=== asac_ is now known as asac
ArneGoetjepitti: typo. it's iain-farrell. Will change it later, when LP is out of read-only mode again.01:03
rickspencer3stupid read only launchpad ate the work items I added01:34
rickspencer3and I didn't save my work :(01:34
rickspencer3pitti, looks like I failed to get the work items on, I'll do it tomorrow01:34
faganrickspencer3: ouch01:36
* fagan is using vb at the moment which I hear rickspencer3 had some hand in making :)01:38
rickspencer3fagan, nah, not really01:38
rickspencer3I worked on the Visual Studio shell for a while starting in '9801:39
rickspencer3and did a lot of work with C# language V2 and the Visual Studio editor01:39
rickspencer3all as a usability engineer01:39
faganrickspencer3: Oh cool01:39
* fagan is just doing a crappy college project and would prefer to be using quickly01:40
rickspencer3of course01:40
rickspencer3launchpad emailed me a diff before it blew away my changes01:48
rickspencer3still read only though, I'll add 'em back tomorrow01:49
rickspencer3good night all, see you tomorrow01:49
* Amaranth starts thinking extreme for compiz01:50
faganAmaranth: how extreme?01:51
* fagan is going to miss wobbly windows when gnome-shell comes in01:51
Amaranthload these plugins: core,move,resize,place,decoration,animation,ccp,gnomecompat,text,commands,wall,snap,expo,staticswitcher,regex,workarounds,fade,session01:51
Amaranthno more scale, zoom, dbus, etc01:52
AmaranthI'd cut expo from the defaults too, actually01:52
Amaranthperhaps regex so long as our default settings don't use it01:52
Amaranththen you can get rid of text too01:52
faganAmaranth: well cutting stuff that isnt used is always good :)01:52
Amaranthand you end up with a crappy metacity01:52
Amaranthfagan: I just cut everything that the metacity compositor doesn't implement01:53
faganmetacity is awesome01:53
AmaranthExcept we'd need an extra level in Visual Effects if we did this01:54
AmaranthBecause Extra should then change it back to what we were using before01:54
faganand it should be called awesome01:54
AmaranthBut in that case I dunno how much it'd actually help because we'd still have the plugins installed which means ccp reads the data for them and loads the gconf settings for them01:54
brycefagan, only if the monitor physically shakes when you use it01:54
faganbryce: I think thats compiz's drunk mode01:55
AmaranthI really hope splitting up the plugins into separate packages and only installing what we use gets compiz fast enough to keep it in01:55
Amaranthbecause if I have to backport the 0.9 plugin system upstream is going to hang me01:55
faganAmaranth: Well if gnome-shell is ready for lucid+1 then this will more than likely be the last release of compiz in ubuntu01:56
Amaranthfagan: From what I've heard I sort of doubt that01:56
Amaranthfagan: Probably lucid+201:56
* fagan doesnt think it will be ready for another year01:57
faganlucid+3 at the latest01:57
Amaranthat which point I start throwing together bits from XFCE and GNOME 2.x and make my own spin using compiz :)01:57
Amaranthoh, and gnome-do/docky :)01:57
faganAmaranth: http://shanefagan.com/2009/12/03/more-gnome-shell-messing-from-me/01:58
fagandocky + gnome-shell01:58
Amaranthgnome-shell still feels too much like it's getting in my way01:59
faganI like it, plus when you get used to the super key its very easy01:59
Amaranthand I disagree with the architecture01:59
AmaranthIt would not have been nearly as difficult as they say to have gnome-shell and mutter be separate02:00
AmaranthIf they did that I could make it work with compiz in no time02:00
AmaranthBut when I asked about it they told me to go use XFCE. I then worked on XFCE for a couple months and got commit access :P02:00
faganId say it wouldnt take a lot of work to get it using compiz but since shell is using clutter the window manager could use it02:01
faganAmaranth: I think when shell is released lots of distros are going to stay with panel02:01
AmaranthThe thing that makes me wonder if we're going to see gnome-shell used at all is the lack of 3D drivers for everything02:01
AmaranthWhat do you use if gnome-shell can't run?02:02
AmaranthThe funny thing is gnome-shell/mutter is dropping non-3D users while compiz git works right now without compositing and xrender compositing is planned02:02
Amaranthfagan: So we have to ship metacity, gnome-panel, and friends forever but they are completely unmaintained. Awesome.02:03
brycewhy do they call it 'gnome-shell' anyway?02:03
brycewhat is shelly about it?02:03
faganbryce: its a new UI shell for gnome02:03
AmaranthI dunno, they keep using the term "desktop shell"02:03
faganit makes sense02:03
AmaranthI know Windows calls the explorer process a "shell" so I guess they get it from there02:04
faganAll I know is that we need a redesign weather or not gnome-shell is right is debatable but at least its something new02:05
faganPanel is very dated02:06
bryceI would just like something which behaves itself when I do xrandry things02:06
faganI just want something a little more fault tolorent than panel. If panel crashes it trys to start itself again but if its a major crash it keeps trying to start again and makes the desktop completely unusable02:08
faganpanel is a pain when it crashes02:09
faganI think the crash problem is solved (slightly) by gnome shell not allowing applets embedded in the panel02:11
faganbryce Amaranth I would like if we just shipped ubuntu netbook edition as the default desktop because its a lot sleeker and a lot more stable than shell02:14
Amaranthfagan: Every app marked to be restarted automatically by the session does that02:16
