[bug#66968] [PATCH] gnu: linux-libre: add linux-libre-stable & linux-libre-longterm

[Maxim Cournoyer]

Hi Leo,

Leo Famulari <leo@famulari.name> writes:

> On Fri, Sep 13, 2024 at 11:36:10PM +0900, Maxim Cournoyer wrote:
>> >  (define-public linux-libre-headers linux-libre-headers-5.15.49)
>> > +(define-public linux-libre-stable linux-libre-headers-6.5)
>> > +(define-public linux-libre-longterm linux-libre-headers-6.1)
>> 
>> Sounds like a reasonable thing to have.  Leo, Wilko, what do you think?
>
> The package named 'linux-libre' is by definition the latest stable
> kernel:
>
> https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/linux.scm?id=22a34ea792ef0df15fd30d46f557b572c61d5404#n513
>
> Should we add an alias for it?

This wasn't requested... I guess the main use here is using older kernels.

> We also have a package called 'linux-libre-lts', which is supposed to be
> the most recent kernel with long-term support.

Michael, did you know about 'linux-libre-lts' ? It seems it would suite
your 'linux-libre-longterm' request well?

> I think the phrasing "longterm" is more idiomatic for Linux, so we could
> replace the -lts package with a -longterm package if people prefer that.

Sorry, I had forgotten we already had -lts.  I guess we're already
covered then.

For the linux-libre-headers, I agree that the current situation is odd
at best; it should be renamed to linux-libre-headers/pinned or
something, since it's propagated by our build toolchain if I recall
correctly and entails a world rebuild.

linux-libre-headers could then point to the latest and greatest,
in sync with linux-libre.

Does that make sense?  Renaming the package variable shouldn't cause any
rebuild.

-- 
Thanks,
Maxim