitti>>> for u in up:13:40
pitti...     print u.display_name13:40
pittiand just printing u gives me "https://api.launchpad.net/1.0/ubuntu/oneiric/+upload/3198332" [...]13:40
seb128pitti, http://pastebin.ubuntu.com/694454/13:41
seb128is what I'm trying to run13:41
pittiseb128: it seems it doesn't work with login_anonymously()13:41
pittidarn, seems we need to login authenticated again13:41
pittii. e. login_with(), and supply a cookie the first time13:42
pittiI don't know13:42
seb128I don't want to start having to maintain a cookie for version13:42
seb128I guess I should ask on #launchpad, seems a bug13:42
seb128pitti, well in any case that's why the versions queue use is broken13:43
pittiah, ok13:43
seb128since that's what it does to list items in the queue13:43
pittiseb128: we could just leave the cookie on ~/.launchpadlib/ on people, and give it a readonly priv?13:43
seb128pitti, wfm13:43
seb128but we need an account to use?13:44
seb128or do you want to use the retracer one?13:44
pittiI'm happy for it to use my creds, as long as it's readonly and public13:44
seb128pitti, want to set it up?13:44
pittiseb128: sure13:44
pittiseb128: done, next cron job should get it13:51
pittiin 15 ins13:51
seb128pitti, danke! ;-)13:51
cyphermoxseb128: would I need an FFE for the libnm-gtk split (what is needed for g-c-c) ?14:04
seb128pitti, ^14:04
cyphermoxI'm about to write one for NM too.14:04
pitticyphermox: they split a source pakcage?14:04
pittiseems fine14:04
pitticyphermox: any difference in the binary packages?14:04
cyphermoxwait no14:05
cyphermoxthey split the binary package14:05
pittii. e. source only split, or does it create new library packages?14:05
cyphermoxnew library14:05
seb128is there an easier way to specify configure option to dh rules than using "override_dh_auto_configure:"?14:05
pittithat'll mean a transition with potentially many rebuilds, so we should have an FFE14:05
seb128i.e should I14:05
pittiseb128: I don't think there is14:05
seb128dh_auto_configure --disable-icon-update14:05
cyphermoxpitti: nothing was using any of it before :)14:05
seb128dh_auto_configure --  --disable...14:05
seb128pitti, ok, thanks14:05
cyphermoxpitti: it's been split to make wifi/mobile dialogs available to other apps instead of relying on nm-applet14:06
pitticyphermox: seems fine14:06
cyphermoxpitti: my ffe for NM is https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/85559214:18
ubot2Ubuntu bug 855592 in network-manager "FFE: NetworkManager" [Undecided,New]14:18
pitticyphermox: updated14:21
pittiseb128: http://people.canonical.com/~platform/desktop/versions.html> darn, no difference14:24
seb128pitti, "Last updated: Wednesday September 21 2011 13:50:06 +0000"14:24
seb128pitti, it was before your change14:24
pittioh, versions.py stopped running, weird14:24
* pitti runs in foreground, perhaps it crashed14:25
seb128maybe it's waiting for lplib authentification?14:25
seb128or something went wrong with the cookie14:25
pittino, I did that already, and checked14:25
seb128there is a reason I prefer anonymous ;-)14:25
pittitemporarily disabling cronjob (in case I forget)14:25
dobeypitti: can we go back to the old, non-broken, python-gojbect? :)14:26
pittidobey: what is broken?14:26
pittiseb128: running in foreground now; lp login etc. went fine14:27
dobeypitti: 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 bindings14:27
pittidobey: how was the latter any different?14:28
pittie. g. Soup didn't have static bindings, so you couldn't use it before either14:28
dobeypitti: back when we could mix static bindings and gi bindings, we could actually have software that works reliably14:28
pittithat actually never worked reliably14:28
dobeypitti: right, but we could mix static/gi with old gobject14:28
pittiit was a source of weird crsahes14:28
dobeyi'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 :-/14:30
dobeyand my particular points of pain right now are Unity and twisted14:32
dobeynot Soup14:32
pittidobey: you can still do it by importing all GI bits first, and then the static ones14:32
pittibut you get to keep both halves when it breaks14:32
pitti$ python -c 'import gobject; from gi.repository import Gtk'14:32
pitti-> ImportError('When using gi.repository blabla')14:33
pitti$ python -c 'from gi.repository import Gtk; import gobject'14:33
