Hi,

On Tue, 14 Apr 2026 at 05:36, "mr.doojin" <[email protected]> wrote:

> What about `bash:minimal` or `bash<minimal>` instead of
> `bash-minimal`? Either option makes it easier to recognize package
> variant names.
>
> `/`, `:`, `<`, and `>` are legal characters in guile symbols.
>
> This means package names can also be `emacs/magit:minimal` or
> `emacs/magit<minimal>` instead of `emacs-magit-minimal`.

Somehow, I’m repeating [1]:

         I’m not sure to get what is different from what it’s already there.

        Maybe.  However, this change comes with a lot of modifications that
        appear to me impossible in practise.  Well, 1. it would be tedious to
        have a transition plan – we cannot break all the manifests
        overnight. ;-)  And 2. this cost does not seem compensated by the
        readibility it would bring, IMHO, aside “category” is always a can of
        worms. ;-)

Well, I don’t know if this is discussed elsewhere in parallel.  If yes,
please point me where, so that I can read more context. :-)

That said, replacing the hypothetical package name `emacs-magit-minimal`
by something else isn’t the question.  The question is about the
transitional plan for the current package name of variants to other
package names.

BTW, having a fully documented scheme for package variants and switch to
it – hence coming with a cost! – needs to answer the ambiguity.

Considering the hypothetical `emacs/magit<minimal>`, what is minimal?
’emacs’ or ’magit’?  Here the example is difficult to resonate about.
Let try to point my point. :-)

Consider that Magit uses ’emacs-build-system’ which relies on
’emacs-minimal’.  Now, I would like to build Magit with the
’emacs-build-system’ but using another Emacs than ’emacs-minimal’.  How
do you capture this information with the scheme?


Cheers,
simon



1: Re: A new package naming convention?
Simon Tournier <[email protected]>
Mon, 30 Mar 2026 12:00:54 +0200
id:[email protected]
https://lists.gnu.org/archive/html/help-guix/2026-03
https://yhetil.org/guix/[email protected]

Reply via email to