[05:54] <Dandel> Does anyone know what it would take to add the opengl icd dev files to the xorg edgers branch on precise? ( it exists in all newer releases )
[05:54] <Dandel> * Opencl
[05:57] <Dandel> also, There is probably a good reason to integrate this patch ( http://cgit.freedesktop.org/xorg/xserver/commit/?id=7ecfab47eb221dbb996ea6c033348b8eceaeb893 ) into the older server set for lts since there is plans to remove the old entry proints in the newer driver updates
[05:57] <hyperair> looking at the build-deps, probably a backport of khronos-opencl-headers and ocl-icd
[05:58] <Dandel> hyperair, I did look into that... quantal and up all have ocl-icd-libopencl1
[05:59] <Dandel> opencl-headers ( exists in precise ) but there is no direct lib
[05:59] <hyperair> you need version 1.2
[05:59] <hyperair> which isn't in precise
[05:59] <hyperair> at least not according to rmadison.
[06:00] <Dandel> ya... and honestly it's a combination of both update opencl-headers to 1.2 and to add ocl-icd-opencl-dev 
[06:00] <Dandel> there is already at least four different packages that can benefit from this ( with slight updates )
[06:01] <Dandel> nvidia and amd binary drivers both can take full advantage of this change
[06:01] <Dandel> and as for testing packages, Piglit has full opencl support
[06:02] <Dandel> I know that opencl support in piglit is not enabled specifically because of compilation problems caused by the lack of an opencl icd lib ( like what libgl.so does for opengl )
[06:05] <Dandel> Also, The piglit package that got disabled was disabled not due to a new mesa requirement... it was a bug with detecting the posix function strndup.
[06:34] <Dandel> hyperair, opencl headers 1.2 do introduce the icd concept and supporting the icd for ubuntu lts is good since there is some patches to the mesa opencl implementation ( clover ) that enables the icd support.
[06:36]  * hyperair shrugs
[06:36] <hyperair> i'm not too familiar with opencl
[06:37] <Dandel> opencl is a varied language... lets you write programs to do specific tasks on the graphics card and regular old cpu.
[06:38] <hyperair> yeah i got that much
[06:38] <hyperair> but i'm not familiar with how the components involved, etc.
[06:38] <hyperair> (nor am i the one handling the opencl packaging, or the xorg-edgers ppa)
[06:39] <Dandel> except for some minor libc changes there is not many components involved in fixing this package.
[06:40] <hyperair> libc changes?
[06:40] <Dandel> I don't know what all changes are from precise (2.15-0ubuntu10.5) to quantal ( 2.15-0ubuntu20.2 ) on libc but i don't think it's a lot... since the major versioning did not change.
[06:49] <Dandel> hi ricotz 
[06:50] <Dandel> it looks like the daily piglit builds will resume although some verification needs to happen.
[06:52] <Dandel> ricotz, I think it may be time to update waffle ( to version 1.3.0 ) since there is a patch for piglit that will land sometime this next week that will knock out the builds.
[06:56] <ricotz> Dandel, hi, i see
[06:56] <ricotz> Dandel, so your patching should work for now?
[06:57] <Dandel> ya, but updating waffle is also a longer term fix.
[06:57] <ricotz> Dandel, thanks!, will look into waffle
[06:58] <Dandel> ricotz, the patch set: http://lists.freedesktop.org/archives/piglit/2013-October/007763.html
[07:01] <Dandel> ricotz, based on what I can see, there should be no changes when building waffle 1.3 ( mainly fixes )
[07:05] <Dandel> ricotz, the updated patching also should include fixes for lesser "known" tasks like generating an html viewable output of piglit results.
[07:14] <ricotz> Dandel, will push it in a bit
[07:24] <ricotz> Dandel, doesnt seem your refreshed patches apply
[07:26] <Dandel> ricotz, looks like it failed to handle the set 0o'
[07:27] <ricotz> did they work for you locally?
[07:27] <Dandel> they where written to the latest tree.
