Hi Simon,
On Tue, 27 Oct 2026 at 12:22, Ludovic Courtès <[email protected]>
wrote:
This growth problem is not entirely new: we wouldn’t have all
of CRAN or
PyPI in Guix proper. In the former case, it’s good that the
separate
Guix-CRAN channel exists, and that Guix itself doesn’t aim to
provide
all of the 30k packages or so of CRAN. Yet, vibe coding gives
the
problem a new dimension and could be the basis of a simple
package
inclusion criterion.
About this, it reminds me some ideas [0,1] discussed months ago:
moving
peripheral packages to channels (still maintained by Guix, but
as
separate repositories).
Good candidates to be extracted seem the current Guix proper
packages
from CRAN and/or Bioconductor, then humanly maintain them into
one or
more dedicated channels – unrelated to the fully automated
guix-cran and
guix-bioc channels.
CRAN and Bioconductor packages could certainly move. Inclusion in
Guix came first, the separate channels second. Now I think it
would be fine to move them all to the separate channel, as long as
we keep a distinction between those that have been "audited" and
those that are just automatically generated with no guarantees at
all.
The big obstacle to even using the guix-cran and guix-bioc
channels is that it makes Guix operations rather slow.
That said, R packages are very unusual in that there are very few
other packages that depend on them. Extracting other packages to
separate channels is likely much more problematic.
--
Ricardo