Amaranthwhen working on some nautilus patches I actually exhausted the number of connections I could have to the X server due to nautilus restarting and crashing over and over and apport showing a dialog for each crash02:16
faganAmaranth: I know but its a real pain if you have a segfault02:16
Amaranthhad to restart X to fix it02:16
brycelet's port UNE to Wayland :-)02:17
Amaranthwayland is awesome :)02:17
Amaranthand actually still needs a "compositor"02:17
AmaranthIt does most of the work itself though so what it really needs is an app to implement effects02:17
faganWhats wayland?02:18
Amarantha display server written from scratch to take advantage of things like gallium, KMS, GEM, etc02:18
brycefagan, http://www.phoronix.com/scan.php?page=article&item=xorg_wayland&num=102:20
Amaranthit's fast, small, and completely flicker free02:20
Amaranthoh, and everything is client side02:20
bryceAmaranth, have you run it?02:21
Amaranththe server just composites pixmaps together and blits them02:21
Amaranthbryce: nah, never got a chance to get it all setup02:21
Amaranthbryce: Looks like you could do it in vmware or qemu now though02:21
bryceI'm working on getting it packaged02:21
bryce(off and on)02:21
faganbryce: awesome02:21
faganx.org is starting to look a little out of date02:22
Amarantheh, wayland is an experiment, really02:22
AmaranthIf anything you'll end up running wayland but all your apps will be running in rootless X servers02:22
bryceactually X.org is refactoring itself slowly towards modernity02:23
Amaranthor wayland will run under X and let you do neat transitions for user switching02:23
Amaranthbryce: Right, I was saying if wayland gets used at all by a wide audience that'll be how02:23
bryceAmaranth, yep, that's my thought too02:24
Amaranthwayland would help with boot times though ;)02:24
bryceanother use case I could imagine is for really limited functionality devices02:24
Amaranthhurry up and package it so we can switch :D02:24
brycelike maybe some of the arm stuff02:24
Amaranthgotta shave those seconds02:24
brycehey I found out that my failsafe-x mode boots X in <1 sec02:25
bryceso if we all just switch to vesa...02:25
ajmitchno thanks02:25
ajmitchthough I did barely notice for awhile that my desktop was using nv instead of nvidia02:26
bryceajmitch, wow interesting02:26
Amaranthajmitch: If you don't use compiz you probably won't notice at all when you're using nouveau02:26
bryceajmitch, well we'll break that for you once we get nouveau in02:26
AmaranthExcept things will actually be fast02:26
ajmitchAmaranth: I only noticed when I tried to turn compiz on :)02:26
Amaranthnouveau 2D acceleration is amazing02:27
brycedinner... bbl02:27
ajmitchperhaps I was just used to my desktop being slow anyway02:27
* fagan uses the nvidia proprietary drivers 02:27
lifelesscompiz? is african for 'make-my-desktop-laggy'  ?02:27
* lifeless bites his sarcastic fingers02:27
faganlifeless: but it looks nicer02:27
faganand notify-osd fades a lot better02:28
faganand alt+tab looks nicer02:28
lifelessyes, but I like being able to alt-tab quickly02:28
Amaranthnouveau makes using the metacity compositor pleasant02:28
faganand dont forget wobbly windows02:28
Amaranthlifeless: It's pretty fast here...02:28
ajmitchAmaranth: but will it play world of warcraft in wine yet? :)02:28
Amaranthajmitch: depends02:29
* fagan loves wobbly windows and misses them when he is in windows land02:29
Amaranthyou need the exact same chip and such as the developer who got it working, basically02:29
ajmitchI doubt that the memory management of nouveau's 3d driver is adequate for wow02:29
Amaranthdid you beat WoW yet?02:30
ajmitchthat would require me spending far too much time on it02:30
ajmitchso 'no thanks' :)02:30
faganAmaranth: Some guy got all but one achievement02:31
faganThat dude must have 0% life02:31
faganHe must be able to kick everyones ass02:31
* fagan gos back to making the single worst program ever made 02:32
fagan(college VB project)02:32
ajmitchdo you feel dirty yet?02:33
faganajmitch: I feel like I just ran threw a ditch02:34
faganplus its 2.3002:34
faganthe project is due in 12.30 hours02:34
faganand I have barely anything done02:35
ajmitchgood luck, and you should ignore irc for a few hours :)02:35
faganajmitch: I know im just on a tea break02:35
faganPlus I have taxi driver in the backround02:35
faganSo im happy02:35
=== eeejay is now known as eeejay_away
=== bjf is now known as bjf-afk
=== eeejay_away is now known as eeejay
pittiGood morning07:30
faganI found a little bug in open office thats just for the Irish local08:05
faganWe should be using the UK dictionary instead of not using any08:05
chrisccoulsongood morning everyone08:32
seb128good morning here08:39
chrisccoulsongood morning seb12808:42
seb128hey chrisccoulson08:43
chrisccoulsonhow are you today?08:44
seb128good, you?08:45
chrisccoulsonyeah, not too bad thanks. glad it's friday :)08:45
seb128my pc is beeping again today08:53
seb128ie loud and annoying speaker sound08:54
seb128it's half pitti's fault half mine on this one09:03
* seb128 fixes now09:03
seb128pitti, you didn't commit your libgnome change to bzr09:03
seb128and I didn't notice the revision was not there when merging09:04
seb128so the "turn bell off" got dropped09:04
pittioops, I'm sorry09:04
seb128we really need hooks to check that the bzr version match the archive one09:04
seb128I forget to push sometimes too...09:04
chrisccoulsoni've been caught out before by people forgetting to push too :)09:05
chrisccoulsonalthough, i should check the versions match too09:05
seb128bryce, pitti: the xorg syncs breaked installability for gtk...09:24
pittioh darn, how?09:24
seb128depwait on libx11-dev (>= 2:
seb128and we have09:24
seb128“libx11” 2:1.2.2-1ubuntu1 source package in Ubuntu09:24
seb128pitti, things don't build because our libx11 is too old09:24
