/srv/irclogs.ubuntu.com/2010/05/15/#ubuntu-x.txt

=== PyObjectPtr is now known as static_name
=== static_name is now known as apachelogger
hyperairSarvatt: did you notice xserver-xorg-video-intel's last two snapshots being terrible?02:54
Sarvatthyperair: you mean the last 6?04:22
Sarvattthe fact i uploaded 6 new snapshots in a day says yes I did :D04:22
Sarvattstill have to force disable pageflipping too :(04:23
Sarvattoh goodie 11 new commits too, i'm scared to update it again since its at least mostly working and it's been broken every day :D04:24
hyperairSarvatt: oh heh. xD04:26
Sarvatti'll update it at midnight (30 minutes away) and make sure its not screwed before uploading it this time04:26
hyperairSarvatt: ok cool.04:26
hyperairSarvatt: i've got an old deb which works =p04:26
hyperair-rw-r--r--    1 hyperair hyperair     554338 2010-05-15 05:02 xserver-xorg-video-intel_2%3a2.11.0+git20100511.8562b7bc-0ubuntu0sarvatt_amd64.deb04:26
Sarvattthe one on there now isn't working for you?04:26
hyperairer artifacts all over.04:26
Sarvattah thats the 5-08 one04:26
hyperairyesterday's one just had issues with compiz's shadows.04:26
Sarvatti just said screw it and rebranded the 05-08 as 05-11 because it was all kinds of messed up on 5-1004:27
hyperairso the shadows went missing, and then some of the selected items' borders got bolder.04:27
Sarvattwhat one did you try?04:27
Sarvattbecause the 5-14 is working ok here04:27
Sarvattall of the 5-13 ones were messed up, i was adding commits to that instead of doing a new checkout as they went in04:27
hyperair 2:2.11.0+git20100513.8de09a07-0ubuntu0sarvatt~lucid and 2:2.11.0+git20100512.6c27f6e4-0ubuntu0sarvatt04:27
hyperairaha 5-14 eh.04:28
Sarvatt2:2.11.0+git20100514.4be8d7eb-0ubuntu0sarvatt - not having any problems with that outside of the drawer applet going black sometimes04:28
hyperairdrawer applet..  haven't used that.04:28
Sarvattbut theres 11 new commits since that one now too04:28
hyperairlol04:29
Sarvatti think a new quarterly release is due soon, they usually shove in a ton of things that break this way every time right before it :D04:29
Sarvattat least its fixed at the end of the day every day the past week, instead of being broken for the whole month of november last time :D04:30
hyperairheh i remember soemthing of that sort04:31
hyperairi switched from aptitude forbid-version to aptitude hold04:31
Sarvattlemme know if 5-14 works for you, better off holding off until tomorrow to update when the fixes for these 11 commits hit if it does04:31
hyperairokay, i'll go install it now.04:32
hyperairokay, shadows are missing, the panel icon for the main menu in gnome-panel turns black occasionally04:35
Sarvatti try not to let it stay broken though even if it means i have to upload an old working snapshot04:35
hyperairbut other than that, it's fine.04:35
Sarvattok will update it and see how it goes then04:35
* Sarvatt jumps ahead a time zone and updates things04:35
hyperairlol04:35
Sarvattwhat shadows are missing? I don't see that at all04:40
Sarvattuploaded 05-1504:41
hyperairSarvatt: gtk-window-decorator usually puts drop shadow.s04:42
hyperairshadows*04:42
Sarvattah i use mutter04:42
hyperairheh04:42
Sarvattwoke up to it looking like this this morning - http://sarvatt.com/downloads/screenshot.png04:43
Sarvattcouldn't screenshot the nasty constant flicker though :)04:43
Sarvattif this one is screwed up i'm just going to put the 05-08 (thats the same as your 05-11 one) back on there for a bit04:45
hyperairi see.04:49
