GNU bug report logs - #37868
[PATCH] guix: Allow multiple packages to provide Linux modules in the system profile.

Previous Next

Full log

View this message in rfc822 format

From: Danny Milosavljevic <dannym <at> scratchpost.org>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: Mark H Weaver <mhw <at> netris.org>, 37868 <at> debbugs.gnu.org
Subject: [bug#37868] [PATCH v6] system: Add kernel-module-packages to operating-system.
Date: Sun, 15 Mar 2020 23:09:04 +0100

[Message part 1 (text/plain, inline)]

Hi Ludo,

On Sun, 15 Mar 2020 22:00:04 +0100
Ludovic Courtès <ludo <at> gnu.org> wrote:

> I don’t think #:allow-collisions?, #:locales? and #:relative-symlinks?
> are needed, so I’d recommend removing them.

Removing allow-collisions.

Otherwise the defaults are different.

I'm pretty sure that we don't need locales for Linux kernel modules,
for example :)

That said, I can do it--but it would increase build dependencies.

> > +            (let* ((inputs '#$(manifest-inputs manifest))
> > +                   (module-directories #$(input-files (manifest-inputs manifest) "/lib/modules"))
> > +                   (directory-entries
> > +                    (lambda (directory-name)
> > +                      (scandir directory-name (lambda (basename)
> > +                                                (not (string-prefix? "." basename))))))  
> 
> also one-word identifiers are preferred for local
> variables.

I'd like to do that but it would lose information here.

"modules" would be too vague.  "directories" would be non-unique.
(What "module-directories" means is "'/lib/modules'-directories", literally)

"entries" would be too vague too.  Entries of what?
(Especially since that's a procedure).

I'll make it say "directory" instead of "directory-name" there.

Note:

The "existing-files" procedure exists only in order to allow us to
build Linux kernels without any modules (neither in linux-libre nor anywhere
else) and have the profile hook succeed.

Maybe it's written in an overly general way for that?  What do you think?

(It's actually kinda bad that I ignore kernel-loadable-modules
which have no "/lib/modules" in it (better would be an error)--but I wasn't
sure whether manifest-inputs is guaranteed to keep the original order of
the entries--which would be: linux-libre first)

[Message part 2 (application/pgp-signature, inline)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.