Alex Kost <alez...@gmail.com> writes:
> Mathieu Lirzin (2016-01-26 00:41 +0300) wrote:
>
>> Finally, There is some code redundancy for the emacs UI but If I
>> remember correctly there is a reason for that. (Alex?)
>
> If there is redundancy in the code, it should be removed. I didn't find
> a procedure to make a name specification at the time, so I wrote
> 'make-package-specification' in "emacs/guix-main.scm". I think we can
> add a general procedure to construct a name specification to (guix
> utils):
Sorry, my mind got confused. ;)
--8<---------------cut here---------------start------------->8---
(define (full-name->name+version spec)
"Given package specification SPEC with or without output,
return two values: name and version. For example, for SPEC
\"foo@0.9.1b:lib\", return \"foo\" and \"0.9.1b\"."
(let-values (((name version output)
(package-specification->name+version+output spec)))
(values name version)))
(define (name+version->full-name name version)
(string-append name "@" version))
(define* (make-package-specification name #:optional version output)
(let ((full-name (if version
(name+version->full-name name version)
name)))
(if output
(string-append full-name ":" output)
full-name)))
--8<---------------cut here---------------end--------------->8---
I overlooked that this code handles “full names” without having a
<package> object as an argument.
> (define* (package-name-specification name #:optional version output)
> (let ((full-name (if version
> (string-append name "@" version)
> name)))
> (if output
> (string-append full-name ":" output)
> full-name)))
>
> Then 'package-full-name' from (guix packages) and the code from
> "emacs/guix-main.scm" can use it. WDYT?
It seems a good idea to me now.
Thanks,
--
Mathieu Lirzin