* hyperair notes that Sarvatt's machine is called asuka.04:49
hyperairNGE?04:49
Sarvattyep :)04:50
hyperairheheh04:50
Sarvattahh go figure huge build queue for chromium/mozilla :)04:53
Sarvattseems to be working fine04:57
Sarvattshadows are messed up, I see what you mean04:57
Sarvattswitched to compiz and i see a light border a few pixels away from the edge of windows and where the shadow should be is completely transparent04:58
Sarvatthttp://sarvatt.com/downloads/screenshot2.png04:59
Sarvattcan see the grey border there04:59
Sarvatti probably would have never noticed that since i use everything maximized anyway :)05:05
Sarvattstinks that its impossible to disable vsync now on intel without an env variable05:07
Sarvatttried setting vblank_mode to 0 in /etc/drirc and ~/.drirc but it doesn't stick, vblank_mode=0 glapp works though05:08
hyperairenv variable?05:10
hyperairwhat's wrong with vsync? i have it turned on anyway05:11
hyperairand you can set your .xprofile/.gnomerc to export the var05:11
Sarvattwell for one its totally screwed up on 94505:13
Sarvattif you dont move the mouse constantly to generate interrupts it lags to all hell05:13
Sarvatt20 fps glxgears ftw! :)05:14
hyperairouch05:15
Sarvattouch, libdrm in xorg-edgers picked up an epoch somewhere in the past month? how the heck did that happen, i dont see forcing one in my history and dont think anyone else uploaded it05:30
Sarvattargh I did do it when I updated to xserver 1.8 since i reused a command updating a driver - ./auto-xorg-git *-e 1* -d origin/ubuntu -g -p libdrm -a 0ubuntu0sarvatt -t +05:35
Sarvattguess it'd be more evil to delete it and reupload without an epoch since it'd strand people with the epoch version unless they manually downgraded and anyone switching away *needs* to use ppa-purge05:38
Bernardogood morning05:39
Bernardohi lucazade08:17
lucazademorning08:17
lucazadei've read that 2.6.34 should fix /proc/acpi/video, in that way ignoreacpi could be removed08:19
Bernardoand it seems there is a fix for the psb module in the intel mailing list, we should look at that08:20
Bernardothis seems similar, maybe we can do the same kind of patch - http://www.gossamer-threads.com/lists/linux/kernel/1224507 ?08:24
BernardoI don't have 2.6.34 here, maybe I'll be able to install it and test later today08:24
lucazadelooking at08:27
=== |Bernardo| is now known as Bernardo|away
hyperairSarvatt: i seem to have gotten an upgrade for xserver-xorg-video-intel which fixes the black-icon syndrome but not brought back shadows.11:21
jcristauRAOF: fyi i took your vesa patch and added http://git.debian.org/?p=pkg-xorg/driver/xserver-xorg-video-vesa.git;a=commitdiff;h=8c3fb055a63c2eca23b3ba6b585484f010e2ee06 to get a package dependency on libdrm212:18
Sarvatthyperair: shadows fixed in git now, just uploaded it13:20
=== Bernardo|away is now known as Bernardo
Sarvattok looks like FBC is just plain broken on a large number of 8086:2a42 devices as well, I was thinking it might be lenovo bios specific there - https://bugs.edge.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/53864816:07
ubottuLaunchpad bug 538648 in xserver-xorg-video-intel "[Intel GM45] Irregular sync flashes (Lenovo T500, Dell Insp. 1545)" [Medium,Triaged]16:07
Sarvatti dont think i can write up a patch to disable just FBC on that specific chipset since the structs encompass the whole generation, should we just quirk the powersave module parameter on that pci id for now?16:10
Sarvattricotz: ahh that dh_gconf problem is hitting your maverick mutter ppa too huh?17:30
