RAOF | Technically that's argb windows rather than csd. I don't expect darkroom is implementing csd at this point. | 00:01 |
---|---|---|
Sarvatt | ahhh there we go, needed to enable compiz blur to make this manageable | 00:08 |
Sarvatt | heh and i get a nice 1 FPS while moving a window with that :) | 00:09 |
Sarvatt | and then compiz dies | 00:10 |
RAOF | At least your drivers support the needed extensions; intel doesn't, last I checked :) | 00:15 |
Sarvatt | i'm on intel :) | 00:18 |
Sarvatt | oh wow gnome-color-chooser | 00:18 |
RAOF | Oh, maybe you're using mipmap blur. | 00:21 |
Sarvatt | gaussian alpha blur works at least, slow as heck though | 00:21 |
Sarvatt | didnt try mipmap | 00:22 |
RAOF | Maybe our shiny new mesa enables it. | 00:23 |
Sarvatt | i dont have the fake opengl 2.0 support enabled in driconf either | 00:23 |
Sarvatt | looks like it was fixed here - http://cgit.freedesktop.org/mesa/mesa/commit/?id=0dc700850acb81c7088ab740959441521f8d38d9 | 00:26 |
Sarvatt | fixed in karmic i guess - http://macslow.net/?p=386 | 00:32 |
RAOF | notify-osd looks significantly better with blur enabled. I should have turned it on sooner! | 00:41 |
Sarvatt | ack, chromium controls moved over to the left now too, evil! :) | 00:44 |
Sarvatt | is it horribly slow on 965 too? | 00:45 |
Sarvatt | moving rgba windows around is a slideshow on 945 | 00:45 |
RAOF | I really like that chromium's over the other side; that makes it consistent. Enabling gaussian blur is a significant performance hit, but not abysmal. | 01:10 |
=== Amaranth_ is now known as Amaranth | ||
Sarvatt | RAOF: so you're saying lucid + a 2.6.34 kernel is working for nouveau for you too? | 02:38 |
Sarvatt | thats what cnd was saying and I was trying to make sure they aren't actually using nv without realizing it | 02:38 |
RAOF | Sarvatt: Yes. | 02:39 |
Sarvatt | what the heck | 02:39 |
RAOF | The libdrm ABI break doesn't really affect the DDX in everyday use. | 02:39 |
Sarvatt | sure it did, i tested the heck out of it when it first went in, libdrm with 0.0.15 + 0.0.16 kernel did not work | 02:40 |
RAOF | Yeah, but the new libdrm has the 0.0.16 interface. | 02:40 |
Sarvatt | in *lucid*? | 02:40 |
Sarvatt | stock lucid? | 02:41 |
Sarvatt | 2.4.18 with the abi reverts | 02:41 |
RAOF | Oh. Um. | 02:41 |
Sarvatt | they are saying its working in lucid | 02:41 |
RAOF | ECONTEXT | 02:41 |
Sarvatt | *maverick* has the 2.4.20 libdrm with the new abi, but in #ubuntu-kernel and on the kernel team list tgardner and cnd are saying nouveau is working with the maverick backported kernel in lucid | 02:42 |
RAOF | I am equally perplexed; it totally should be dying with a drm mismatch. | 02:42 |
Sarvatt | they are just using -nv without realizing it as far as I can tell, can't get an xorg.0.log out of anyone :) | 02:42 |
RAOF | I thought this was one layer up the stack; with the ABI of the libdrm_nouveau that the DDX is linked to changing underneath it. | 02:43 |
RAOF | I thought nouveau died in a more spectacular way when that happened. | 02:43 |
Sarvatt | apparently it unregisters cleanly now and nv loads if they are saying it works :D | 02:50 |
Sarvatt | i can't for the life of me figure out what else could be happening | 02:51 |
Sarvatt | thats awesome if that works though, backported kernels aren't supposed to be supported on the desktop anyway they were saying so at least they have nv | 02:54 |
RAOF | Given how much I was looking forward to making a godawful libdrm symlink farm hack, that would be awesome, yes. | 02:55 |
Sarvatt | was really hoping to speak to them about building the agp modules into the kernel at UDS :( | 02:57 |
RAOF | Damn, yeah. | 02:58 |
