Hey everyone!
Making Nix solve people’s problems 
- @solene got a grant from NLNet NGI0-entrust to build Genealogos, a CycloneDX SBOM generator from Nix packages, including a public website that will contain pre-generated SBOM for every package in nixpkgs, with a human friendly viewer.
Documentation 
- @jlesquembre opened a PR to display versioned manuals on nixos.org
-
@fricklerhandwerk and @balsoft merged their PR to display
nix-storeandnix-envsubcommands on separate manual pages, including all common options and environment variables- check it out on the latest online manual
- it also works for
man nix-store-<subcommand>andnix-store --<subcommand> --help
- @infinisil, @fricklerhandwerk, and the newly formed documentation team devised and submitted a project proposal for Google Season of Docs 2023
Python bindings
Thanks to some funding from Antithesis, @infinisil started working a few weeks ago on providing python bindings to Nix. @yorickvp recently joined the effort, bringing in his deeper knowledge of the internals of Nix.
Making Nix work reliably everywhere 
-
@connorbaker continued his work on CUDA
- Rebased and refactored cudapackages.nccl: refactor to fix #220340 and #221895 #217619
- Now also closes new issue cudapackages.nccl: only ever builds with default version of cudaPackages #221895
- Merged tiny-cuda-nn,python3Packages.tiny-cuda-nn: init 1.6 #218166
- Created and merged python3packages.torchvision: switch to backendStdenv.cc from cudatoolkit.cc #221950
- Provided guidance on python3.pkgs.mmcv: init at 1.7.1 #219815
- Reviewed (several times) and suggested fixes for python3packages.torch-bin: 1.13.1 → 2.0.0 #221652
- Updated and merged magma: fix #220343 and separate CUDA build/run-time dependencies #220366
- I am now the Magma maintainer
- Closed magma: lexicographic sorting of CUDA capabilities is not future-proof #220343
- Created python3packages: torch-bin and friends should move to autoPatchelfHook and autoAddOpenGLRunpathHook #222473
- Preliminary review of python3packages.torch: 1.13.1 → 2.0.0 #222273
- Merged opencv: misc CUDA-related updates and fixes; add enableLto #221370
- Rebased and refactored cudapackages.nccl: refactor to fix #220340 and #221895 #217619
-
@thufschmitt and @infinisil helped to move RFC136 forward, with the hope that consensus could be reached within a couple of weeks
-
@infinisil is looking for some help to shepherd the RFC140
-
@balsoft and @thufschmitt made it possible for the new CLI to connect to SSH hosts with password authentication
-
While working on the above, @balsoft made the integration tests slightly lighter to use
-
@thufschmitt worked on fixing an issue with the Nix profiles location in the latest Nix release, through trying to add a compatibility layer and fixing an overlook regarding root’s channels profiles
Making Nix ubiquitous 
Blog posts and media 
- @infinisil published two Nix Hour recordings
- @djacu is working on the next “What’s new in Nix” video
Meetups
- @arsleust organized a Nix meetup in the Paris Tweag office gathering some twenty passionate Nixers from all around Europe
Community organization
- @thufschmitt wrote an early draft of a trademark policy to open it up for community discussion
- @thufschmitt reworked an old draft clarifying the role of the foundation in the ecosystem to address some remarks on it
- @thufschmitt and the rest of the foundation board started prioritizing for the coming quarter
Nickel
- The team worked on adding missing stuff to the stdlib, as well as making a pass on the documentation (#1195, #1196, #1203)
- @vkleen is adding right now this documentation on the website https://nickel-lang.org (in progress)
- @vkleen found a non-trivial interaction between lazy contracts, recursive fields and record mapping. We worked to try to find a (semantically) clean solution, which he implemented (#1194)
- @yannham opened an issue about merging arrays not being idempotent, and started preliminary work to fix it (#1203, #1211, #1213)
- @yannham started to improve the implementation of the type checker again, in particular to follow the bidirectional type checking discipline more closely (part 1: #1193)
- @ebresafegaga fixed a bug in the LSP when an import failed to type check (#1191)
- @ebresafegaga worked on an emacs mode for Nickel
- @matthew-healy continued to update string functions to use Unicode grapheme clusters as the atomic unit (#1200)
- @dpl0a and @yannham continued to work on incremental evaluation, focusing on merging for now. It’s almost there, but they are hunting the last train of bugs.
may Nix become more visible 