Sarvattlooking at the debhelper changelog it should have been fixed but the fix isn't working17:31
SarvattThe following packages have unmet dependencies:17:31
Sarvatt  mutter-common: Depends: gconf2 (>= 2.28.1-2) but 2.28.1-0ubuntu1 is to be installed17:31
ricotzSarvatt, hadnt looked really into it yet, but it seems to be another problem 17:34
ricotzSarvatt, did you had a look at failure protocol of gnome-shell?17:37
Sarvattnope i dont use gnome-shell at all, looks like crap on a netbook17:38
ricotzok, just thought so, because you mentioned this problem17:39
Sarvattit's dh_gconf in debhelper for sure bringing in that dep,                         addsubstvar($package, "misc:Depends", "gconf2 (>= 2.28.1-2)") unless $gconf_dep;17:39
Sarvattits run by /usr/share/cdbs/1/class/gnome.mk17:40
jcristaui guess maverick will get new gconf soon enough17:42
ricotzSarvatt, thanks for bringing this up17:44
Sarvattricotz: its affecting a ton of other packages in maverick right now, just brought it up in #ubuntu-desktop17:45
Sarvattbut UDS and all.. :D17:45
Sarvattreally wish xchat would release a new version already, it's insane how much faster svn is17:46
ricotzright ;-)17:46
Sarvatttakes close to a minute to start and load the scrollback for 32 channels on this atom cpu, 4 seconds with one of the commits backported :D17:47
Sarvatthaving the same gconf problem with that though thats why i looked into it17:47
ricotzsounds like a hell of an improvement17:49
ricotzisnt there a ppa for xchat yet?17:49
Sarvattso much stuff fails to build with gcc 4.5 though17:49
Sarvattwell I built it here - https://edge.launchpad.net/~sarvatt/+archive/ppa/+sourcepub/1133313/+listing-archive-extra17:50
Sarvatthttp://xchat.svn.sf.net/viewvc/xchat?view=rev&revision=1386 thats the commit that made such a huge difference17:50
ricotznice!17:51
ricotzperhaps you can think of backporting the patches for mesa from nvfx-next-6b ;-)17:52
ricotzg2g, sorry17:53
Sarvattthat probably needs a lot of work to bring it up to date for master by now?17:53
Sarvattjust run auto-xorg-git like normal, control+z at the patching pause, delete debian/ and replace with git://sarvatt.com/mesa.git and git pull that branch into it and then let it finish17:55
Sarvattor ./auto-xorg-git -H hooks -g -d origin/ubuntu -t "~" -p mesa -a 0ubuntu0ricotz -b nvfx-next-6b17:56
Sarvatt(and do the normal delete debian/ and replace with the other at the patching pause)17:57
Sarvattjust saying that incase he reads the irc logs :)17:57
Sarvattricotz: if you read this later I uploaded the nvfx-next-6b branch to https://edge.launchpad.net/~sarvatt/+archive/nouveau/+packages18:05
LLStarksyo sarvatt, alpha channel broken again in edgers?18:34
Sarvattits been in various stages of brokenness all week, latest one is working fine here though now18:39
LLStarksD:18:40
Sarvattif 2:2.11.0+git20100515.9c3da713-0ubuntu0sarvatt3 is screwed up for you and you have the old debs around the 05-11 one should be fine18:43
jcristau2 new commits 5 minutes ago...18:43
Sarvattthats the 0508 checkout with a later date since the breakages started on 051018:43
Sarvattsheesh I looked when he asked too!18:44
SarvattLLStarks: new version about to be uploaded..18:44
LLStarksi see18:48
LLStarksthanks18:48
LLStarkswhenever something breaks in edgers, it's either alpha channel or gallium being a noob18:49
Sarvattyep, lucky the breakage doesn't last for months at a time anymore though :)18:50
LLStarksmaverick will be mesa except for nouveau, right?\18:51
Sarvattwhat do you mean?18:51
Sarvattmesa classic?18:51
LLStarksyup18:51
