IRC channel logs

<zimoun>«Most package managers need a dependency solver, but dependency solving is an NP-hard problem, and writing a correct solver from scratch is difficult to do correctly, let alone a fast solve». If the first statement is refused, what happens?

<zimoun> https://pretalx.com/packagingcon-2021/talk/LJ9HJK/

<zimoun>Because of solvers, Spack is not “reproducible” by design.

<zimoun>Spack is “just” modulefiles on HUGE steroids from my understanding.

<civodul>+ build system, tooling, convenient UIs, etc.

<zimoun>yeah, HUGE steroids. ;-)

<zimoun>I mean, Spack will have the same issues as Modulefiles, at one point or another. It is not necessary a bad thing (each cluster is different) but a fact. :-)

<zimoun>Other interesting slots https://pretalx.com/packagingcon-2021/talk/H8FDJL/

<zimoun> https://pretalx.com/packagingcon-2021/talk/J3CHVC/

<zimoun> https://pretalx.com/packagingcon-2021/talk/K8GPRA/

<zimoun>Wow! Many interesting materials :-)

<zimoun>I would recommend you https://pretalx.com/packagingcon-2021/talk/K8GPRA/ about Pkg.jl the Julia package manager which implement content-address via UUID.

<zimoun>civodul, cool you pushed --check. :-)

<civodul>yeah, fearlessly!

<civodul>we can fine-tune over time

<rekado>I’m getting a huge guix inferior backtrace in my pigx-web application. Nothing that looks useful, though.

<rekado>so, today I gave a talk that I had not practiced before. I used pdfpc (newest version from core-updates-frozen) and I can recommend it.

<rekado>it’s standard presentation software in that it shows your notes and the next slide on your laptop (“presenter view”) and the current slide on the second monitor.

<rekado>I had pretty good notes, so I just followed those during the talk.

<zimoun`>rekado: pdfpc looks nice, I used Emacs and pdftools – poor man solution. :-)

<rekado>yeah, that’s also what I used before, but that only works if you know what you’re doing and what you’re going to say :)

<rekado>I got sick and could rely 100% on my notes

<zimoun`>that’s cool! the version on master does not own enough features?

<rekado>it has all the features I use, but on core-updates-frozen there are a lot more that I haven’t tried yet

<rekado>e.g. embedding of videos, subtitles, etc

<rekado>transitions too.

<rekado>I only remembered pdfpc because it didn’t build when I upgraded my packages after switching to core-updates-frozen.

<rekado>and upgrading it was the only good way around the build failure.

<rekado>the webinar I attended today had a proprietary solution for “reproducible” genomics analysis scheduled two slots before my talk on our new PiGx web thing.

<rekado>and of course one of the questions was about comparing Guix to (bio)conda, which always feels like a category error to me.

<rekado>but I think I accidentally came up with a fair response that acknowledges the role bioconda plays in the bioinfo community

<rekado>another odd error: the info-dir derivation fails, but there’s no output at all, so also no logs. No idea why it fails.

<rekado>looks like this: https://elephly.net/paste/1635265663.html

<zimoun`>rekado: could you quickly sketch the arguments Conda vs Guix you used?

<zimoun`>about info-dir, no idea. Is it from core-updates-frozen or master?

<rekado>this is on a somewhat older version of the master branch

<rekado>pro conda: *very* widely used, so you’ll likely encounter it as the supported installation method for new bioinfo tools. So conda can be used to quickly test if the tool is worth using.

<zimoun`>rekado: I see, thanks.

<civodul>rekado: re info-dir.drv, that vaguely rings a bell; could it be https://issues.guix.gnu.org/50264 ?

<rekado>civodul: I’ll pull and see if that fixes it.

<civodul>does pdfpc work with exwm?

<civodul>i can't seem to get it to display notes

<civodul>i have a feeling that these tools are always almost there, but not quite

<drakonis>so, i'm having some weirdness regarding guix-science right now

IRC channel logs

2021-10-26.log