RAOF | That sucked all round. | 02:58 |
Sarvatt | mesa needs libglew-dev to build now and its in universe :( | 03:10 |
RAOF | You mean mesa-demos, right? | 03:12 |
RAOF | I wonder how many people care about 3D graphics under VMWare. | 03:13 |
Sarvatt | nope mesa, think they removed the bundled version when they split demos out | 03:13 |
Sarvatt | no idea what part of the build actually needs it though.. | 03:13 |
RAOF | Doesn't mesa *build* libglew, though? | 03:14 |
Sarvatt | bundled version is still there, hmm | 03:15 |
Sarvatt | the dri target fails to build | 03:15 |
Sarvatt | checking for GLW... configure: error: Package requirements (x11 xt) were not met: | 03:15 |
Sarvatt | No package 'xt' found | 03:15 |
Sarvatt | ok where the heck did i get glew from | 03:16 |
* Sarvatt needs to get off the pc :) | 03:16 | |
RAOF | Sarvatt: No, that's it checking a bunch of pkg-config files to get what it needs to *build* GLW. | 03:16 |
Sarvatt | yeah needs xt, i had a commit open where it was checking for glew.pc and got mixed up, sheesh | 03:17 |
RAOF | apt-file says that xt.pc does not exist in Maverick. | 03:17 |
Sarvatt | libxt-dev | 03:17 |
Sarvatt | i've got it installed, hmm | 03:18 |
RAOF | So, my apt-file cache is broken. | 03:18 |
Sarvatt | Filename: pool/main/libx/libxt/libxt-dev_1.0.7-1_i386.deb | 03:23 |
Sarvatt | MD5sum: 26737bd9e76f4893d2fe7643c1a95da3 | 03:23 |
Sarvatt | Archive: maverick | 03:23 |
Sarvatt | maybe because it was just copied over from lucid? | 03:23 |
RAOF | Your guess is as good as mine. Maybe it's just baroke. | 03:24 |
Sarvatt | this libgles-qemu is interesting | 03:29 |
RAOF | It's a bit strange. | 03:29 |
Sarvatt | even debianized | 03:31 |
Sarvatt | the headers are all we need really, they still need the blob libs for each platform packaged to actually do anything with them | 03:35 |
RAOF | Which isn't terribly useful if we want to actually, say, test things on our desktop :) | 03:38 |
Sarvatt | they're different on each platform and testing it on the desktop isn't really useful though, powervr has a package where you can test what it'll actually be like on the devices in the sdk | 03:39 |
RAOF | Sure the extensions will be different, but you'll get the same core APIs, right? | 03:39 |
RAOF | And you're not going to be performance testing on the desktop. | 03:40 |
Sarvatt | the extensions look crazy different on the different devices i looked at | 03:40 |
RAOF | But if you're not using the extensions it doesn't matter; and since the extensions are crazy different you can't use them if you want to be portable. | 03:41 |
bryceh | btw, I noticed that versions-current.html has been broken. Surprised no one noticed, maybe it's not being used anymore? Anyway, in case it's still useful it's back online - http://www2.bryceharrington.org:8080/X/Reports/ubuntu-x-swat/versions-current.html | 03:41 |
Sarvatt | i dunno just having different extensions available can make things take different code paths | 03:41 |
bryceh | problem was that apt was exhausting its cache, I guess due to too many sources. Odd. | 03:41 |
Sarvatt | bryceh: it was working a few days ago at least | 03:41 |
bryceh | oh ok. I don't have a way to see how long it's been broken. | 03:42 |
Sarvatt | i go to it a lot, thanks so much for fixing it :) | 03:43 |
Sarvatt | we're pretty much caught up, just waiting for xserver to be uploaded and the dozens of packages to build against it and such, dont think anyones looking forward to that :) | 03:45 |
Sarvatt | most of those green ones in maverick are in dep wait | 03:45 |
RAOF | We need to merge -intel from experimental, but I can do that once mesa is done. | 03:46 |