seb128we are sort of in the middle of a transition09:25
pitti-> #u-devel09:25
seb128nothing desktopish build now09:25
pittiI need to disappear for some 2 hours09:33
seb128pitti, see you later09:34
seb128maybe archive will be fixed by time you come back ;-)09:34
seb128hey Keybuk09:45
seb128ok, let's try the daily dist-upgrade and look at login10:46
pittiseb128: should bug 460535 be assigned to slomo instead of you?12:16
ubottuLaunchpad bug 460535 in gst-plugins-base0.10 "Unable to play any but the first track of CD" [Medium,Fix committed] https://launchpad.net/bugs/46053512:16
sabdflseb128: want to talk about the bottom panel piece, i can chat now if you want12:33
seb128just back from lunch12:36
chrisccoulsonpitti - did you mean to assign bug 491162 to ubuntu-desktop (or desktop-bugs?)12:36
seb128hey sabdfl12:36
ubottuLaunchpad bug 491162 in gdm "gdm does not start X unless remove "tty-device-added KERNEL=tty7" from upstart gdm.conf" [High,New] https://launchpad.net/bugs/49116212:36
seb128pitti, ^12:36
pittihey sabdfl12:36
seb128sabdfl, was your concern about the 2 bars against 1? or about a particular applet?12:36
pittisabdfl: so the original idea was to drop some of the less popular applets and the starters, and merge the two panels (an idea which has been thrown around for a loong time already, but we never got to it)12:37
pittithis particularly helps with smaller screens like laptops12:37
pittiso you want to keep the two panels?12:38
pittichrisccoulson: I didn't assign it in the first place; not sure whom to give it to yet12:38
chrisccoulsonpitti - ah, you're right, it was bryce i think. i thought it was you because i only got an e-mail when you changed the milestone (and not when it was assigned)12:39
chrisccoulsonsorry ;)12:39
seb128pitti, does it need to be assigned at all?12:41
pittior unmilestoned12:41
seb128pitti, it will be in the next upstream version so in lucid12:42
pittiI gave Keybuk a quick ping, so far I don't understand it at all yet, I'm afraid12:42
pittiseb128: oh, you mean gstreamer, sorry12:42
pittiseb128: we just usually fix SRUs in the dev release first, that's why I milestoned it12:42
seb128I prefer to stay on sync with Debian12:42
seb128I'm fine being assigned to it12:43
seb128milestone for beta rather if you want12:43
* pitti moves to alpha-212:43
* fagan thinks 2 panels are nice12:43
Keybukseb128, pitti: I replied on ubuntu-devel12:43
KeybukI'd rather not touch that line right now12:43
pittiKeybuk: thanks12:43
Keybukbecause you might hide the problem12:43
KeybukI get failure to start gdm right now most boots ;)12:44
Keybukand I can definitely say it's not that line12:44
Keybukbecause I can debug upstart, and it says that match is TRUE :p12:44
Keybukthere's certainly a bug because you removed the "and started hal"12:44
KeybukI've had a few failures where gdm runs out of respawns12:44
Keybukbut that's not it either12:44
Keybukso far my *theory* is that the framebuffer (graphics) device shows up12:44
Keybukbut X wants the DRM device12:44
Keybukand that hasn't shown up yet12:45
Keybukand again, you run out of respawns12:45
* pitti hopes the answer isn't "put back hal" :)12:45
pittiI take it the "on hal" just delayed the startup far enough for the fb to get active?12:46
KeybukX genuinely needs HAL :p12:48
Keybukyou didn't fix that12:48
pittiKeybuk: ... at it12:48
Keybukbryce could upload X with those patches, of course12:48
pittithe libraries were synced last night12:48
tjaaltonwe are close..12:48
KeybukHAL MUST DIE!12:49
pittithose are needed first12:49
tjaaltonpitti: reminds me, two more syncs; libxcb and libxtst12:49
tjaaltonand libxvmc12:49
pittitjaalton: don't bother about filing bugs; doing right now12:49
pittitjaalton: experimental or unstable?12:49
tjaaltonpitti: done already :) libxcb - unstable, libxtst - exp, libxvmc - unstable12:50
Keybukpitti: I think we're just genuinely missing something12:50
Keybukit's on my plate to debug12:50
* fagan hugs pitti for killing hal hitman style12:50
Keybukand will probably look into that next week during alpha 1 freeze12:50
pittitjaalton: synced; can you please close the bugs?12:51
tjaaltonpitti: yep, on it12:51
pittithanks to you12:51
pittiRiddell: want to add some Kubuntu bits to https://wiki.ubuntu.com/DesktopTeam/ReleaseStatus for today's release team meeting?13:09
sabdfli got drawn into something else13:35
sabdflwe do have a design plan to get to one panel, but in order to do that we have to provide substitutes for the functions in the bottom panel today13:35
sabdflthose will likely be here for 10.10, but not 10.0413:35
sabdfli don't want two panels, but I don't want to bastardise the top panel either, and in the interim the plan is to live with two till we have good solutions13:36
pittisabdfl: we thought to move the task switcher into the top panel13:36
pittisabdfl: ok, that's fine13:36
pittisabdfl: having two panels isn't actually a speed issue, we just thought that when we are at reducing the applets anyway, we can just as well get to that13:37
pittibut let's defer that then13:37
sabdflok thanks13:37
sabdflwith Unity, trash is in Places, the "what's running" is handled by the launcher, and desktop reveal is part of task switching13:38
pittiso those two should stay, too?13:38
=== robbiew-afk is now known as robbiew
seb128pitti, the plan we had was to move task list and workspace switcher to the one bar13:57
seb128and drop trash (which is in nautilus places anyway) and show desktop button13:57
seb128we changed the clock to not display the date too13:58
seb128that made the one panel fit ok on small screens and reduce some login load due to the extra applets we cleaned13:58
seb128sabdfl, ^14:00
seb128pitti, sorry that was not for you ;-)14:00
