bug#49288: [PATCH] gnu: services: Export provenance-entry and sexp->channel

[Andrew Tropin]

Ludovic Courtès <ludo@gnu.org> writes:

> Hi!
>
> Andrew Tropin <andrew@trop.in> skribis:
>
>> These functions are need for Guix Home.
>> ---
>>  For now functions are copy pasted, but it seems better to reuse
>>  existing functions.
>>  
>> https://git.sr.ht/~abcdw/rde/tree/057888b2604b05d555fd1b67f3d356644f9a134d/gnu/home-services.scm#L418
>>  
>>  gnu/services.scm | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/gnu/services.scm b/gnu/services.scm
>> index 8d413e198e..53b358c63b 100644
>> --- a/gnu/services.scm
>> +++ b/gnu/services.scm
>> @@ -91,6 +91,9 @@
>>              ambiguous-target-service-error-service
>>              ambiguous-target-service-error-target-type
>>  
>> +            provenance-entry
>> +            sexp->channel
>
> This ‘sexp->channel’ exists “for historical reasons”.  Could you use
> (@ (guix channels) sexp->channel) instead, together with
> ‘channel-instance->sexp’?

Sure!)

> Actually, could you define it like so:
>
> --8<---------------cut here---------------start------------->8---
> (define home-provenance-service-type
>   (service-type
>    (inherit provenance-service-type)
>    (extensions
>     (list (service-extension
>          home-service-type
>            (service-extension-compute
>           (first (service-extensions provenance-service-type))))))
>    (default-value #f)))
> --8<---------------cut here---------------end--------------->8---

Looks good to me.

> More generally, perhaps you could have a procedure that takes a service
> type and “rewires” it so that it extends ‘home-service-type’ instead of
> ‘etc-service-type’, say.  Does that make sense?

Thank you for the idea, it can be useful, I'll consider it.

signature.asc
Description: PGP signature