dobeyunfortunately not possible14:33
pitti-> no complaint14:33
pittibut even this simple one already breaks in glib14:33
pitti(Warning: g_boxed_type_register_static:..)14:33
pitti$ python -c 'from gi.repository import Soup; import twisted'14:35
pittidobey: ^ that particular combination might work?14:35
pittiah, twisted itself doesn't use gobject, that was just a particular submodule14:36
seb128chrisccoulson, 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 weird14:37
pittidobey: is there a particular library you need? building a gir is not exactly rocket science, so we can help upstream to build one14:37
dobeypitti: from twisted.internet.glib2reactor import install14:37
chrisccoulsonseb128, yeah, it seems to work ok here14:37
dobeypitti: the problem isn't that the .gir/typelib is missing; the problem is that glib/gio/gobject stuff isn't totally bound yet14:37
seb128chrisccoulson, hum ok, I'm testing on the 10v with an install I just did an hour ago14:37
chrisccoulsonseb128, actually, maybe not - https://launchpadlibrarian.net/80459240/farnsworth-oneiric-20110920-16.png14:38
chrisccoulsonthat doesn't make a lot of sense ;)14:38
seb128your changes in the ppa won 1s on the boot there14:38
chrisccoulsonnice :)14:38
chrisccoulsonseb128, is that a fresh install?14:38
dobeypitti: or well, i'm trying to port the glib2/gtk2 reactor to gi/gtk3 APIs with gir, but alas, lack of upstream bindings14:38
seb128chrisccoulson, well "fresh", rebooted 10 times with some minutes in session14:38
seb128chrisccoulson, i.e it should have done the ureadahead caching14:39
chrisccoulsonthere's still the case where it regresses if you've ever changed your monitor settings, which sucks14:39
chrisccoulson(ie, it adds back another reprobe)14:39
chrisccoulsonwhich there isn't really any need for :)14:39
seb128it's "down" to 26.8s on the 10v14:39
chrisccoulsonseb128, what was it in lucid?14:41
chrisccoulsonstill a way to go then ;)14:42
seb128but it was 37s 1 week ago on oneiric14:42
seb128so 10.2s in a week is already something ;-)14:42
seb128it's not only us14:43
seb128gnome-session used to start 6s into the boot, it starts at 9s now14:43
chrisccoulsonah, so it's not all our fault then ;)14:43
seb128I need to debug nautilus14:44
seb128it was taking 3 seconds of cpu, it's taking like 10s now14:44
seb128and I've moved all the .so out, so it's not the u1 integration or something else being stupid14:44
chrisccoulsonouch, that's quite a bit14:44
seb128the other offender is compiz,unity14:45
chrisccoulsonyeah, i think those are a lost cause for oneiric ;)14:45
seb128also we use to run the wm directly14:45
seb128not delay it to after g-s-d14:45
seb128i.e on http://people.canonical.com/~pitti/bootcharts/daniel-lucid-20100305-1.png14:45
seb128lot of small things taking 1s as well14:46
seb128the helper to detect if we should run 3d or not14:46
pittiseb128: hm, who knows -- http://people.canonical.com/~platform/desktop/versions.html worked now14:46
seb128zg taking over a second cpu14:46
seb128pitti, looks better ;-)14:46
didrocksseb128: the helper souldn't take any time now, (at least, you should pay the price before gnome-session is there)14:47
seb128didrocks, it's autologin14:47
didrocksseb128: even with autologin14:47
pittiseb128: indeed, much more useful again \o/14:47
didrocksseb128: it's run *before* gnome-session14:47
seb128didrocks, so it's buggy, it takes 2s of blue bar14:47
chrisccoulsonso, we'll all be discussing boot speed at UDS then ? :)14:47
seb128didrocks, not on my charts14:48
pittiseb128: ah, it marks the unapproved ones with a little "Q"14:48
seb128pitti, right14:48
didrocksseb128: can you upload your chart?14:49
seb128didrocks, yeah, I'm copying them over14:49
didrocksif you are on autologin14:50
didrocksyou should see it running twice14:50
didrocksone before gnome-session, paying the price14:50
didrocksand a short time after gnome-session (just checking a file)14:50
didrocksseb128: ls /tmp/unity_support_test*14:51
pittididrocks: oh, I was going to ask you about this -- what creates /tmp/unity_support_test.0 ?14:51
pittididrocks: it's a predictable file name, so whatever creates it must be *very* careful to not make this a symlink attack vector14:51
didrockspitti: unity_support_test creates it :)14:52