istazsabdfl: pitti doest it mean 10.10 won't be using gnome-shell by default?14:00
seb128istaz, that has not been discussed yet, gnome-shell is not ready for that let's see how it goes in the next 6 months14:01
seb128istaz, right now gnome-shell works only on machines with fast graphic and compositing and has no fallback and we will need to solve that issue anyway14:02
seb128istaz, but that's discussions for next cycle14:02
seb128in any case we will need to figure something for installs which don't allow running gnome-shell14:03
istazI will be quite happy if you are still going to provide support for my older computer which won't support gnome-shell :)14:03
seb128see ;-)14:03
tedgkenvandine: Could you set up the bustle data collector to start with the session start so then we could get all the dbus traffic to align with the bootcharts?14:31
kenvandinei haven't looked at it yet14:31
kenvandinei am still trying to get that patch into gtk :/14:32
kenvandinewith 2.19.1 it fails make check...14:32
kenvandineso fixing those tests now14:32
tedgkenvandine: I went to the session at GCDS on it, and it's cool.  I'm not sure how to put it "really early" into gnome session though.14:32
kenvandinetedg, yeah, i'll look at that14:32
kenvandineunless seb128 wants to take a stab at it on the target hardware :)14:32
kenvandinei also want to use it to look at gwibber14:33
tedgkenvandine: Don't do that, it'll make you cry ;)14:33
seb128kenvandine, hey, how is empathy 2.29 going btw?14:33
seb128kenvandine, I don't know how to do this dbus thing14:33
kenvandineseb128, haven't touched it yet... soon though :)14:34
kenvandineseb128, i haven't looked at bustle yet14:34
kenvandinewell, read about it14:34
seb128ok, me neither14:34
tgpraveen11Hi all14:34
kenvandineseb128, i need to get the docs building for the app indicators too14:34
tedgseb128: http://www.willthompson.co.uk/bustle/  basically  "bustle-dbus-monitor >foo.bustle" will basically dump all the stuff off the session bus into "foo.bustle"14:34
seb128and I'm busy with other changes14:34
tgpraveen11Sup guys?14:34
seb128I'm fine doing testing but I'm too busy to learn what to do14:34
seb128ie give me instructions or look for somebody else14:35
kenvandineseb128, this gtk patch has turned into a pile of work :/14:35
kenvandineseb128, i'll figure out a way to make it start early and you can just do it :)14:35
seb128kenvandine, I was expected it to be some work indeed14:35
seb128kenvandine, ok thanks14:35
* kenvandine finally moved the actually building of gtk to another box... so at least my laptop doesn't overheat, this weekend i really need to take it apart and check it out14:36
kenvandinemaybe a lose heat sink or something14:36
* kenvandine thinks evolution filtering has gotten way slower14:37
seb128tedg, just curious why do you ask us to get those dbus infos where you could get those?15:04
seb128tedg, you don't get the cpu use on your box?15:04
seb128chrisccoulson_g1, chrisccoulson: what delay did you investigate exactly? the one before or after gnome-settings-daemon?15:05
seb128or both?15:05
chrisccoulsonseb128 - before gnome-settings-daemon15:05
seb128the delay before compiz is 2 seconds there15:06
seb1281 before g-s-d15:06
seb128one after g-s-d15:06
chrisccoulsonKeybuk - if a process loads at boot, and then at some point calls execvp() and turns in to something else, what process name will be shown in bootchart (or does it get a new line)?15:06
seb128I'm wondering what is the 1 second between seahorse and compiz15:07
Keybukchrisccoulson: the second process name I think15:07
Keybukit may be the first15:07
chrisccoulsonseb128 - that delay is just g-s-d loading isnt it?15:08
tedgseb128: I don't get as much.  But I haven't run a bootchart.  I wasn't looking at the bustle data as much for just me, I thought it'd be useful for the desktop optimization in general.15:08
chrisccoulsonKeybuk - thanks. i've been looking at the 1 second delay in gnome-session. I know where most of it is wasted now, but there is still a small amount unexplained15:08
chrisccoulsonbut then i just realised something :)15:08
chrisccoulsonactually, no, what i realised only applies to my own setup, as i have seahorse-agent installed15:09
chrisccoulsonseahorse-agent starts up and then calls execvp() and runs gnome-session15:09
seb128chrisccoulson: I don't know there is no color in the bar during that second15:10
seb128which is weird15:10
Keybukbootchart can be confusing here15:10
Keybuksometimes it helps to get the tgz and regenerate the bootchart with --no-prune15:10
Keybukthen you get a whole load more detail15:10
chrisccoulsonKeybuk - that's what I thought. so, i'm not sure gnome-session is _really_ gnome-session for some of that time:)15:10
seb128Keybuk, btw do you still have daily measures?15:10
chrisccoulsonbut i could be clutching at straws ;)15:10
seb128Keybuk, could you remember me the url?15:11
chrisccoulsonseb128 - yeah, it's wierd that there is no colour. what i need to do is some profiling in g-s-d really to work out how long it really holds the session up for15:12
chrisccoulsonbut i haven't got around to that yet:)15:12
Keybukseb128: http://people.canonical.com/~scott/daily-bootcharts/  ?15:12
chrisccoulsoni'm trying to get gconf to start early, but it's more difficult than i thought, as it needs the session bus15:13
chrisccoulsonbut, i'm sure i can make it work :-/15:13
seb128Keybuk, thanks15:13
seb128Keybuk, weird I don't have similar numbers15:13
rickspencer3didrocks, thanks for getting the work items into the quickly blueprints15:13
Keybukseb128: smaller with?15:13
seb128Keybuk, http://people.canonical.com/~seb128/bootchart/seb128-dellmini-lucid-20091203-1.png15:14
seb128Keybuk, it's over one second difference15:14
