IRC channel logs


back to list of logs

***rekado_ is now known as rekado
<zimoun>A reviewer pointed to me
<zimoun>and BioNix
<zimoun>civodul: since a post about the differences betwen Nix and Guix is still waiting, the reviewer is asking what is the advantage for Guix? Legitimate question. :-) Well, I have some arguments but maybe not the main ones. What is for you the 3 or 4 main differences?
*civodul thinks
<civodul>you need differences relevant to the paper(s) at hand
<civodul>since the BioNix spends time explaining the Nix language and shows code sprinkled with Bash snippet, you can already say something about the uniform language and semantic clarity
<civodul>then, if you're talking about GWL, you can say that BioNix is naive: everything flows through the store, there's no batch scheduler integration
<civodul>you can mention typing: Guix allows you to define disjoint data types, for packages, workflows, etc.
<civodul>BTW, it sucks that there's no GWL paper
<civodul>they cite GWL though, but what they write about it is irrelevant
<civodul>other differences: you could say Guix simplifies interoperability with "guix pack"
<PurpleSym>zimoun: I don’t see any old wip-haskell branch right now. I believe it was deleted when the merge was complete.
<zimoun>civodul: thanks.
<zimoun>usually, without deeply knowing Nix, I say that Guix is a continuum from declarative config to core using the same “language“.
<zimoun>Guix core and packages move at the same time which allows inferior; Nix does not have this feature out-of-the-box.
<zimoun>Ah, I did not know about ‘guix pack’, that Nix does not have such equivalent.
<zimoun>I do not know the status of SWH and Nix, any idea?
<zimoun>PurpleSym: I see wip-haskell gnu: ruby-pandoc-ruby: Disable tests.
<zimoun> wip-haskell-updates gnu: ghc-pandoc-citeproc: Build statically linked executable.
<zimoun> wip-haskell-updates-2 gnu: ghc-pandoc-citeproc: Build statically linked executable.
<PurpleSym>I think wip-haskell-updates are very very old and wip-haskell has been removed from the official repo (maybe not your clone).
<zimoun>Ah indeed, ‘git fetch --prune’ removed wip-haskell.
<PurpleSym>Alright, let’s move the topic out of bug#54729 to guix-dev, asking for more things that need to be fixed in Haskell-land, create a branch, fix the importer as much as possible and do a `guix refresh -t stackage` afterwards.
<zimoun>PurpleSym: the fixes of the importer can go directly to master. The aim of branch seems fix the build system and update stackage. Well, the fixes for the importer are bonus. :-)
<PurpleSym>Sure, importer fixes can go to master, but they are not a bonus. I’m not doing that stuff manually again.
<civodul>zimoun: Nix does have tools to build Docker images, but it's a "library" in the Nix language, so not as convenient
<civodul>and "library" really is too strong a word because Nix lacks modules, etc.
<zimoun>civodul: thanks
<zimoun>PurpleSym: yeah, I understand.
<zimoun>civodul: you pinged about patch#53882. I reviewed Julien patches and fixes minor things, then have just sent a v2 ready to push. Let me know.
<civodul>zimoun: oh nice, thanks!
<civodul>it's team work :-)
<civodul>i should be able to apply it later today
<zimoun>civodul, petit cachotier ;-) Or I missed your announce. Is it accepted? Or is it this one with «unfriendly review»?
<civodul>zimoun: heh, it's supposed to be published soon
<civodul>it's an "invited" article based on the --tune article, so that one was easy :-)
<civodul>funny how people eventually find out about things you put on-line
<zimoun>civodul: well, somehow you announced indirectly. ;-) Because you boosted which mentions the paper. :-)
<civodul>true! but i guess they found it via a search engine or something