Sarvatt | RAOF: http://sarvatt.com/downloads/patches/copy-fb.patch | 03:47 |
Sarvatt | (updated copy-fb patch for 2.11) | 03:47 |
Sarvatt | its in x-updates | 03:47 |
RAOF | Ta. | 03:48 |
Sarvatt | been trying to get fedora to upstream it in #intel-gfx since its broken yet again in intel master | 03:49 |
Sarvatt | the bgnr stuff needs to be upstreamed in xserver though first | 03:49 |
Sarvatt | oh duh fglrx needs the 2.6.34 support patch I'm sure for maverick | 03:52 |
RAOF | And then it'll break with the new X server? :) | 03:53 |
Sarvatt | heheee | 03:53 |
bryceh | I've added a hook to look up the nvidia-graphics-drivers version too | 03:58 |
Sarvatt | does it check git repos or ftp or http directories? | 04:08 |
Sarvatt | you could use say ftp://download.nvidia.com/XFree86/Linux-x86/latest.txt for the upstream versions | 04:09 |
Sarvatt | oh you already do :) | 04:09 |
Sarvatt | was going to say if it checked git the nvidia-settings git repo has the newest ones | 04:10 |
Sarvatt | http://people.freedesktop.org/~aplattner/nvidia-versions.txt | 04:11 |
RAOF | Alright. Time to head out & collect my new glasses. Also, luncheon. | 04:12 |
* RAOF wonders what decided to eat all his ram _just_ as the build was about to finish. | 07:13 | |
hyperair | RAOF: a browser is a likely candidate. | 07:24 |
hyperair | if i leave deluge's web UI on chromium for more than two days, it can drive my machine into OOM-massacre mode | 07:24 |
=== schmidtm_ is now known as schmidtm | ||
* hyperair wails at Sarvatt. | 19:59 | |
hyperair | it broke again! | 19:59 |
hyperair | x-x-v-i, i mean | 19:59 |
* hyperair downgrades before cautiously re-enabling gdm | 19:59 | |
Sarvatt | when did you upgrade | 19:59 |
Sarvatt | i disabled page flipping a few days ago | 19:59 |
Sarvatt | err disabled the disable page flipping patch! lol | 20:00 |
Sarvatt | are you using an old kernel or something? | 20:00 |
hyperair | just did | 20:00 |
hyperair | 2:2.11.0+git20100526.03bbb4c8-0ubuntu0sarvatt~lucid | 20:00 |
Sarvatt | which was the good one? | 20:00 |
Sarvatt | that you had before | 20:00 |
hyperair | no, my kernel is 2.6.34 | 20:00 |
Sarvatt | or do you upgrade every day? | 20:00 |
Sarvatt | i just want to be sure its page flipping thats screwed up for you because i'm disabling it for everyone just based on you :) | 20:02 |
Sarvatt | and not that the drivers broken in other ways after the update which it probably is | 20:02 |
hyperair | i upgrade every day. | 20:03 |
hyperair | 2:2.11.0+git20100525.b645ec83-0ubuntu0sarvatt~lucid | 20:03 |
hyperair | this is the one i'm currently using | 20:03 |
Sarvatt | so its not page flipping | 20:04 |
Sarvatt | drivers just busted, ok | 20:04 |
Sarvatt | updating it :) | 20:04 |
Sarvatt | mesa is a mess right now, haven't been able to update it in like a week | 20:05 |
hyperair | well come to think of it, that driver went bust on me and blackscreened, which is why i lost my uptime and had to reboot in the first place. | 20:05 |
hyperair | which is how i ended up using the new driver. | 20:06 |
hyperair | which then proceeded to go bust on me three times in a row with less than half an hour's worth of uptime each round | 20:06 |
=== Amaranth_ is now known as Amaranth | ||
Sarvatt | ok ok new one uploading with disable-pageflip.patch :) | 20:08 |
Sarvatt | i keep trying to drop it because they are actually trying to fix it | 20:11 |
hyperair | i dunno really, it could have been a random error that's not related | 20:12 |
hyperair | after all, my uptime was ~3 days | 20:12 |
hyperair | compared to the previous half an hour uptimes | 20:12 |
Sarvatt | i needed to update to todays git checkout anyway, no biggie :) | 20:13 |
