IRC channel logs

<rekado>does anyone here know the upstream status of ESS? It has become very hard to try to build ESS in Guix, but the project doesn’t appear to be dead.

<rekado>ACTION unlocked a couple hundred more packages for guix-bioc

<civodul>wo0t!

<civodul>70%

<civodul>(and still building)

<civodul>thought https://guix.bordeaux.inria.fr/workers is idle hmm

<civodul>*though

<civodul>fun, unattented upgrade had “upgraded” to the previous version of the config

<civodul>because mcron hadn’t been restarted (so it was still spawning the “old” upgrade job)

<civodul>the workers were all close to ENOSPC, which is why they were idling, waiting for the next “guix gc” job

<civodul>i wonder if ‘cuirass remote-worker’ should automatically GC when that happens

<civodul>a bit radical, but that’s usually what one would expect

<wdkrnls>I'm curious how researchers are using R on Guix.

<wdkrnls>Right now I have a manifest with R + Emacs packages.

<civodul>ACTION doesn’t use R

<wdkrnls>But that is less than ideal. I would have liked to Emacs to be a bit more separated.

<wdkrnls>err... liked for...

<civodul>i guess you could make two different manifests?

<wdkrnls>That obviously creates the shell environments I need. It's just that I'm not really sure how to get Emacs to use them.

<wdkrnls>Presumably it's somehow a matter of using tools like direnv or buffer-env.

<wdkrnls>I think it's just a common issue that if I do the same thing over and over then I figure it out.

<wdkrnls>When I stop doing it, then I forget completely.

<wdkrnls>In the past I had a nice way of connecting to a remote R session.

<wdkrnls>I probably could repurpose that separate R projects.

<civodul>yeah, buffer-env might work

<civodul>anyway, that’d make a good entry for the cookbook!

<wdkrnls>I agree. I keep getting distracted and then forgetting all my git skills :(

<wdkrnls>err patching skills.

<wdkrnls>But I'm getting ahead of myself. :)

<rekado>wdkrnls: are you using ESS to connect to a remote R?

<wdkrnls>I was in the past. Now I'm just taking the same approach.

<wdkrnls>But replacing all that magically ssh incantation stuff with calls to guix shell.

<civodul>hmm workers are approaching ENOSPC again, there must be a set of disk-consuming builds

<rekado>maybe the 1000+ new bioconductor builds?

<rekado>I expanded the set of packages to include experiment and annotation packages and they tend to be larger

<rekado>(the absence of these packages blocked hundreds of regular bioconductor packages)

<wdkrnls>rekado: As simple as https://paste.debian.net/1304859/

<wdkrnls>Now I just need to make an isolated version.

<wdkrnls>I know the cookbook already has that part :)

<wdkrnls>And it would certainly be convenient to have a helper function which creates well-commented template manifests for me.

IRC channel logs

2024-01-21.log