Re: Astronomy related software in Guix

[Hugo Buddelmeijer <hugo@buddelmeijer.nl>]

Hi Oleg, Ole,

I love Guix, and astronomy, so I'm willing to do some of the work to
get astronomy software working.

I concur with Ole in the priority list; to summarize and to add some:
- Python stuff, which is mostly already in Guix, or can otherwise (for
now) be installed through pip
- ds9, wcslib, wcstools, cfitsio, ftools-fv
- Source-Extractor and other astrometic [1] software (SWarp, SCAMP for me)
- Java tools, Topcat [3] / Aladin [4], those are also just a jar-file
away in the meantime
- ESO (European Southern Observatory) software [2]: cpl, esorex
- Older stuff, IRAF, x11iraf, pyRAF, ESO-MIDAS

My plan was/is to contribute some of these packages, but so far have
not yet gotten to it. Most of my 'Guix-time' is spent trying to get
everything that already exists to work the way I want. For example,
getting power management to work on my laptop is next on my list.
However, what I would certainly do is send patches to keep these
packages updated if they would be in guix, because I would use most of
them.

I did manage to compile some of the software on a clean Guix system
(like cfitsio IIRC). I noticed that some of the old libraries rely on
/bin/cp and similar paths. I recall finding some release notes from
the 90s explicitly stating these paths are hardcoded to prevent
problems that probably don't exist anymore. I didn't know how to fix
those path-dependencies properly in a Guix package since at that point
I never made a Guix package, so I just added symlinks in my
config.scm. I realized that starting with converting those old
libraries would be biting off more than I could chew, so I'm planning
to first create some simpler Guix packages, like some more Python
libraries; I did my first 'guix import' experiments last week.

I do co-maintain some astronomy related conda-forge packages (some
with Ole), and those recipes (=build scripts) occasionally contain
some comments about problematic behaviour of these software, e.g. for
cpl [5]. I suppose the debian packages might also include some
workarounds that we could consult when creating Guix packages.

FWIW, I didn't mind becoming a conda-forge maintainer, because almost
everything is automated. E.g. a bot automatically creates a merge
request when a new version of a package is out and it is usually only
a single click ('merge') to get the conda package updated. I realized
I need at least some automation to commit to being a package
maintainer (even though that doesn't seem to be an official role in
Guix). So my goal is to first set up my own Cuirass server, and then
see what I can automate from there; but I haven't gotten to that yet,
because I'm still in the get-my-work-machine-comfortable-phase.

Hugo

[1] https://www.astromatic.net/software/
[2] https://www.eso.org/sci/software.html
[3] https://www.star.bris.ac.uk/~mbt/topcat/
[4] https://aladin.cds.unistra.fr/
[5] https://github.com/conda-forge/cpl-feedstock/blob/main/recipe/meta.yaml

On Mon, 16 Oct 2023 at 20:59, Sharlatan Hellseher <sharlatanus@gmail.com> wrote:
>
> Hi,
>
> Thank you Ole for your reply.
>
> I'd pin this thread in Guix Science mailing list to attract anyone interested in this subject.
>
> Some of the listed projects are already available (cfitsio, wcslib, wcstools, astropy, sunpy, jwst, ...) some are still in review (GDL, plplot).
>
> Working in collaboration with Iliya to bring Nemo to Guix.
>
> Regards,
> Oleg
>
> On Mon, 16 Oct 2023, 14:09 Ole Streicher, <olebole@debian.org> wrote:
>>
>> Dear Oleg,
>>
>> when we started with Debian Astro, we mainly had the same problem. At
>> the end, it depends on your needs, and the needs of your (potential)
>> community. I started with what seemed to be needed in my institute; that
>> was mainly ds9, wcslib, wcstools, cfitsio, ftools-fv. ds9 and ftool-fv
>> give the main functionality to display FITS images and tables; however
>> they are quite old now, and wcslib+cfitsio are the basic libraries.
>>
>> Then, there is the astropy ecosystem, which is today one of the building
>> blocks for data analysis; including ipython, jupyter notebooks,
>> matplotlib, scipy, numpy etc. This is however not a "game changer"
>> because people can (and will) install it from PyPI.
>>
>> Definitely worthwile to package is also Source-Extractor, and maybe
>> other software from astromatic.
>>
>> One other constituent is Topcat, which bring together with the program
>> itself brings a whole Java infrastructure to handle astronomy data files.
>>
>> Finally people are very happy to get older legacy packages ready-to-run:
>> IRAF, x11iraf, pyRAF, ESO-MIDAS. The GNU Datalanguage is a replacement
>> for the legacy IDL, which is also helpful for some people.
>>
>> I hope this small list helps you somehow :-)
>>
>> Cheers
>>
>> Ole
>>
>> On 10.10.23 02:50, Sharlatan Hellseher wrote:
>> > Hi Ole,
>> >
>> > I hope this message finds you well. I came across your name as a contributor to
>> > Debian Astro.
>> >
>> > I follow your updates  in https://salsa.debian.org/debian-astro-team
>> > and would like to ask for advise on priority you picked to prepare packages.
>> >
>> > I am also interested in using Guix for astronomy software and was wondering if
>> > you could offer some advice. Specifically, as an expert in astronomical software
>> > priority list for packaging, I was wondering if you could provide a list of
>> > priority astronomical/astrophysical software that would be beneficial to include
>> > in Guix. I understand that this may vary depending on the needs and interests of
>> > the community, but any guidance would be greatly appreciated.
>> >
>> > Guix provides source base destribution of software and
>> > which may be reproduced very strictly. It was successfully deployed in
>> > Bioinformatics researches already.
>> >
>> > Thank you for your time and contributions to Debian Astro.
>> >
>> > Regards,
>> > Oleg
>> >
>> > P.S. Some reference links
>> > - https://guix.gnu.org/en/about/ About the project.
>> > - https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/astronomy.scm
>> > Available packages definition related to astronomy, astrophysics, solar dynamics
>> > or galaxy dynamics.
>> > - https://packages.guix.gnu.org/ Packages search (26472 availalbe)
>> > - https://git.sr.ht/~hellseher/ffab/tree/main/item/TODO.md My personal list of
>> >    software queued to be packed.
>> >
>> > --
>> > … наш разум - превосходная объяснительная машина которая способна
>> > найти смысл почти в чем угодно, истолковать любой феномен, но
>> > совершенно не в состоянии принять мысль о непредсказуемости.
>> >