hyperair | well if you don't get other complaints you could always enable it and i can just aptitude hold =p | 20:13 |
Sarvatt | i locked up 4 times in the past 3 days that i'm sure was because of it :) | 20:13 |
hyperair | ahah | 20:14 |
hyperair | lol | 20:14 |
Sarvatt | the freezes when i dont have that disable page flipping patch are always the same, pc hard locks with the image still on the screen, not pingable and sysrq doesnt work | 20:14 |
hyperair | hmm | 20:15 |
hyperair | i see. | 20:15 |
hyperair | oh well. | 20:15 |
hyperair | ....shit. now i've got a corrupted git repository. | 20:16 |
Sarvatt | i lost my /etc git repo last time :( | 20:16 |
hyperair | ouch =\ | 20:17 |
hyperair | well, i blame ext4 | 20:17 |
hyperair | nothing on my btrfs volume went missing. | 20:17 |
Sarvatt | RAOF: wow you've been busy with mesa huh? I'm scared to ask how much longer this takes to build now :) | 21:00 |
Sarvatt | hmm, radeong_dri.so stopped being built in the past few days for some reason it looks like, or it wanted to make my life hard and is called r300_dri.so now | 21:07 |
Sarvatt | haha | 21:11 |
Sarvatt | always great when 7 days of headaches are caused by a tiny stupid change you made :) | 21:12 |
Sarvatt | commented out a line in a confflags-dri section in mesa's debian/rules, everything after it wasn't getting passed :D only reason i needed all these other build deps was because of that | 21:14 |
Sarvatt | llvmpipe is quite annoying to build during development cycles apparently, oprofile is broken every few days when binutils gets updated making llvm-dev uninstallable | 21:20 |
Sarvatt | RAOF: are you sure you want to use the --with-dri-searchpath method? for libgl1-mesa-dri-gallium? I need to build what you've pushed to see how its set up but there are things hardcoded to use /usr/lib/dri only that dont honor it we've found :( | 21:26 |
tormod | heya sarvatt! | 21:33 |
Sarvatt | heyo tormod! how the heck have ya been man? | 21:33 |
tormod | remember keyboard layout went a bit wild sometime when running xserver 1.7? do you know the reason? | 21:34 |
tormod | I mean, in xorg-edgers on karmic I think | 21:34 |
tormod | there's an upgrade regression that reminds me of that, bug 583037 | 21:36 |
ubot4 | Launchpad bug 583037 in xorg-server (Ubuntu) "arrow-left maps to ISO_Level3_Shift (affects: 1) (heat: 8)" [Undecided,Incomplete] https://launchpad.net/bugs/583037 | 21:36 |
tormod | bryce, btw your bug robot tagged that bug "hardy" without reason AFAICS | 21:37 |
Sarvatt | that looks like a bug that'll go away once he drops the xorg.conf to me :D i have no idea though, i'm really sketchy in the xkb area since I just use english and occasionally japanese and things just work so I haven't had to fix anything to learn more about it :) | 21:44 |
bryceh | tormod, well I'm sure it has a reason, but probably not a good one. I'll investigate. | 21:46 |
bryceh | btw, if anyone finds other issues with the automated scripts, file against http://bugs.launchpad.net/arsenal/ and that'll help me keep track of them | 21:47 |
bryceh | Current Operating System: Linux nx6125 2.6.34 | 21:48 |
bryceh | what is nx6125? | 21:48 |
Sarvatt | wow llvmpipe *stinks* on a CPU with only SSE2 | 21:49 |
Sarvatt | compaq laptop model number is my first guess :) | 21:49 |
Sarvatt | yea google agrees | 21:49 |
bryceh | oh that's the machine name, duh | 21:49 |
bryceh | tormod, bug 586547 | 21:50 |
ubot4 | Launchpad bug 586547 in arsenal "Tagged bug as 'hardy' inappropriately (affects: 1)" [Undecided,New] https://launchpad.net/bugs/586547 | 21:50 |
tormod | bryceh, I have also noticed bug status being changed and then back in one go, like in bug 468413 (after comment 31) | 21:54 |
