Nix flake

[Eveeifyeve]

Description

• Followup from pr Nix flake #283 to take over from @nicholaschiasson comment
• Adds nix flake
• Changed nix flake to use flake parts over utils and flake compact for old nix compatibility
• Adds back the cargo lock file which is needed for nix run package and recommended by rust team.

Testing

N/A since there is no code change.
I simply ran nix run locally to test that it is working.

Please follow our Coding Guidelines

[Eveeifyeve]

I have one more thing to do for this pr and that is adding an github action to update the flake.lock over time.

[Eveeifyeve]

Finished. Note this will open a PR every time this updates.

[kralverde]

I'm not a nix user but I'm seeing a lot of relative paths in the script, is there away to anchor them to the script's directory?

Also I'm not a big fan of the workflow (as a concept) why do you need periodic automatic updates? Can you just add a contrib/script to re-generate the file that's ran as needed?

[Eveeifyeve]

I'm not a nix user but I'm seeing a lot of relative paths in the script, is there away to anchor them to the script's directory?

No sadly since it requires the same versions for the flake compact and paths needed for package is required for version and name and toolchain

Also I'm not a big fan of the workflow (as a concept) why do you need periodic automatic updates? Can you just add a contrib/script to re-generate the workflow that's ran as needed?

Because of dependancy changes from nixpkgs and rust-overlay I wanted to do this so you would have developers worrying about updates to dependancy updates with flakes everything is locked to what is specified in flake.lock like cargo lock but it doesn’t update until you do nix flake update.

[nicholaschiasson]

Isn't it a better idea to only ever intentionally update the flake lock file and manually check that into version control? I would think that having it automatically updated on a cron workflow could introduce breaking dependencies accidentally sometimes, and not to mention maintainers using the flake locally would be forced to move to a new derivation much more often, which is perhaps not a huge problem but a little inconvenient imo. Is this best practice for flakes right now?