Keybukseb128: are you running with any patches?15:14
Keybukor PPAs etc.?15:14
seb128not that I know about15:14
seb128but I don't reinstall the box15:14
seb128I dist-upgrade daily15:14
Keybukyour X starts much faster15:15
Keybukyou're into the session by 10s15:15
Keybukso I am15:15
KeybukI see15:16
Keybukpitti: is that still not fixed?15:16
seb128ah right15:16
seb128Keybuk, thanks15:17
Keybukbecause mine are fresh installs every time, I get jockey penalty on that boot15:17
KeybukI also don't see dkms on yours, do you have the bcmwl thing installed?15:17
seb128are we supposed to measure standard boot or first boot after install?15:17
seb128first boot after install probably has some other extra work15:17
seb128like gnome-panel writting their config etc15:18
Keybukthis is second boot15:18
Keybukwhich is why the jockey-gtk confuses me15:18
KeybukI allow a first boot for everything to sort itself out15:18
seb128I don't think I've bcmwl installed no15:18
Keybukthat could be it too I guess15:18
Keybukwill grab pitti when he returns15:18
pittiargh, that15:19
sabdflseb128, i don't think we can drop trash, though we could probably improve performance (like do lazy detection of whether there's anything in the trash)15:24
seb128sabdfl, ok, I had the impression it was not so useful since you can open from nautilus easily but I've no strong opinion15:25
seb128sabdfl, the other reason to reduce the number of applet is that each one has a start cost for gnome-panel15:26
seb128sabdfl, we should have a look to that anyway but that will go down to the bonobo use at some point and we can't replace that easily15:27
seb128though the one slowing login most now is the indicator message one15:28
sabdflseb128: jump on Ted for that :-)15:33
sabdflhe promised me he'd make it *blindingly* fast15:33
sabdflso, hold him to that15:34
seb128right, I've annoying him for a few days about it now ;-)15:34
seb128I've -> I'm15:34
pittisabdfl: we make the trash applet so fast that you won't even see it!15:35
chrisccoulsonseb128 - we could add trash to the places menu, couldn't we?15:40
rickspencer3for trash, I was thinking that we could cache the current icon (full or empty) and then not init the trash until the user clicks on it15:40
pittithat wouldn't make too much sense to me15:40
pittithe sense for having it on desktop is to be a drag'n'drop target15:40
seb128chrisccoulson: wouldn't solve the issue we would still start the backend to display the icon15:40
pittiotherwise it's easier to just press delete or right-click15:41
rickspencer3but the problem with trash, as I understand it, is the init function to get the icon15:41
chrisccoulsonah, i wasn't aware15:41
chrisccoulsonseb128 - if it was in the places menu, you wouldn't have to start the backend until you opened the menu15:41
rickspencer3I would think 99.99% of the time the state would be the same as when the user logged off last15:41
seb128rickspencer3, wouldn't work, the trash is supposed to update when you select an icon on the desktop and press del15:41
pittiseb128: however, I don't have a trash applet, and gvfsd-trash is still running; I suspect that nautilus starts it anyway15:41
rickspencer3I don't want it to be so that when the user clicks on menu it doesn't open immediately15:41
rickspencer3that's not really a good expereince15:42
chrisccoulsonpitti - yeah, nautilus will start it when you move something to trash15:42
seb128anyway it doesn't cost so much on the stock install since the trash is empty15:42
seb128it costs for users who have things in the trash15:42
rickspencer3seb128, so can't the trash just init at hat point?15:42
seb128it could I guess15:43
rickspencer3seb128, I think that logic could probably be used by UNE, so could be a double win, desktop and netbook users would benefit15:44
didrocksrickspencer3: y/w :) (I told you yesterday that I'll work on that this afternoon ;))15:45
didrocksI have to run away, have a nice week-end!15:46
rickspencer3bye bye didrocks15:46
rickspencer3thanks again15:46
seb128didrocks, have fun15:47
rickspencer3kenvandine, where does gwibber store credentials currently?15:47
kenvandinekeyring i think15:53
kenvandineeww... gconf15:54
kenvandinethat needs to be fixed15:54
gorgapor how would I find out where the gnome-terminal color schemes are located, like tango, rxvt, xterm, linux console, etc? I'm sure they are some xml file somewhere, but I can't find them16:07
dobeykenvandine: yeah, i think i convinced segphault that we should store that stuff in keyring at uds16:11
kenvandinedobey, good16:16
chrisccoulsonseb128 - the latest GDM has debugging on by default now, and dumps loads of stuff in syslog16:16
chrisccoulsondo we want to switch that off?16:16
chrisccoulson(it might mess up the bootchart numbers) ;)16:16
seb128or new gnome-icon-theme uploaded16:17
seb128we didn't have an icon cache for the gnome icon theme16:18
seb128I noticed while tracing nautilus16:18
seb128it was opening some 360 times the outdated cache there16:18
seb128seems to win almost one second16:18
seb128we are around 11 seconds for desktop now16:18
chrisccoulsonawesome :)16:19
pittiseb128: *hug*16:19
rickspencer37 seconds to go16:19
seb128thank you guys ;-)16:20
seb128the bottleneck will be gnome-panel at some point16:20
seb128it takes longer than 4 seconds of cpu to start16:20
rickspencer3maybe we can cache images of gnome panel and defer loading them until the user clicks on them16:23
* rickspencer3 just kidding16:23
seb128it has quite some blanks in middle though16:23
seb128let's see what robert_ancell figure16:23
seb128he said he would investigate slowness there16:23
seb128not for the blanks specifically but for gnome-panel16:23
vuntzseb128: gnome-panel is blocking the login for 4 seconds?16:25
