IRC channel logs

2022-05-23.log

back to list of logs

***Server sets mode: +nt
<zimoun>hi!
<civodul>hello!
<PurpleSym>What’s our policy regarding R packages depending on some texlive packages? My problem right now is that r-prereg wants to `knitr` a PDF by default, but it lacks the proper LaTeX packages to do so and fails. Should they be added as propagated-inputs?
<PurpleSym>I don’t see `r-knitr` itself depending on any LaTeX stuff.
*civodul doesn't know
<civodul>looks like we have Julia build failures on x86_64: https://guix.bordeaux.inria.fr/build/11822/log/raw
<civodul>well test failures, numerical ones even
<zimoun_>civodul: hum, what is the CPU of Bordeaux? Because it looks like the kind of errors dependent on the microarch.
<civodul>zimoun_: good point! one of the build machines is an old thing that sometimes causes problems
<civodul>Intel(R) Xeon(R) CPU E5-2650L 0 @ 1.80GHz
<civodul>so yes, that may well be the problem
<rekado>PurpleSym: I’d prefer to only add the texlive-* stuff to r-prereg
<PurpleSym>rekado: I tried that, but unfortunately it fails with: `mktexpk: don't know how to create bitmap font for ec-lmcsc10.` 😕
<PurpleSym>The same set of packages works on the command-line.
<zimoun_>PurpleSym: I am back to work. :-) Any progress on the Haskell LTS upgrade? I am going to resume this week.
<PurpleSym>zimoun_: I fixed some issues with the importer (see https://debbugs.gnu.org/cgi/bugreport.cgi?bug=54760 and https://debbugs.gnu.org/cgi/bugreport.cgi?bug=54752), but they are not pushed yet.
<PurpleSym>(Feel free to review.)
<PurpleSym>Progress is slow, because I can only work on that during the weekend right now.
<zimoun_>PurpleSym: Thanks. I will give a look so.
<zimoun_>civodul: Julia builds fine on my workstation. It is highly probable that the issue comes from the hardware. And I even bet that the failures are not deterministic reproducible. ;-)
<civodul>uh, could be!
<efraim>works fine for me™!
<zimoun_>civodul: do not miss this 20min of French https://replay.jres.org/w/f1rZTkh84cNimWf4aXHxN1 :-) About deployment of Jupyter.
<rekado>next week I’m supposed to present Guix to MDC folks; ideally with a workshop component where people get to do a few exercises.
<rekado>do you happen to have some training materials that I could liberally copy from?
<rekado>(I always seem to schedule these things on the first day after my holidays, and I don’t really want to prepare so much during my time off)
<drakonis> https://sr.ht/~lle-bout/awesome-guix/#learning-resources
<rekado>thanks, I’m familiar with the links there.
<rekado>I was looking for something tailored to scientists / HPC users
<drakonis>ah
<drakonis> https://git.savannah.gnu.org/cgit/guix/maintenance.git/tree/talks
<drakonis>highly suspect that some of the presentations on the matter haven't been uploaded
<drakonis>not very useful then
<zimoun_>rekado: I did some stuff but in French https://gitlab.com/zimoun/jres22-tuto-guix
<zimoun_>I will convert some in English but not next week.
<zimoun_>it is an introduction. Without exercise.
<zimoun_>The structure is : 1. CLI search/show/install/remove and present environment variables.
<zimoun_>2. present declarative manifest
<zimoun_>3. use this manifest for introducing transformation
<zimoun_>the slides (French) are there https://zimoun.gitlab.io/jres22-tuto-guix/presentation.pdf
<zimoun_>rekado: how long is your presentation?
<rekado>1.5 to 2h
<zimoun_>bionfo background or else?
<rekado>yes, bioinfo people
<civodul>rekado: i watched the video and zimoun showed it all: guix package, generations, rollbacks, guix shell, manifests, describe/time-machine, etc.
<civodul>it's at https://replay.jres.org/w/3TuYmocHwKtzs7q1VtL1GB
<civodul>but i gather you're supposed to be on holidays :-)
<civodul>anyway i really enjoyed the way zimoun_ presented it all
<rekado>zimoun_: this is great, thank you! I’ll study your materials and watch the recording.
<rekado>(It’s a pity I don’t understand French… I’m missing out!)
<zimoun_>I am planning to transform the material into English, but not for next week.
<zimoun_>FWIW, I think the progression from day-to-day to core concepts is adapted to scientific practitioners. Well, it is rooted in Python and Numpy but I would like to have a variant rooted in R and bioinfo.