[07:27] <Dandel> failed on: Applying patch 40-piglit-install-directories.patch
[07:28] <ricotz> and they applied cleanly with fuzz
[07:28] <ricotz> *without*
[07:28] <Dandel> earlier, it was broken with Applying patch 30-change-template-dir.patch
[07:29] <Dandel> looks like a slight conflict since 40-piglit-install-directories.patch and 10-no-rpath-in-build.patch both modify cmakelists.txt
[07:33] <ricotz> ok, could you fix it? waffle 1.3.0 is up an built
[07:33] <ricotz> bbl
[07:33] <Dandel> ricotz, the issue is the lines it is looking for moved
[08:07] <ricotz> Dandel, keep an eye on tabs vs spaces
[08:08] <ricotz> pushed a fixed 40_...
[11:06] <zzippy> tseliot: hi! you have a minute regarding nvidia-prime package in saucy?
[11:06] <tseliot> zzippy: sure
[11:09] <zzippy> on my machine (and a few others in our forum) in 12.04.3 nvidia-prime is automatically installed when selecting the nvidiablob from restricted drivers. On the same machine in 13.10 this does not work.
[11:09] <zzippy> no driver suggested.
[11:10] <zzippy> (no problem to install manually) just wanted to let you know.
[11:12] <zzippy> also, since bumblebee is in the sources of 13.10, it might be better to mark this as "conflicted" in debian control file.
[11:12] <tseliot> zzippy: in 13.10 the UI doesn't support hybrid graphics, so that's expected. I hope to fix this in 14.04
[11:12] <tseliot> zzippy: good point
[11:12] <zzippy> ok, so I misunderstood your blog article ..
[11:13] <zzippy> ..thought that restricted drivers would also handle nvidia-prime in saucy.
[11:13] <tseliot> zzippy: that's explained in the wiki page, I don't remember if I mentioned it on my blog
[11:19] <zzippy> tseliot: ok, thanks, so I will update our wiki at ubuntuusers.de concerning software-properties-gtk in saucy.
[11:20] <zzippy> anything in work to make it possible to restart X on Intel (when battery stamina is needed)? We have to use ugly script to manage this in the moment ..
[11:21] <zzippy> (and steal bbswitch from bumblebee)
[11:22] <tseliot> zzippy: I'm working on it
[11:23] <tseliot> zzippy: and I'll make sure it's available in 12.04.4 and 14.04
[11:26] <zzippy> nice to hear. if we could do anything to help (testing)? Quite a few optimus machines around in our forum ..
[11:34] <tseliot> zzippy: sure but I expect to release my work in early December
[13:17] <Dandel> ricotz, last rebuild "worked" but didn't
[13:21] <ricotz> Dandel, i know
[13:22] <Dandel> it's actually not that bad of a bug :)
[13:22] <ricotz> i pushed a fix, but can't testbuild it here
[13:22] <ricotz> don't trigger a new build though
[13:23] <ricotz> g2g
[15:42] <Dandel> wb ricotz 
[15:42] <Dandel> the build worked :)
[15:43] <Dandel> now all that is needed is the relevant packages to add the opencl icd support and opencl 1.2 support to precise and the opencl support for piglit can be turned on.
[16:21] <Dandel> ricotz, is it me, or did the piglit build sequence skip a distro release? ( The 12.10 package was not rebuilt )
[16:27] <ricotz> Dandel, quantal was disable and i kept it so
[16:27] <ricotz> *disabled
[16:40] <Dandel> ricotz, No big deal... anyways, quantal did have one thing that precise lacks that all newer releases have. ( namely opencl 1.2 headers and opencl icd support ) 
[16:41] <Dandel> the only reason opencl testing is not enabled for piglit is the fact it'll break precise build badly.
[16:42] <Dandel> like all piglit builds, Opencl has it's own build param ( PIGLIT_BUILD_CL_TESTS )
[16:43] <Dandel> actually, on the arm devices it is possible to install piglit :) ( in theory )... just set PIGLIT_BUILD_GL_TESTS to false, and then leave the other build options enabled.