didrockspitti: hum, it runs under the user or lightdm name, indeed14:52
pittididrocks: does it need to be in /tmp/ ?14:52
didrockspitti: it needs to be somewhere where it's removed at startup14:52
seb128didrocks, doh, ok, my fault14:52
didrockspitti: basically, that's the speed improvment I made last week i discussed on the team report14:53
seb128didrocks, there is a gnome-fallback... on the chart, I though it was the fallback 3d session upstream thing14:53
didrockspitti: when unity-greeter is running, the support tool test is ran and drop that file depending on the result14:53
seb128but it seems it's gnome-fallback-mount-helper from g-s-d14:53
didrocksseb128: ahah! :-)14:53
seb128didrocks, pitti, chrisccoulson: http://people.canonical.com/~seb128/oneiric-Inspiron-1011-oneiric-20110921-6.png14:54
didrockspitti: then the scripts in /etc/X11/Xsession.d/ before logging try to pick it14:54
seb12810v beta2 chart with the desktop team ppa14:54
pittiseb128: 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.png14:54
pittiseb128: however, the desktop PPA should make little difference14:55
seb128why does it call modprobre and blkid in serie?14:55
pittias both CPUs are fully used all the time, the extra wait time doesn't matter14:55
seb128pitti, it makes a ~1s difference there14:55
didrockspitti: 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 reboot14:56
pittididrocks: so it can't go into ~, and it needs to be predictable14:57
didrockspitti: knowing that it can be run from multiple users on the machcine (and even lightdm)14:57
pittididrocks: and needs cleaning up on boot?14:57
didrockspitti: indeed, if you install a new driver or whatever14:57
pittididrocks: then the only locations are /tmp/ and /run/lock/, both of which are world-writeable14:57
didrocksI picked /tmp for that reason14:57
pittididrocks: so we indeed need to be careful; I'll have a look at the code, I guess it's in lp:unity?14:57
didrockspitti: no, it's lp:nux14:58
didrockspitti: my code is pretty silly, but yeah, I will leave a second look :)14:58
pittididrocks: I can just run that tool without harm, right?14:58
pittididrocks: what's .0, the $DISPLAY?14:58
didrockspitti: yeah, it won't hurt you, I promess!14:58
didrockspitti: no, it's the result (0 for ok, 1 for ko)14:58
pittiright, it's not user specific14:59
chrisccoulsonnice - http://chevrel.org/temp/Capture-trailer_400p.png :)14:59
didrockspitti: http://bazaar.launchpad.net/~unity-team/nux/trunk/view/head:/tools/unity_support_test.c14:59
pittididrocks: dang, it's vulnerable14:59
pitti$ ln -s pwned /tmp/unity_support_test.015:00
pitti$ /usr/lib/nux/unity_support_test15:00
pitti$ ls /tmp/pwned15:00
pittididrocks: I'll create a branch15:00
didrockspitti: thanks, can push it in next nux release15:00
didrockspitti: I just fopen(resultfilename, "w+") and then fclose it15:00
pittididrocks: right, you need to use the Unix open() with O_CREAT|O_EXCL15:01
pittithe latter is the important bit here, to atomically guard against symlink attacks15:01
didrockspitti: ok, not really good at security on this, will look at some documentation with your branch :)15:02
didrocksok, basically, the idea is to fail is the file is already there?15:02
pittididrocks: yes15:03
pittididrocks: the attach scenario is that you can overwrite/kill any file which is owned by the user who calls support_test15:03
seb128didrocks, 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
pittii. e. lightdm or you15:03
pittididrocks: you can e. g. point it to your keyring, or the lightdm configuration, etc.15:03
pittididrocks: I'll create a branch/MP15:03
didrockspitti: oh ok, so some kind of "nullify" attack15:03
didrockspitti: 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
didrocksseb128: will do15:04
seb128didrocks, thanks15:04
didrocksseb128: still some compiz churn, but will be ok afterwards, I hope15:05
seb128didrocks, no hurry, I doubt we will unfreeze before friday morning15:05
seb128didrocks, but feel free to claim on the pad for later ;-)15:05
didrocksyeah, I'm afraid of that too, I'll try have some iso testing as well tomorrow morning15:06
mdeslaurdidrocks, pitti: is there a possibility of someone malicious creating the file and breaking the GUI on a multiuser system?15:06
didrocksmdeslaur: well, define "breaking"15:07