vuntzseb128: or it's active during 4 seconds?16:25
pittiAFAICS it uses 4 blocks of CPU, 6 seconds together16:26
seb128vuntz, time between gnome-panel start and stop of loading for it is some 11 seconds16:26
pittibut spread out across 11 seconds16:26
vuntzseb128: would be nice to see the impact of out-of-process applets (remove them), and in-process applets (also remove them), and the the applications menu (remove it) :-)16:27
seb128using a plain color background spare a 1 to 2 second cpu use for nautilus16:27
vuntzjust to know what is expensive16:27
seb128vuntz, most applets are around 0.3 seconds16:27
seb128or less16:27
seb128the menu one is around 2 seconds16:27
seb128or less now with gnome-menus caching16:27
seb128let me do some new charts about those16:28
seb128vuntz, I'm wondering if there the 0.3 seconds is basically the bonobo ping-pong cost16:28
vuntzwell, the interesting part would be "what's left when you remove everything from the panel" :-)16:28
vuntzdoes it take 5 seconds?16:28
seb128I'm just doing that one now16:29
seb128it's booting16:29
vuntzlovely, thanks!16:29
seb128you're welcome ;-)16:29
vuntzrickspencer3: you're lucky to have good slaves like seb128 ;-)16:29
rickspencer3vuntz, I think you have that relationship backward16:30
seb128vuntz, empty panel uses cpu for 1 second16:31
seb128let's add the clock and then the other in process in order16:33
vuntzrickspencer3: heh16:35
=== bjf-afk is now known as bjf
vuntzseb128: cool, that's a good data point. Still way too much, but I was afraid it'd be more ;-)16:36
seb128vuntz, http://people.canonical.com/~seb128/bootchart/seb128-dellmini-lucid-20091204-7.png16:39
seb128that is only the clock16:39
seb128and it's already interesting16:39
seb128you have loading16:39
seb128over a second of nothing16:39
seb128and then not so busy use16:40
pittiseb128: http://people.canonical.com/~seb128/bootchart/seb128-dellmini-lucid-20091204-5.png16:40
seb128ie doing some things for a while16:40
pitti-> SHIP IT16:40
seb128pitti, ;-)16:40
vuntzseb128: waiting for e-d-s, I'd guess16:40
pittiseb128: this still has gvfsd-trash16:40
pittiI think nautilus just starts it16:40
seb128pitti, right16:40
seb128vuntz, could we delay that until menu opening?16:41
seb128ie first click16:41
pittioh, doesn't it?16:41
seb128vuntz, that would also avoid the "something ask for keyring password at login"16:41
vuntzseb128: "let me click, hrm, why do I have to wait for one second for the click to work?"16:42
pittiit seems to me that the first click on the clock takes ages16:42
seb128seems that happens when you are private calendars16:42
pittiI guess since it's talking to google cal, etc.16:42
seb128vuntz, couldn't you start init on mouseover?16:42
vuntzmaybe, yeah16:42
vuntzin all cases, it should be delayed in some way16:42
pittiwell, at least until everything else is ready16:42
seb128that should be quick enough to not have that issue16:42
pittino reason not to start pulling calendars when the panel/nautilus are set?16:42
rickspencer3bryce, hi, are you available at 10:30 for the bughugger call?16:44
seb128vuntz, is there any extra work if weather is on?16:44
seb128seems to be half of that if I toggle those off...16:45
seb128those = weather and temperature options there16:45
seb128I would assume those would not add anything if there is no location set though16:45
vuntzseb128: yeah, the weather does some stuff, obviously16:46
seb128vuntz, but I've no location...16:46
vuntzhrm, interesting :-)16:47
pittidoesn't it default to "Eiffel Tower, Paris, France" if you don't have one set?16:48
* vuntz wishes he had more time to look at all this :/16:48
vuntzpitti: nah. I live in Grenoble!16:48
vuntzthat's a bit too visible to put by default, though ;-)16:49
seb128only the menus16:49
seb128it's still using quite a bit even with caching16:49
seb128and there is this long blank16:49
seb128I'm wondering if that's due to the 2 cores being busy16:49
* seb128 removed nautilus from session16:49
seb128hum no16:53
seb128vuntz, that's weird, it does load, then sit there doing nothing for a some seconds and then start loading the menu16:53
seb128I'm wondering what is the some seconds about16:54
vuntzthe menu is loaded in the idle loop16:54
seb128well it's idle16:54
seb128ie no activity16:54
vuntzit might be using 0.1% of the cpu to, say, draw itself on the screen16:54
seb128I'm wondering if we should build the menu in an agressive way16:55
seb128rather than during idle16:55
seb128Amaranth, hello16:58
seb128Amaranth, any chance you would upload that 1 liner compiz today?16:59
seb128Amaranth, we are sort on waiting on that for 3 days now...16:59
seb128or tell me if you are too busy I can do it16:59
chrisccoulsonhmmm, i can't access my folder at people.ubuntu.com. i get a permission denied error :-/17:02
chrisccoulsonseb128 - i might ask you to test some things i'm currently experimenting with shortly, if that's ok17:03
seb128chrisccoulson: sure17:07
chrisccoulsonseb128 - i've patched gconfd now to parse the sources when it loads, rather than waiting for a client to read a value. i'm hoping that with the gnome-session changes, and the g-s-d change i'm about to do, that the sources can be parsed without delaying the rest of the session17:09
chrisccoulsonbut i don't know how well that's going to work yet ;)17:09
chrisccoulsonseb128 - the patched gnome-session build you tested for me the other day loads g-s-d almost instantly btw17:10
chrisccoulsonwhen you look at http://people.ubuntu.com/~chrisccoulson/gnome-session-startup-time/tests/seb128-dellmini-lucid-20091130-6-gnome-session-improvement.png17:10
chrisccoulsongnome-session _really_ starts in line with dbus-launch there17:11
