Flakes are now in stable and the writing is on the wall that they are the way forward. It's a standardize API that offers faster shelling and is easier to update nix flake update instead of fussing about the nixpkgs Git repository, finding a commit hash, and then getting it's sha256 hash. The approach in this merge request offers backwards compatibility with the Flake-skeptical and older systems via flake-compat.
What needs testing: anything related to darwin. I don't have access to any of those types of machines.
Opinionated fix: move to nixpkgs-fmt over nixfmt. Like my gripes with prettier, nixfmt moves your code around too much to achieve its opinionated goals which causes unnecessary merge conflicts. String now needs one extra character? Better break all the lines apart and make a massive diff vs. nixpkgs-fmt whose goals are better for maintainability than aesthetics. One could easily compare this to moving from purescript-tidy from purty (though sadly no config file).
Flakes are now in stable and the writing is on the wall that they are the way forward. It's a standardize API that offers faster shelling and is easier to update
nix flake update
instead of fussing about thenixpkgs
Git repository, finding a commit hash, and then getting it's sha256 hash. The approach in this merge request offers backwards compatibility with the Flake-skeptical and older systems viaflake-compat
.What needs testing: anything related to
darwin
. I don't have access to any of those types of machines.Opinionated fix: move to
nixpkgs-fmt
overnixfmt
. Like my gripes withprettier
,nixfmt
moves your code around too much to achieve its opinionated goals which causes unnecessary merge conflicts. String now needs one extra character? Better break all the lines apart and make a massive diff vs.nixpkgs-fmt
whose goals are better for maintainability than aesthetics. One could easily compare this to moving frompurescript-tidy
frompurty
(though sadly no config file).