In reply to @terru:raccoon.college
Question about environment variables which nixpkgs reads to configure itself: these all have the prefix NIXPKGS_ or NIXOS_, except NIX_ABORT_ON_WARN. I find this very confusing, as it suggests that this variable is read by nix itself, which it is not. It was introduced in #140763, apparently without consideration of this inconsistency. Is there any good reason why it should not be called NIXPKGS_ABORT_ON_WARN instead, and if not, what is the process to change it? rename it, and for a release cycle still also accept the old variant, but produce a trace if it is set?

Maybe?:

concurrency:
  group: ${{ github.ref }}
  cancel-in-progress: true

I was goofing around with this tool last night https://discourse.nixos.org/t/genix7000-nix-project-logo-generator/15937/37

nix run github:cab404/genix7000#to-image -- 5flake.svg \
  --num 5 \
  --thick 20 \
  --rotation 339 \
  --angle 120 \
  --clipinv true "\#5277C3" "\#7CAEDC" "\#6989D8" "\#6A93CE" "\#819DD1"

I have an open issue regarding Python where I think it's worth considering the overall implications for nixpkgs infrastructure.
See the bit under the heading Reducing evaluation overhead.

This is basically an instance of https://discourse.nixos.org/t/1000-instances-of-nixpkgs/17347, just inside nixpkgs.

Python isn't the only ecosystem where the packageOverrides idiom comes and bites us performance wise, but it is the most common source of such issues.