seb128we just moved the gconf delay on that one17:11
chrisccoulsonseb128 - yeah, that's what i'm working on to avoid now17:11
seb128btw do you think we could land your other optimization in lucid?17:11
seb128the dkpower one17:11
chrisccoulsonseb128 - yeah, i can do that17:11
seb128did you send it upstream btw?17:11
chrisccoulsonseb128 - do you want me to do that GDM change too? (to switch off debugging)17:13
seb128chrisccoulson: ok, I think the g-s-d 1 second delay is due to the xrandr option17:40
seb128or at least half of it17:40
seb128I'm wondering if it's doing some xorg queries it could avoid if there is no config on disk to apply anyway17:41
chrisccoulsonyeah, i'm not too sure yet, as I've not done much with g-s-d.17:43
chrisccoulsonit does start all the plugins before the rest of the session starts loading though17:43
chrisccoulsoni was going to patch it to load xrandr and xsettings (hardcoded, to avoid reading from gconf), and then let the rest of the session start whilst it loads the other plugins (using gconf)17:44
chrisccoulsonbut that will mean that users would never be able to unload xrandr/xsettings, but i'm not sure how much that matters really17:44
seb128I doesn't I would say17:45
chrisccoulsonthats the next thing on my list to work on anyway :)17:45
chrisccoulsonafter GDM is uploaded17:46
pittiseb128: jockey fix uploaded now17:49
pittishould avoid calling jockey17:49
pitti(except at first boot)17:49
seb128pitti, thanks17:49
seb128I didn't have jockey on my charts but scott did17:50
chrisccoulsoncool, gdm uploaded now17:52
chrisccoulsontime for some g-s-d hacking now17:52
seb128https://bugzilla.gnome.org/show_bug.cgi?id=586276 is interesting17:55
ubottuGnome bug 586276 in general "Don't spawn xrdb" [Normal,Unconfirmed]17:55
chrisccoulsonseb128 - i can possibly have a look at that too, if we think that xrdb is too heavy18:03
=== sabdfl1 is now known as sabdfl
seb128would probably be nice to clean but that's probably not the main issue we have now18:04
seb128chrisccoulson: http://mail.gnome.org/archives/performance-list/2009-April/msg00000.html18:34
seb128that might interest you as a reading18:35
=== dobey_ is now known as dobey
chrisccoulsonseb128 - thanks, i'll take a look at that shortly18:38
seb128http://mail.gnome.org/archives/performance-list/2009-March/msg00001.html too18:39
brycerickspencer3, http://bryceharrington.org/X/Data/18:41
seb128there is an interesting comment in this one18:42
seb128the login only need a part of the schemas and value18:42
seb128which leads to wonder if the xml could be splitted18:42
seb128one part read before loading in a quicker way or something18:42
seb128chrisccoulson: http://mces.blogspot.com/2008/11/improving-login-time-part-2-gnome.html too18:43
=== robbiew is now known as robbiew-afk
halflinechrisccoulson: hey, so i'm looking at bug 59847618:57
ubottuError: Launchpad bug 598476 could not be found18:57
halflinechrisccoulson: trying to figure out under what conditions this problem happens18:58
halflinei haven't been able to reproduce at all...18:58
halflinechrisccoulson: do you think you could get a trace with --sync ?18:58
chrisccoulsonhalfline - it wasn't reproducible all the time, even for me18:58
halflineright, but you'd think i would be able to make it reproducible 100% of the time by adding a sleep in the right place18:59
chrisccoulsonhalfline - there'a already a trace with --sync on the bug report18:59
halflineoh Trace 218314 IS with sync?19:00
chrisccoulsonit is :)19:00
halflineohhh yea, i guess XFreePixmap wouldn't normally cause a roundtrip19:00
halflinei'm basically just trying to get a good understanding on this issue, because it's probably going to have to go to vendor-sec etc as a CVE19:01
chrisccoulsoni need to have a quick look at the code again, because it's been a while since i looked at it19:02
halflineit's all very strange for me because...19:03
halflinethat gdk_pixmap_finalize is happening in end_implicit_paint19:03
halflinewhich makes me think it's the double buffer pixmap19:03
halflinebut that's not going to cause an X error19:03
pittigood night everyone, have a nice weekend19:06
halflinechrisccoulson: yea i'm definitely confused.  frame 6 in the trace is just calling XFreePixmap on a pixmap created from the local X connection...19:11
halflineit's not like a foreign pixmap or anything...19:12
halflinechrisccoulson: actaully19:13
halflinechrisccoulson: look at the XID passed to XFreePixmap19:13
halflinethat's a bogus number19:13
chrisccoulsonhmmm, i'm confused as well now :) Where would the bogus number come from?19:14
halflineit's the xid field of the gdk pixmap object19:16
halflineso i think we must be dealing with some sort of memory corruption here19:16
chrisccoulsoni am confused now then. i thought i understood it before19:16
chrisccoulsonbut the change we've got in ubuntu seems to fix it completely :-/19:17
halflinebut i wonder if it's because of a side effect of code churn19:18
halflineor changes the timings or something19:18
halflinedon't get me wrong your patch looks reasonable regardless19:18
halflinei just want to understand this issue and its implications19:18
chrisccoulsonyeah, it would be nice to understand it fully19:18
chrisccoulsoni'll try and take a look at it again at some point, but i won't get a chance tonight19:19
halflinei'm going to run it through valgrind and see if it says anythign19:21
=== bjf is now known as bjf-afk
=== eeejay is now known as eeejay_away
bigonwhat are the project for emapthy geolocalisation support? will lucid have map and geoclue support?20:15
brycedoes bzr have an equivalent to git show?20:17
brycee.g. I'd like to do `bzr show 321`20:17
brycebzr diff -r 321 doesn't quite do it20:17
bigonbzr diff -c ?20:18
brycebigon, aha thanks20:18
