[01:03] <Sarvatt> johanbr: i'll add that patch and upload now, it'll probably take forever to get pushed to master
[01:03] <Sarvatt> then again it'll probably take 24 hours to build on launchpad too :D
[01:04] <johanbr> great, thank you
[01:04] <johanbr> I just finished building the package myself
[01:05] <johanbr> off to test if it works now :)
[01:12] <johanbr> Sarvatt, yep, the patch works fine... no more segfault
[01:17] <Sarvatt> glad to hear it, added it as a hook for auto-xorg-git too for the time being, its nice not having to manually edit anything before upload every new commit :D
[01:44] <Sarvatt> oops evdev 2.3.1 overrided git master, guess we're gonna have to fake the version number for master :D
[07:49] <tjaalton> upgraded, but the mouse and keyboard didn't work if I just logged out
[07:58] <tjaalton> and got failsafe after reboot
[07:58] <tjaalton> huh..
[11:19] <mac_v> if i add <  radeon.modeset=0  > to the kernel line , it would turn off KMS?
[11:20] <tjaalton> yes
[11:20] <mac_v> thanks :)
[11:25] <tjaalton> jcristau: so, we discussed with pitti about the udev rules and triggers. the drivers should really run udevadm, since for instance here xserver-xorg was configured before evdev, resulting in no kbd&mouse after a logout (sure, it yelled to reboot, but..)
[11:26] <jcristau> ok
[11:27] <jcristau> feel free to change that in git
[11:27] <jcristau> and add Depends: udev [linux-any] i guess
[11:27] <tjaalton> sure
[11:29] <tjaalton> our 64-xorg-xkb.rules didn't work at all, fixed :)
[11:30] <tjaalton> gconf settings overrode the defaults though, so I didn't notice it
[11:30] <jcristau> heh
[11:35] <jcristau> tjaalton: are you also adding xserver-xorg-input-{synaptics,evdev}-dev?
[11:36] <tjaalton> jcristau: for the headers?
[11:36] <jcristau> .h and .pc, yes
[11:37] <tjaalton> maybe I could
[11:38] <tjaalton> is there a postinst.in template I should use? looking at the xserver-xorg-core one..
[11:40] <jcristau> yeah stealing that seems fine
[11:49] <tjaalton> jcristau: ok, pushed evdev postinst with the trigger
[11:52] <jcristau> looks good
[11:52] <jcristau> i guess the server and synaptics need the same thing
[11:52] <tjaalton> yeå
[11:52] <tjaalton> uh
[11:52] <tjaalton> *yep
[11:53] <jcristau> :)
[11:53] <tjaalton> I need lunch.. the fingers are getting stiff :)
[12:00] <tjaalton> jcristau: and there goes evdev-dev
[12:00] <tjaalton> now lunch ->
[12:04] <jcristau> tjaalton: thanks!  i pushed a small fixup on top.  enjoy lunch :)
[12:23] <tjaalton> jcristau: oh yeah, copied the deps from some lib :)
[12:24] <jcristau> i figured ;)
[14:03] <mac_v> why is KMS turned off by default in .32-7 ?
[14:05] <tjaalton> because userspace is not ready
[14:21] <tjaalton> jcristau: wondering about the version comparison in xorg/evdev/synaptics. shouldn't the trigger run every time postinst is run?
[14:22] <jcristau> only when the rules changed i think
[14:22] <tjaalton> oh, good point
[17:44] <Sarvatt> oh good, the chrome beta build for linux is from right before all the rendering errors started happening in chromium. i thought they were intel related until i saw they happened on nvidia as well
[17:45] <Sarvatt> well it started around 4.0.260 in chromium and the chrome beta is based off of 11 days before that
[21:16] <notlistening> Hi, I am using the bleeding edge x release from PPA, i have two atom 330 machines both with ion graphics, one with the updated x-server and the other without and i am playing back a HD movie and getting about 90% CPU usage on the one with the PPA and 150% on the one without. Is this the updated version making a difference or do I have to look elsewhere? This is on Ubuntu 9.10 with the latest Beta Nvidia Drivers 
[21:18] <notlistening> oh and using lucid 
[21:19] <tormod> notlistening, you mean xorg-edgers PPA?
[21:20] <notlistening> yeah sorry i was not being exact with names there
[21:21] <tormod> lucid main has xserver 1.7, lucid xorg-edgers has xserver 1.8-pre-alpha, but I didn't know the nvidia blobs work with either
[21:21] <notlistening> Just trying to track down the performace differences
[21:22] <notlistening> oh I was just trying my luck :D
[21:22] <tormod> check the logs to see if there is any big difference there
[21:22] <notlistening> they wirjs fine so far as i can tell but i am not testing them beyong a media centres usage
[21:23] <notlistening> oh ok, I do remember enabling DRI2 as well i think
[21:24] <notlistening> should i report my findings?
[21:26] <tormod> only if you can single out some clear change, so far you findings are "anecdotal" to me :)
[21:27] <tormod> for instance, change between 1.7 and 1.8 on the same machine
[21:28] <notlistening> Ok i will try that now as i have the setups, are the test to run like glxgears and look for performance differences etc?
[21:28] <tormod> glxgears is famous for not being a good benchmark
[21:29] <tormod> playing the same movie and measure CPU is a bit better
[21:29] <tormod> is this a very new nvidia blob?