ubot4 | Launchpad bug 468413 in xserver-xorg-video-ati (Ubuntu) (and 1 other project) "Radeon module loaded before agpgart (affects: 17) (heat: 108)" [Undecided,Incomplete] https://launchpad.net/bugs/468413 | 21:54 |
bryceh | tormod, actually in that particular case I did that manually | 21:55 |
bryceh | but I have the scripts doing it automatically sometimes too | 21:55 |
bryceh | it does this when the bug is in state Incomplete, since once there is a comment it turns into Incomplete (with response), but it needs to be in Incomplete (without response) | 21:56 |
bryceh | I don't know of any way to achieve getting it into that state except by moving it to some other state and then back to Incomplete | 21:56 |
tormod | oh I see | 21:56 |
bryceh | I have a bug about this open against Launchpad | 21:56 |
bryceh | but it got marked Low, which means it'll probably not get attention | 21:56 |
bryceh | it's easy enough to work around in arsenal so dunno that I'll try to fix it in LP myself | 21:57 |
Sarvatt | yay even more headache, nouveau_class.h was removed from libdrm, mesa 7.8.1 needs it still for nouveau, only master works there :D | 22:07 |
Sarvatt | so debian-experimental doesn't build | 22:07 |
Sarvatt | http://sarvatt.com/downloads/mesa_7.8.1-2_amd64.build | 22:09 |
Sarvatt | RAOF must have built it on lucid with libdrm 2.4.18 | 22:10 |
bjsnider | what's the git command that downloads the current snapshot but doesn't grab the entire history? | 22:20 |
hyperair | git clone --depth=1 | 22:24 |
hyperair | you can change --depth accordingly to control how many levels of history to download | 22:24 |
bjsnider | i thought there was an actual command separate from clone | 22:25 |
hyperair | no, i don't think there was | 22:25 |
hyperair | there's also fetch | 22:26 |
hyperair | but that's after cloning | 22:26 |
bjsnider | like the "export" command in subversion | 22:26 |
hyperair | git archive? | 22:26 |
bjsnider | what does that one do? | 22:26 |
hyperair | makes a tarball or zip | 22:26 |
hyperair | out of the tree | 22:26 |
hyperair | you can git archive --prefix=asdf/ HEAD | tar -x | 22:27 |
hyperair | which is equivalent to svn export | 22:27 |
hyperair | or you could just cd into another directory and GIT_DIR=/path/to/whatever/.git git checkout HEAD -- . | 22:27 |
Sarvatt | argh, my bad, nouveau_class.h was removed post 2.4.20. commits after the nouveau_class.h removal were cherry picked into debian, I thought they did a git checkout | 22:58 |
RAOF | Sarvatt: No, debian-experimental doesn't build because I made a typo fixing up a trivial bug. | 23:44 |
Sarvatt | well it didn't build for me because I built it against git libdrm, didn't get to whatever the typo was yet :) | 23:45 |
RAOF | Ah, right. | 23:46 |
RAOF | Sarvatt: What's hardcoded to /usr/lib/dri? | 23:49 |
Sarvatt | when X starts it loads whatever's in /usr/lib/dri and would just load swrast if nouveau_dri.so was missing, not sure if you put that into the gallium package or left it in dri | 23:51 |
RAOF | Ah. | 23:51 |
Sarvatt | like if you install libgl1-mesa-dri-gallium a ton of GLX visuals and GLXFBConfigs go missing compared to when the gallium dri is installed to /usr/lib/dri/, and some people on ati say it causes compiz lockups all over the place | 23:53 |
Sarvatt | did you rename all of the gallium drivers btw? | 23:53 |
Sarvatt | still haven't finished building it, got distracted testing out libkms stuff | 23:53 |
RAOF | No. | 23:53 |
Sarvatt | they dont get used without renaming :( | 23:54 |
Sarvatt | radeong_dri.so = r300_dri.so, swrastg_dri.so swrast_dri.so | 23:55 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!