dobeybryce: or diff -r 1..220:36
brycedobey, ah, I tried -r 1:2 but it didn't like that20:41
=== bjf-afk is now known as bjf
=== eeejay_away is now known as eeejay
AmaranthI would have done a compiz upload if launchpad hadn't been down for 12 hours instead of 90 minutes21:50
rickspencer3bryce, looks like you json isn't quite valid, at least according to python json module21:56
brycerickspencer3, oh?  I generated it using simplejson fwiw21:57
rickspencer3would it be hard to format the pages so that they are on long list of dictionaries?21:57
brycerickspencer3, shouldn't be hard, no21:57
rickspencer3is simplejson python?21:58
brycecurrently they're formatted to be readible21:58
bryceimport simplejson as json21:58
rickspencer3well, I think the change would be the page would start with "[", and each }{ would be },{21:58
bryceprint json.dumps(records, indent=4)21:59
* rickspencer3 tries simplejson21:59
* rickspencer3 assumes it's loads21:59
brycewhich json file in particular are you looking at?21:59
rickspencer3picked at random22:01
bryceyep that one is generated by simplepython22:01
rickspencer3ValueError: Extra data: line 11 column 1 - line 78 column 1 (char 217 - 1708)22:02
rickspencer3I am assuming this is caused by finding the } and then WAIT, there's more???22:02
rickspencer3I'll try fidling with the format a bit22:02
rickspencer3but I've done lots of json on web clients, so I'm pretty sure this is it22:03
bryceI see what's wrong22:03
bryceI am not printing out the entire structure, but looping over each element and printing a json for it22:03
rickspencer3so pack them into a list and dump that22:03
rickspencer3then I can slurp it in22:03
bryceright, should be easy to fix, hang on22:04
=== robbiew-afk is now known as robbiew
brycethere we go, fixed22:09
brycetry now22:09
* rickspencer3 tries22:09
brycemost of the rest of the json files will update over the next hour22:10
brycethe upstream reports will update overnight; I opted not to hammer upstream's bugzilla more frequently than that ;-)22:10
rickspencer3[{'status': 'Confirmed', 'tags': ['upgrade'], 'importance': 'Wishlist', 'title': 'New upstream 173.14.18 available', 'package': 'nvidia-graphics-drivers-173', 'id': 361856}, {'status': 'Confirmed', 'tags': ['upgrade'], 'importance': 'Undecided', 'title': '[lucid] New nvidia 190.42 driver available', 'package': 'nvidia-graphics-drivers-180', 'id': 404135}, {'status': 'New', 'tags': ['merge', 'upgrade'], 'importance': 'Undecided',22:10
rickspencer3'title': 'Please merge nvidia-settings 185.18.31-1 (main) from Debian unstable (contrib)', 'package': 'nvidia-settings', 'id': 417410}, {'status': 'New', 'tags': ['upgrade'], 'importance': 'Undecided', 'title': 'Please update xfs to 1.1.0 version', 'package': 'xfs', 'id': 414272}, {'status': 'Confirmed', 'tags': ['8.10', 'intrepid', 'sis', 'upgrade', 'x.org'], 'importance': 'Undecided', 'title': 'after upgrade to 8.10 (new x.org)22:10
rickspencer3 SiS driver is not working well (issue with color depth?)', 'package': 'xserver-xorg-video-sis', 'id': 291294}]22:10
rickspencer3bryce, at your leisure22:10
* Amaranth uploads new compiz that does not depend on compiz-fusion-plugins-extra22:11
rickspencer3we would factor bug gravity into a separate module so that we can use that on the server22:11
rickspencer3would = should22:11
rickspencer3I'll try to do that while you are out next week22:12
bryceyeah I'm presently working on extracting brian's code to get a listing of team bugs22:12
Amaranthso according to seb128 my last two uploads of compiz should have shaved 2.5 seconds off login time :)22:12
rickspencer3I've got it so that you can add URL search and now the json parsing is working ;)22:13
brycebtw, xserver 1.7 is still on track for getting in by alpha-1 but it's currently blocked waiting on a few things being done in debian.  Timo's going to keep working on it next week22:13
brycerickspencer3, sweet :-)22:13
rickspencer3so I'll get it loaded into the bugpane next (probably this weekend)22:13
rickspencer3bryce, great about 1.7 (and thanks to timo)22:13
rickspencer3it occurs to me that the "Add a URL" UI needs to be a management console thingy, that's take me like 20 minutes to do with couchgrid!22:14
* rickspencer3 steps away from computer for an hour or so22:14
rickspencer3bryce, btw, the gravity system is in LaunchpadUtils.py in lp:bughgger if you want to give it a whack22:16
bryceok will do22:17
rickspencer3no worries if you don't get to it though22:17
bryceI just got "team-pkg-list <team>" done22:17
rickspencer3I'll just stub it out if you don't22:17
brycenow to see if I can pull some desktop stuff :-)22:17
bryceneed to figure out what everyone's team names are now22:17
* rickspencer3 will try when he gets back22:17
rickspencer3that's canonical-desktop-team22:17
rickspencer3there's code for that in bughugger22:18
rickspencer3I'll check when I get back22:18
rickspencer3bryce ... have a great and fun holiday if I don't catch you before you leave today22:18
bryceyeah I'll be on later22:19
Amaranthwhee, compiz _still_ fails to build on armel22:27
Amaranthbut the armel buildd is apparently idle because it told me immediately after uploading22:27
Amaranthstill KDE junk22:28
brycerickspencer3, if you're back yet... which format do you prefer?23:02
bryce1.  Data/<team>/<query>.json23:02
=== eeejay is now known as eeejay_away
bryce2.  Data/<query>/<team>.json23:03
bryce3.  Something else...23:03
brycefwiw, I've got #1 coded up currently for 'upgrade' bugs:  http://www2.bryceharrington.org:8080/X/Data/23:03
=== eeejay_away is now known as eeejay
=== eeejay_away is now known as eeejay
=== robbiew is now known as robbiew-afk

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