[Slackbuilds-users] RFC: nvidia-driver organization
gerardo.zamudio at linux.com
Fri Jul 8 15:34:01 UTC 2016
-----BEGIN PGP SIGNED MESSAGE-----
On 07/08/2016 01:44 AM, King Beowulf wrote:
> The nvidia-driver has become a horrible mess to maintain. Nvidia
> is moving to GLVND and Vulcan, and several libs, binaries have
> moved to F/OSS, and Nvidia keeps adding/renaming some libs. These
> changes, along with compat32 support (not supported in SBo but it's
> a fan favorite), are making it confusing to generate a consistently
> bug-free script. I'm thinking of splitting it up into multiple
> nvidia-kernel will remain the same.
> 1. nvidia-utils: -Contains all the open source binaries. -Scripts
> ts similar to nvidia-switch to backup and restore Mesa and X.Org
> components that Nvidia overwrites, as well as dealing with
> 2. nvidia-driver: A much simpler script that just assumes a pure
> x86_32 or x86_64 installation.
> 3. COMPAT32 provided either via Alien Bob's multilib
> convertpkg-compat32 or my own script in nvidia-utils
> 4. Finally, instead of all the "install' lines in the SlackBuild
> script, leverage the nvidia-installer or the NVIDIA-*.run to
> perform the install into /tmp/SBo... for subsequent packaging. This
> will have the advantage of making the nvidia-driver.SlacBuild less
> brittle for simple version bumps. The disadvantage is that this
> will break a few things the script now does.
> I know a many who simply run the NVIDIA-*.run and are happy. For
> those on this list who run my scripts, what say you? Also, let me
> know (off list) if you are willing to assist in thrashing your
> system for testing. I have just the one GTX600 system atm, so
> testing other GPUs and desktops will be nice.
> Thanks, Ed
Thanks for your efforts, Ed. I use your NVIDIA scripts exclusively for
my system and I'm in favor of your approach. I would prefer COMPAT32
be provided by your script.
-----BEGIN PGP SIGNATURE-----
-----END PGP SIGNATURE-----
More information about the SlackBuilds-users