Ludovic Courtès <l...@gnu.org> writes:
> Hello!
>
> Ricardo Wurmus <rek...@elephly.net> skribis:
>
>> it seems that it is impossible to nest with-imported-modules. A gexp
>> that is wrapped in multiple layers of with-imported-modules won’t depend
>> on the list of all mentioned modules but only on the outermost.
>>
>> This is because with-imported-modules sets the current-imported-modules
>> parameter without checking if the parameter already has a value.
>>
>> Should nesting be supported? It seems useful.
>
> Is it? :-)
>
> My impression is that one would always write:
>
> (with-imported-modules … #~(…))
>
> or possibly combining multiple expressions:
>
> (define e1 (with-imported-modules … #~(…)))
> (define e2 (with-imported-modules … #~(… #$e1 …)))
>
> in which case modules are appended.
>
> I couldn’t think of a use case where one would literally write:
>
> (with-imported-modules …
> (with-imported-modules …
> #~(…)))
No, I’m not writing this literally, but I have an expression that uses
(guix search-paths), for example. Then I have another procedure that
takes this expression and wraps it with an expression that uses (gnu
build linux-container).
--8<---------------cut here---------------start------------->8---
(define (containerize exp)
(with-imported-modules (… (gnu build linux-container) …)
#~(begin
(use-modules (gnu build linux-container))
…
#$exp)))
(define (generate-exp)
(define exp
(with-imported-modules (… (guix search-paths) …)
#~(begin
(use-modules (guix search-paths))
…)))
(if container? (containerize exp) exp))
--8<---------------cut here---------------end--------------->8---
The observed effect was that only one of the with-imported-modules calls
had any effect on the final expression.
--
Ricardo
