Hi Arun,
I fully agree with your pain, and I don't know the solution.
On 30/6/26 18:28, Arun Isaac wrote:
But, I find that surprisingly many researchers want to use fairly niche
software packages that are specific to their domain. These packages are
usually not packaged for Guix, and the researchers end up having to
package it themselves. As a result, they are suddenly swamped by a full
package definition in Scheme. After a few heroic attempts at making it
work, they give up and go back to conda.
This is conflating two different things, because those conda packages
already exist.
Creating a conda package is significantly harder than creating a guix
package. Both in a "just for me" hackish way and in a "for the
collective" proper way.
At least for the few conda packages I maintain; their Guix counterparts
are much simpler.
And conda(forge) even does everything right. There is a limit to what
YAML files can reach, and conda has reached it. A major selling point
of Guix is that through scheme it transcents that limit.
> I even tried to write such tooling but was deterred by the
> complexity of programmatically editing arbitrary Scheme code.
The corollary of that is that we can write transformation and such; we
can modify the packages directly in scheme, which gives enormous power.
With conda, you are essentially stuck with whatever decision some
package maintainer has made for you.
(I think `guix style` and `guix import` use `object->string` to convert
the scheme objects back into files.)
But yeah, I'm still using Guix to start podman containers with fedora
and run conda in there, because it is a crazy amount of work to get all
required packages in Guix. Luckily we are finally on Python 3.12!
Hugo