mdeslaurdidrocks: ie: saying the hardware supports 3d, when in fact it doesn't, so desktop sessions don't come up?15:08
didrocksmdeslaur: the support test tool will say "go ahead", then unity needs to fallback if it can't load, which is a whishlist bug opened15:08
didrocksseb128: I'm maybe blind but I don't see the unity_support_tool being run on your bootchart?15:09
seb128<seb128> didrocks, doh, ok, my fault15:09
seb128<seb128> didrocks, there is a gnome-fallback... on the chart, I though it was the fallback 3d session upstream thing15:09
seb128<seb128> but it seems it's gnome-fallback-mount-helper from g-s-d15:09
seb128didrocks, ^15:09
didrocksseb128: yeah saw that, but still, we should it running15:10
mdeslaurthe file should be in a non-user writable location, and maybe the timestamp could be checked to see if it's stale or something15:10
seb128didrocks, well in the session it's normal, if it just checks a file it's too short to register on the chart15:10
didrocksmdeslaur: the file is created by the tool runned by the user though15:10
didrocksseb128: but that means that the initial check is fast as well?15:10
mdeslaurdidrocks: I thought you said the tool is run by unity-greeter?15:11
didrocksmdeslaur: ok, so the tool *can* be run by the unity-greeter15:11
didrocksmdeslaur: there are two cases:15:11
didrocks- 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 /tmp15:12
seb128didrocks, I don't really know, need to check, it seems weird indeed15:12
didrocksthen, Xsession.d (under the new user) check if the file is there, and run the session15:12
seb128didrocks, it shows on my laptop manual login chart: http://people.canonical.com/~seb128/seb-e6410-oneiric-20110920-9.png15:13
didrocksmdeslaur: if the user is using something else or autologin, the Xsession.d won't find the file and will run the tool then15:13
didrocksseb128: hum, interesting15:13
seb128didrocks, it's running, the proc_ps.log has it15:15
seb128didrocks, it's like it was exiting immediatly15:16
seb128didrocks, maybe lightdm close and destroy the process directly in autologin15:16
seb128didrocks, I think lightdm does clean behind him15:16
mdeslaurdidrocks: ok...hrm...15:16
seb128didrocks, lightdm is too light, the helper doesn't have time to run ;-)15:17
didrocksseb128: hum, try logging in slower please :)15:18
didrocksseb128: oh, what you can do15:18
seb128didrocks, it's *auto*login :p15:18
didrocksseb128: reboot, start unity-greeter15:18
didrocksah, you mean, in the autologin case, not the bootchart :)15:18
didrocksso, in the autologin case, lightdm doesn't start it15:18
seb128didrocks, it does15:18
seb128it shows with one tick in the bootchart log15:19
didrocksseb128: unity-greeter is run?15:19
seb128it's autologin15:19
didrocksso, it's not lightdm15:19
seb128no greeter15:19
didrockswell, it's lighdm in same way15:19
didrocksbut not *the* lightdm :)15:19
didrocksit's run by /etc/X11/Xsession.d/50_check_unity_support15:19
didrocksso, it should do the full check and drop the file15:19
seb128well it drops the file15:21
seb128didrocks, it seems fast, it takes 0.1s (user time) (0.04s system time) to run on a session after flush caches15:23
seb128ie echo 3 >  /proc/sys/vm/drop_caches15:23
seb128didrocks, so maybe it's fast enough to not be picked by the bootchart, the graphic filter out "noise"15:24
pittiopen("/tmp/unity_support_test.0", O_WRONLY|O_CREAT|O_EXCL, 0666) = -1 EEXIST (File exists)15:24
didrocksseb128: weird that it's slower when the greeter is there though15:28
seb128didrocks, it's not the same box, the greeter login is my laptop not the 10v15:28
seb128could be that the intel probing is faster on the 10v15:29
seb128quite some people have slow probing issue on modern intel hardware it seems15:29
didrocksseb128: ok, so it seems that the cache (for your laptop) is still useful as winning 0.5s maybe?15:29
seb128didrocks, right15:30
rodrigo_desrt, when binding a GSettings key to a GtkAdjustment value that changes quick, is gsettings clever enough to just not write for every change?15:33
pittididrocks, mdeslaur: https://code.launchpad.net/~pitti/nux/symlink-attack-fix/+merge/7643415:34
mdeslaurpitti: looks ok, thanks15:39
pittiso, good night everyone!15:40
didrocks+1 as well then :)15:40
rodrigo_bye pitti15:40
mdeslaurpitti: good night!15:40
didrockspitti: merging, thanks a lot! good night15:40
