guix-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bug#74633] [PATCH] ui: Search channels for guix extensions


From: Brian Kubisiak
Subject: [bug#74633] [PATCH] ui: Search channels for guix extensions
Date: Tue, 3 Dec 2024 16:59:47 -0800

Hi Carlo,

> On Sun, Dec 01 2024, Brian Kubisiak wrote:
> > diff --git a/gnu/packages.scm b/gnu/packages.scm
> > index 80c22d1d7f..05b8bf8e6d 100644
> > --- a/gnu/packages.scm
> > +++ b/gnu/packages.scm
> > @@ -147,15 +147,16 @@ (define %package-module-path
> >    (let* ((not-colon   (char-set-complement (char-set #\:)))
> >           (environment (string-tokenize (or (getenv "GUIX_PACKAGE_PATH") "")
> >                                         not-colon))
> > -         (channels-scm channels-go (package-path-entries)))
> > +         (channels-scm (package-path-entries)))
> >      ;; Automatically add channels and items from $GUIX_PACKAGE_PATH to 
> > Guile's
> >      ;; search path.  For historical reasons, $GUIX_PACKAGE_PATH goes to the
> >      ;; front; channels go to the back so that they don't override Guix' own
> >      ;; modules.
>
> This comment should be moved onto add-channels-to-load-path!. Possibly
> even as a docstring.

Will do.

> > diff --git a/guix/self.scm b/guix/self.scm
> > index 2652688c71..28239d53f5 100644
> > --- a/guix/self.scm
> > +++ b/guix/self.scm
> > @@ -882,6 +882,7 @@ (define* (compiled-guix source #:key
> >                      ,(local-file "../guix/store/schema.sql")))
> >
> >                   #:extensions (list guile-gcrypt
> > +                                    guile-git     ;for (guix git)
>
> I don't know enough to know if this is a problem, but it's a shame to
> have to add this, given I don't think this change actually ends up using
> any git functionality.

Honestly I'm not sure what the are the downsides of this, since I
don't see how *core-module* would get used without *extra-modules* or
why *core-modules* doesn't include all the dependencies. Maybe Ludo
knows more?

> > diff --git a/guix/ui.scm b/guix/ui.scm
> > index eba12c8616..28690b22bc 100644
> > --- a/guix/ui.scm
> > +++ b/guix/ui.scm
> > @@ -2192,9 +2193,15 @@ (define* (command-files #:optional directory)
> >
> >  (define (extension-directories)
> >    "Return the list of directories containing Guix extensions."
> > -  (filter file-exists?
> > -          (parse-path
> > -           (getenv "GUIX_EXTENSIONS_PATH"))))
> > +  (add-channels-to-load-path!)
> > +  (let ((channels (package-path-entries)))
> > +    (filter file-exists?
> > +            (append
> > +             (parse-path
> > +              (getenv "GUIX_EXTENSIONS_PATH"))
> > +             (map
> > +              (cut string-append <> "/guix/extensions")
> > +              channels)))))
>
> I don't think you need the (append ...). According to the manual,
> parse-path takes another argument as a "tail" for the resulting list, so
> (parse-path (getenv "GUIX_EXTENSIONS_PATH") (map ... channels)) should
> be enough here.

I saw that on your patch, good to know!

> Am I right in thinking that this will look inside the channels at
> /guix/extensions to try to find things?

Correct. I chose this to match existing behavior, since
$GUIX_EXTENSIONS_PATH points at this directory for an extension
installed as a package. (I was able to tkae an existing extension and
pull it in as a channel and everything "just worked" without changing
any code or directory structure in the extension)

> That is, if I wanted to add a command "guix foo" I would need to
> define a module in $channel_dir/guix/extensions/guix/scripts/foo.scm?

This is not required, the command "guix foo" could be defined in
$channel_dir/guix/extensions/foo.scm. The command-files function
doesn't add /guix/scripts if a directory is passed as an argument and
 the run-guix-command function will resolve '(guix extensions foo)
if foo.scm exists in an extension directory
(i.e. $channel_dir/guix/extensions/foo.scm exists).

Thanks for the feedback,
Brian






reply via email to

[Prev in Thread] Current Thread [Next in Thread]