[21:52] <notlistening> tormod, what is a blob?
[21:54] <RAOF> Binary blob - proprietary, closed-source driver.
[21:54] <Sarvatt> the binary driver thats not open source, just a big binary blob packaged up :D
[21:54] <notlistening> thanks sorry quite new here 
[21:55] <Sarvatt> you want to find vdpau enabled versions of the movie playback software and switch the renderer to vdpau on an ion platform though
[21:55] <Sarvatt> the stuff in ubuntu doesn't have it enabled last i checked?
[21:56] <notlistening> the driver was released on 24/11
[21:56] <Sarvatt> that'll let you use the hardware acceleration on the gpu to play things back, but theres restrictions on the formats that it can accelerate so not everything can be accelerated that way even
[21:57] <notlistening> I have added ppa for vdpau playback also
[21:57] <notlistening> and have been trying the new flash 10.1 beta too :D
[21:57] <Sarvatt> last i checked there were some packaging errors for the 190 and 195 nvidia blobs in that ppa that didnt uninstall cleanly though :(
[21:59] <Sarvatt> so are you using mplayer to play back movies, and did you pick vdpau as the renderer in preferences?
[21:59] <notlistening> the file i am trying is an avi file which is using Microsoft mp4 4.2 
[22:00] <notlistening> I am using the built in movie player from XBMC running throught Boxee
[22:01] <notlistening> and have picked Vdpau, i have almost the same machine but have enabled the ppa on one and not the other and was trying to track down the performace enhancement
[22:02] <notlistening> I think i will tkae one step at a time and re do the install and see if i can pin point whats working and whats not
[22:03] <notlistening> and i am using the blob directly from nvidia installed manually in run lvl 1
[22:04] <notlistening> Will report back in a bit
[22:06] <Sarvatt> mp4 and avi are both container types not video formats though and using h264 in avi has some *horrible* limitations.. i wouldnt be surprised if it couldnt even accelerate the sample you're using
[22:13] <notlistening> Well something is giving me 50% less cpu usage so i will try to find out what
[22:21] <notlistening> on the edgers website it says the the current version in lucid is not usable in karmic?
[22:27] <tormod> notlistening, yes you should not install lucid packages in karmic, that means
[22:32] <Sarvatt> think we should enable glsl for r600 in mesa tormod?
[22:33] <tormod> Sarvatt, I did it once on request, didn't seem to break anything :)
[22:33] <tormod> it's not in git master for some reason though
[22:34] <Sarvatt> odd r600_context.h - Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
[22:34] <tormod> all the initial ati 3d work was done for a weather forecast system...
[22:35] <tormod> but that was on r200 or r100, not r600 :)
[22:36] <Sarvatt> yeah it says radeon 8500 after it
[22:36] <Sarvatt> ohhh was that what the 0ubuntu0glsl was a few weeks ago?
[22:37] <tormod> Sarvatt, the glsl is a bit depending on what upstream wants in bug reports, whether they keep saying "please try with glsl" or "oh noes you have enabled glsl"
[22:37] <Sarvatt> i thought someone else on the team actually uploaded something
[22:37] <tormod> it was the "glsl special edition" for phoronix junkies
[22:38] <Sarvatt> surprised they didnt just make that a driconf option
[22:38] <tormod> yes that, or a ENV thing would have been so much nicer
[22:39] <Sarvatt> i've got fake a glsl support option in there on i915
[22:40] <tormod> Sarvatt, since it's a radeon only thing, I can push glsl builds in the radeon ppa
[22:47] <tormod> Sarvatt, once lucid has libdrm 2.4.16, I'll add glsl for radeon to my nightly script
[23:01] <maxb> So... now X is de-HAL-ized, how does one pass config options to input drivers now/
[23:01] <maxb> * ?
[23:01] <Sarvatt> which input driver?
[23:01] <maxb> synaptics
[23:02] <Sarvatt> synclient? xinput?
[23:02] <maxb> I can't survive without my circular scrolling, now I'm used to it
[23:02] <maxb> Aren't those both for run-time tweaks? How do I persist the settings?
[23:03] <Sarvatt> udev rule?
[23:03] <Sarvatt> gpointing-device-settings doesnt have an option for circular scrolling?
[23:03] <Sarvatt> i thought it did for some reason
[23:05] <Sarvatt> yeah gpointing-device-settings lets ya do it
[23:06] <Sarvatt> in system - preferences - pointing device - scrolling
[23:06] <maxb> Hmm... I don't have it installed. I'll try it once my current apt run finishes
[23:06] <maxb> thanks
[23:08] <Sarvatt> can probably just add ENV{x11_options.CircularScrolling}="1" to the synaptics udev rule though
[23:09] <Sarvatt> i think it's going to take input options like that in /etc/X11/xorg.conf.d/ here soon?
[23:09] <Sarvatt> well probably not on xserver 1.7
[23:12] <jcristau> the x11_options in udev might not stay long..
[23:12] <jcristau> i'm kinda tempted to remove that
[23:15] <maxb> Any idea what the replacement will be?
[23:15] <jcristau> xorg.conf
[23:36] <Sarvatt> about Bug #438398.. that guy has a custom gcc symlink pointing to ccache, I have the same problem with my ccache symlinks not working right with DKMS too
[23:43] <Sarvatt> it runs as user nobody and tries to set up a new ccache folder for that user and fails completely yet dkms thinks it worked