Sarvattyeah.. if nouveau is there it'll be a seperate package unless they had other plans at UDS18:52
LLStarkslucid was classic+nouveau?18:52
Sarvattno nouveau in lucid18:52
LLStarksor no 3d18:52
Sarvattnouveau has a classic component too for older cards18:52
LLStarkswait, it didn't make the cut? i thought it did.18:52
LLStarks*it = nouveau18:52
Sarvattno way18:52
Sarvattit would have been crazy to ship nouveau gallium from 7.7.118:53
LLStarksanyway, i'm finding i915/i945 gallium to be terrible. broken desktop rendering in 2d. no compositing whatsoever.18:53
LLStarksno 3d.18:54
LLStarksthis is 0.4 i think18:54
Sarvatt915 is working fine here, you using some crazy compiz plugins maybe?18:55
Sarvattby working fine i mean i haven't seen any artifacts or had any crashes, the classic one is still way better18:56
LLStarksi wish i could enable occlusion and fragment shader on gallium18:58
LLStarksOpenGL renderer string: Mesa DRI Intel(R) 945GM GEM 20100330 DEVELOPMENT x86/MMX/SSE218:58
LLStarksOpenGL version string: 2.0 Mesa 7.9-devel18:58
LLStarksOpenGL shading language version string: 1.2018:58
LLStarkscan only get opengl 1.4 and no glsl on gallium.18:59
SarvattLLVM=1 LIBGL_ALWAYS_SOFTWARE=1 yourapp? :D18:59
LLStarkslemme try.18:59
Sarvattthats with libgl1-mesa-dri-gallium installed19:02
Sarvattwhoops, no llvm dep on that package19:02
LLStarksopengl 2.0 with classic and driconf tweaks. opengl 1.3 with gallium. i don't get it.19:03
LLStarksnice19:04
LLStarkscompiz works now19:04
Sarvattoh probably because you're using the classic dri lib for AIGLX,  it doesn't honor the --with-dri-searchpath thing I'm doing19:04
LLStarksmight be software renderer19:04
LLStarksaiglx and xgl19:04
LLStarkshaven't heard of those since the beryl days19:04
Sarvattneed to move /usr/lib/dri-gallium/i915_dri.so to /usr/lib/dri/19:05
LLStarksthese days, compiz just works.19:05
Sarvattmove the old one out of the way and link it or whatever, haven't worked that part out yet19:05
jcristauis there any point using i915g?19:05
Sarvattnope19:05
Sarvattpeople read phoronix and think its better :)19:06
jcristauthat's what i thought19:07
jcristaulooks like it hasn't changed in a month or so19:07
Sarvattit changed in the last month?!19:07
Sarvattmust have been a big api rework and they just fixed everything :)19:08
jcristauthere was a bunch of commits from jakob mid april19:09
LLStarksfatal x crashes. wonky metacity and gnome-panel.19:10
LLStarksi'll stick to classic for now.19:10
jcristau"for now"...19:10
SarvattLLStarks: just curious, why did you think it was even supposed to be usable?19:10
LLStarksi never expected it to be.19:11
jcristauthen what are you trying to do?19:11
LLStarksget a stable 2d desktop.19:11
Sarvattwith xorg-edgers..? :)19:11
LLStarksi don't see why gallium would affect the desktop when compositing is disabled.19:12
LLStarksedgers can be pretty stable. besides, i love bleeding edge.19:12
jcristau"i love bleeding edge" and "i want a stable 2d desktop".  interesting.19:13
LLStarksi use the word stable lightly19:14
LLStarksthings can be screwed up under the hood, but everything should look right.19:14
Sarvattguess i'll back those two patches back out, sick of uploading intel for tonight :) http://sarvatt.com/downloads/screenshot2.png19:31
Sarvattyeaaah i need to stop updating crap today, i installed the gallium package to mess with llvmpipe there and thats what was causing the rendering problems with mutter :D19:55
=== Bernardo is now known as Bernardo|away

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