[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: How to install GWL?
From: |
Ricardo Wurmus |
Subject: |
Re: How to install GWL? |
Date: |
Thu, 23 Jan 2020 14:12:41 +0100 |
User-agent: |
mu4e 1.2.0; emacs 26.3 |
Roel Janssen <address@hidden> writes:
> This would require us to package GWL in such a way that it doesn't have
> Guix as one of its inputs. Is that possible?
I don’t think so. I suppose the only way to avoid having Guix as an
input is by installing the GWL as a channel when “guix pull” is run, but
that route is fraught with problems. Guix is just *one* of a number of
inputs, and it is not clear how to install all of them via “guix pull” —
nor would that be desirable.
>> How should the GWL be installed for maximum convenience and
>> compatibility? Does it make sense to install it as a channel so that
>> it
>> is tied to the user’s current version of Guix? That would be pretty
>> awkward and less convenient than just typing “guix install gwl”.
>
> Or.. we merge the code from GWL into Guix, so it's automatically there;
> no install needed.
>
> I think the code is quite lightweight, and since it uses the Guix
> modules, it is somewhat tied to a specific version of Guix.
>
> What's the reason for not wanting GWL directly in Guix?
The code won’t remain lightweight forever. The GWL may gain support for
more advanced caching of intermediate results through external
libraries; it may gain support for deployment to virtual machines in the
cloud (e.g. via Guile AWS); it may depend on (or include) grid engine
bindings to provide better support for cluster execution. None of this
really belongs to Guix itself, which already has a very wide scope.
I’d rather see Guix move into the opposite direction and embrace
extensibility by providing a stable API for its core modules. With the
GWL we have an opportunity to demonstrate how Guix can be extended
without having to actually be part of it.
--
Ricardo