seb128Laney, 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
ubot2Gnome bug 657180 in General "tomboy blocks shutdown" [Normal,Unconfirmed]15:44
desrtrodrigo_: no.  it's not.15:44
rodrigo_seb128, I am on #tomboy all day yes, but don't participate much15:45
desrtrodrigo_: it's a difficult problem to solve15:45
rodrigo_but yes, sure, asking now15:45
seb128rodrigo_, thanks15:46
rodrigo_desrt, right, I'm seeing evolution has a gconf-bridge object that does the binding thing that gsettings does15:46
rodrigo_desrt, they use a timeout, which not sure it's the best way15:46
desrtrodrigo_: a good approach may be to introduce a flag at the binding API layer15:46
desrtRATE_LIMIT or so15:46
rodrigo_yeah, like _DELAYED?15:47
desrtor maybe apply the rate limiting by default15:47
desrtand have a flag to turn it off15:47
desrtit seems like a problem that's unique to property bindings15:47
desrtand solving the problem at lower layers is substantially more complicated15:47
desrtrodrigo_: i welcome a bug15:47
rodrigo_desrt, I'll file one, yes15:47
rodrigo_desrt, https://bugzilla.gnome.org/show_bug.cgi?id=65973815:51
ubot2Gnome bug 659738 in gsettings "Need a way to tell GSettings to delay writes for rapidly changing properties" [Normal,Unconfirmed]15:51
rodrigo_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?15:51
desrtrodrigo_: it's a performance hit15:52
desrtparticularly on ext4 with spinning disks15:52
desrtfsync() take a decade there15:52
desrtand dconf does fsync() on each write15:52
desrt(because it has to)15:52
=== tkamppeter_ is now known as tkamppeter
desrtso 100 changes -> 100 fsync() round trips.  each takes like 40ms15:52
desrteven though it's async, your disk ends up active for a few seconds after you stop15:53
desrtthat's bad15:53
seb128tedg, hey15:53
seb128tedg, can you make sure bug #854292 and bug #855557 are assigned and tracked for Oneiric?15:53
ubot2Launchpad bug 854292 in indicator-session "indicator-session disappear after running update-manager" [High,Confirmed] https://launchpad.net/bugs/85429215:53
ubot2Launchpad bug 855557 in indicator-sound "indicator-sound-service interferes with gnome-settings-daemon" [High,New] https://launchpad.net/bugs/85555715:53
desrtseb128: so the mime database stuff in glib...15:53
desrtseb128: 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
seb128desrt, well as commented on the bug the user bug is still the same15:54
seb128desrt, open the info capplet and see .local/share/applications/mimeinfo.list being created with random values15:54
seb128desrt, 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 specific15:55
seb128desrt, but the bug might be somewhere else...15:55
desrtseb128: odd.15:56
desrtseb128: i'll look again, in any case15:56
desrtsince lxde and xfce are impacted as well15:56
seb128well, even for GNOME is seems a not-so-nice bug15:56
seb128open the control center and get gwibber to open photos, gedit to open calendars and another web browser than the one you used ;-)15:56
mvohm, 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 11.1015:59
desrtseb128: so i do as you say, and i get a file16:02
desrtbut the contents of the file are pretty reasonable16:02
desrtjpegs open with eog, videos with totem, oggs with rhythmbox, etc.16:02
seb128desrt, http://pastebin.ubuntu.com/693740/16:05
seb128that's the file I get16:06
seb128desrt, 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 ok16:06
* desrt sort of wonders why merely opening the panel and doing nothing at all is enough to cause a file to be generated16:07
seb128yeah, that's a bug as well in my opinion, bastien seemed to disagree it's one though16:07
desrtreminds me that i have a patch to commit to g-c-c16:07
seb128same, I don't see why it creates "added associations"16:07
rodrigo_out for a bit, bbl16:18
didrockshave a good night everyone16:31
skaetrodrigo_,  seb128 - what is the outlook on bug 832603?   Am stumbling into it on a fresh update.17:00
ubot2Launchpad bug 832603 in gnome-settings-daemon "gnome-settings-daemon crashed with SIGSEGV in g_simple_async_result_complete()" [High,Triaged] https://launchpad.net/bugs/83260317:00
* skaet does fresh update from natty, and booting oneiric, this is first thing seen. 17:01
seb128skaet, rodrigo seems to not be online, I don't personally know but I will check with him when he's there17:01
