Re: [PATCH] Add package definition for GNU XaoS.

[ludo@gnu.org (Ludovic Courtès)]

Alex Kost <alezost@gmail.com> skribis:

> Alex Kost (2015-12-21 15:42 +0300) wrote:
>
>> Fabian Harfert (2015-12-15 23:02 +0300) wrote:
>>
>> [...]
>>> @@ -53,6 +54,7 @@
>>>    #:use-module (gnu packages ghostscript)
>>>    #:use-module (gnu packages glib)
>>>    #:use-module (gnu packages gtk)
>>> +  #:use-module (gnu packages image)
>>
>> Our (gnu packages image) module uses (gnu packages maths).  I wouldn't
>> add this circularity, I think it was better to have xaos in a separate
>> module.  But actually I don't know if it's a real issue and what our
>> policy on such things is.  I hope more experienced guix/guile people
>> will tell if we should avoid such circularities.
>
> Sorry for bumping, just to prevent this package from burying in ML.
>
> Originally Fabian sent a patch for separate "xaos.scm" file.  Then
> Andreas suggested to move it to "maths.scm".  But this will lead to the
> mentioned circularity:
>   (gnu packages image) already uses (gnu packages maths)
> and after this patch:
>   (gnu packages maths) will use (gnu packages image).
>
> Is it OK to leave it like this, or should 'xaos' be left in a
> separate file?

It’s OK to leave it as is.

Module circularity are not a problem, unless there are circular
*top-level* references.

That is, if maths.scm does, say:

  (define foo libpng)

and image.scm does:

  (define bar lapack)

then we have a problem.

Ludo’.