IRC channel logs
2026-01-23.log
back to list of logs
<Kabouik>Hey Guix-HPC. I see in the 2024 activity report that some research institutions have dedicated work hours to Guix, including on some people whom I think do not come from IT fields originally. My institution is not listed, and I am by no means skilled enough in Scheme/Guile/Guix to remotely consider asking my institution to assign such tasks, but would be curious to know what the process was to make those institutions deem Guix contributions one of <Kabouik> their relevant missions, and officially endorse that with some dedicated person-year. <Kabouik>(If I were to become more fluent in Guix, I certainly would be interested in contributing a little bit as part of my work in a French research institution. At least, if they recognized that as a relevant mission, I could more easily justify spending some time to train.) <civodul>contributions to the activity report are purely volunteer <civodul>if you’re real quick, you can contribute to this year’s report, due for publication next week! <civodul>we’d love to know what you’re doing with Guix at your institution <civodul>and of course it makes sense to have it listed under “Personnel” <Kabouik>I think I did already civodul, but just for a very minor contribution (listing a paper that used Guix as reproducibility tool). <civodul>hmm i guess i haven’t figured out the nickname/“real name” mapping <Kabouik>Yeah I didn't really intend to make it straightforward anyway since Guix is at the interface between personal interest and work tool for me, but nothing officially tied to my work missions. <Kabouik>I was interested in seeing that CNRS or some universities seemed to have officially agreed on person-year amounts for some people whom, to my knowledge, are not in positions of HPC admins and such (but I may be wrong). I was thinking about Konrad for instance, but maybe I'm wrong and he has ties with deploying Guix in HPCs. I think it's cool, actually. <Kabouik>Of course seeing Inria or some other establishments, and HPCs strctures show person-years in Guix contributions is no surprise. <civodul>well, people doing the work, not institution executives, are adding a line to “Personnel” <civodul>but yeah, there’s quite a few people working with or on Guix as part of their scientists/engineer/sysadmin position <civodul>(certainly more than appears in the report actually!) <Kabouik>I see, so it all comes to whether those individuals actually upstream that part of their work to their evaluation committees/teams/units when they send their evaluation, and if this mission is actually visible or not outside Guix. <Kabouik>In the 2024 report, the sentence is "As part of Guix-HPC, participating institutions have dedicated work hours to the project", which does make it sound like Guix development/packaging/HPC admin/advocating for Guix as a reproducibility tool is one of the many possible scientific missions in those institutions (again, no surprise for Inria employees or people with strong ties with cluster structures). <civodul>how exactly it works depends on the institution <civodul>CNRS has 40k employees, so whether one person or team works on X is really a local decision <civodul>Inria is 10x smaller, so it’s different <Kabouik>Exactly, that's why I was curious/interested in how it went for others when I noticed INRAE not appearing. <Kabouik>Inria has deep roots into Guix and the whole thing is also very aligned with the institute's scope, I guess, which makes it natural. I believe it'd be cool and encouraging if more institutes (in France at least) were involved in a way or another. <Kabouik>I certainly try to advocate for Guix around me and just started to train students as much as my skills allow. Hopefully it will gain in popularity in my field, but may not bear fruit quickly since Linux users are very scarce here (and the IT department doesn't really encourage it, to say the least). <Kabouik>Things are less hostile at a broader scale though, so there's room to make using/contributing to Guix more visible to the institute for its reproducibility benefits and get good feedback.