skaetseb128, thank you.   Anything you think he might want gathered from the machine before I reboot?17:01
seb128skaet, is g-s-d down, like no theme, etc or do you just get the apport prompt but things work normally?17:01
seb128skaet, add your .xsession-errors to the bug if you can please17:01
seb128skaet, oh and could you check the timestamp of the .crash and see if it correspond to the login?17:02
seb128just to make sure it's not a segfault from the old version on logout before you restarted17:02
skaetg-s-d down,  no theme.   Its fresh install update from clean installed and updated natty.   First boot triggered.17:03
seb128ok, so please add the .xsession-errors that should be enough17:03
skaetseb128,  will do.   Thanks17:03
seb128ok, that was "interesting"17:42
seb128g-s-d was not applied any icon theme, theme, etc17:43
seb128turn 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 greeter17:43
=== eeejay_is_afk is now known as eeejay
micahgseb128: can you please have someone look at bug 84537418:41
seb128micahg, will ping kenvandine tomorrow18:43
seb128but it got only 3 duplicates in 2 weeks18:44
seb128seems like a low priority bug, we get lot of bugs that get that in a day18:44
micahgseb128: really?  maybe people aren't using gwibber then, feel free to lower priority18:46
dobeywhen did GtkBox switch to not having children expanded by default with _add()?18:47
seb128micahg, that bug says you don't need to use gwibber18:48
seb128micahg, gwibber works fine there18:48
seb128micahg, but anyway, I will get kenvandine to look at it18:48
* micahg must be subscribe to weird people with long links then...18:49
seb128micahg, do you use ecryptfs?18:49
micahgseb128: yes18:50
seb128micahg, well maybe it's due ecryptfs filenames limitation18:50
seb128users tend to not run ecryptfs ;-)18:50
micahgseb128: orly, maybe I should have our ecryptfs guy take a look (tyhicks)18:50
seb128well it's still a bug, but it wouldn't be the first thing to be hit by it18:50
seb128evolution has issues due to it as well18:51
seb128the limitation is due to some patents iirc, so not something ecryptfs will fix18:51
kirklandseb128: that's not quite true18:51
kirklandseb128: actually for 12.04, tyhicks and jjohansen are support to fix this18:51
seb128ok, I should better not speak about what I don't know :p18:51
mdeslaurthe filename in the bug isn't long enough to be a ecryptfs issue18:51
kirklandseb128: jjohansen had a functional, though non-perfect solution18:51
seb128kirkland, sorry, I though somebody told me that a year ago18:52
kirklandseb128: there was a slight concern about that, but we cleared it18:52
seb128great to read ;-)18:52
seb128mdeslaur, oh ok18:52
kirklandseb128: :-)18:52
seb128well as said, I will ask kenvandine to have a look when he's back tomorrow18:52
micahgseb128: thanks18:52
seb128it might just be a good old standard gwibber bug ;-)18:52
desrttedg: hey20:33
tedgdesrt, Howdy20:33
desrttedg: did you want to come to montreal?20:33
tedgdesrt, I have passed a request up the chain, but I haven't gotten a response.  Need to ping.20:34
desrttedg: i think olli mentioned that it might be useful for you to be there20:34
tedgdesrt, Yeah, and I had sent him justification.  But, I never got back a yes/no from him.  He's not online now.20:35
desrttedg: he asked me if i thought it would be worthwhile.  i said yes.... so here's hoping =)20:35
=== eeejay is now known as eeejay_is_afk
cyphermoxdesrt: btw, thanks for the invite (got it from jason), I'll definitely be there21:15
=== m_conley is now known as m_conley_away
jbichacyphermox: what do you think of the proposed fix for bug 852961 ?21:19
ubot2Launchpad bug 852961 in network-manager-applet "nm-applet is wrongly started with Gnome Shell" [Undecided,In progress] https://launchpad.net/bugs/85296121:19
cyphermoxjbicha: what should it be set to?21:21
cyphermoxjbicha: hmm.. maybe that's not such a bad idea, I can start to fix one of mpt's bug reports with this21:25
cyphermoxbbl, dinner21:41
=== eeejay_is_afk is now known as eeejay
desrtcyphermox: nice :)22:43
desrtcyphermox: be sure to add yourself to the wiki!22:43
desrtcyphermox: (and ask the others) https://live.gnome.org/Montreal2011/Participants22:44

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