IRC channel logs


back to list of logs

<efraim>I'm pretty sure that's the use case for it but I've been hesitant to go and tag them
<PurpleSym1>civodul: I have to go in a few minutes, but we can have a look at the CRAN blog post later today.
<PurpleSym1>Completely unrelated, but I’ve also been looking into how to get Guix into Windows’ WSL2 as a proper distribution (like Ubuntu). There is a template project here: which could easily be adjusted to Guix. There is no obvious way to build it on Linux, so someone would have to build it on Windows. And then do the app-store dance to publish it.
<PurpleSym1>Not sure if that’s something the Guix maintainers would be willing to support.
<PurpleSym1>But given that alot of people run Windows that would be a good way to “introduce” Guix.
<civodul>PurpleSym1: re the blog post, sounds good!
<civodul>re WSL2, we'd have to ask the maintainers, but i think having an official WSL2 image would be reasonable
<civodul>(same for Docker actually)
<PurpleSym1>civodul: Alright, I see the point now why you referenced that Nature article. Shouldn’t we highlight GWL instead? Guix itself only solves the dependency problem. You can still hard-code paths and end up with the same mess.
<PurpleSym1>Wrt R version guesswork: Absolutely, that’s something Guix will solve.
<civodul>PurpleSym1: so the "source cleanup" is about absolute file names, and yes, this is more GWL's area
<civodul>the R version guesswork is Guix's area
<PurpleSym1>I think that was one of the things they did and something I see quite frequently when reading non-computer scientist R scripts.
<civodul>for clarity i'd focus on what's in Guix's area in this context
<civodul>(same with Jupyter notebooks)
<PurpleSym1>A short section at the beginning mentioning the various issues of packrat/renv would be nice, I agree.
<PurpleSym1>I just tried “reproducing” an renv lockfile, which didn’t work, because CRAN deletes binary packages when a new version is released 🤦‍♂️
<PurpleSym1>And obviously the Windows system had no compiler installed.
<PurpleSym1>I’ll try to come up with an introduction.
<civodul>that you couldn't reproduce an renv lockfile is another good example of how guix-cran improves the situation
<civodul>this kind of hands-on explanation "from the trenches" is invaluable
<civodul>it's concrete and R users in this case can relate to it
<rekado>insert here “awkward look monkey puppet” meme
<rekado>because bioconductor deletes *source* archives and we haven’t yet switched to git