[aur-general] Multiple Kernel modules in the AUR [WAS: Disown request: nvidia-pae]

Oon-Ee Ng ngoonee.talk at gmail.com
Mon Sep 17 23:14:44 EDT 2012

On Tue, Sep 18, 2012 at 3:47 AM, Dave Reisner <d at falconindy.com> wrote:
> On Mon, Sep 17, 2012 at 09:38:30PM +0200, Robert Knauer wrote:
>> Hello,
>> please disown nvidia-pae[1], it's outdated for more than a month now
>> and I mailed the maintainer on 1st of September and got no answer.
>> Thanks,
>> Robert
>> [1] https://aur.archlinux.org/packages.php?ID=40101
> Don't take this personally... you're just the one who happened to bring
> it up. Do we _really_ need all this duplication? Does every kernel in
> the AUR needs its own _from_ _source_ instructions to build kernel
> modules? Really, these should all be about 3-4 lines to change in the
> extra/nvidia PKGBUILD. Instead, we have...
> nvidia-apparmor
> nvidia-bede
> nvidia-bfs
> nvidia-bl
> nvidia-ck
> nvidia-custom
> nvidia-fbcondecor
> nvidia-ice
> nvidia-ll
> nvidia-lqx
> nvidia-mainline
> nvidia-pae
> nvidia-pf
> nvidia-rifs
> nvidia-rt
> nvidia-uksm
> nvidia-zen
> I'm sure these are all unique and beautiful in their own way, but
> really, they're all duplicates as far as I'm concerned. nvidia is the
> biggest offender of this, but it certainly applies to other modules as
> well.
> </rant>
> dave

Firstly - full disclosure - I'm the author and maintainer of the
hackish nvidia-beta-all package (someone else based nvidia-all from
that). It basically builds the nvidia module for every exiting kernel
(including specific patches if needed).

When initially submitting the package there was some discussion about
whether such a PKGBUILD which is non-reproducable (would produce
different results on different machines) violates the AUR's
guidelines. The only alternative to that (if reproducability is
important) would be the various nvidia packages as we currently have.

Of course, I never really agreed with that line of reasoning simply
because as it is compiling any random package on two different
machines could produce different results due to autotools or similar.

More information about the aur-general mailing list