pin nixpkgs registry to pkgs.path via nix.settings.registry

[MatthewCroughan]

Description of changes

This is a suggestion of something that would be needed in order to allow #189048 to function correctly. Otherwise the registry entry the user's nix will generate when faced with a /etc/nixos/flake.nix that does not contain an inputs attribute, will be ambigious and non-reproducible. We set the nix-channel appropriately in channel.nix for this same reason.

I provide two commits for doing this, and I'm not sure which approach is better.

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandbox = true set in nix.conf? (See Nix manual)
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 22.11 Release Notes (or backporting 22.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
  • (Release notes changes) Ran nixos/doc/manual/md-to-db.sh to update generated release notes
• Fits CONTRIBUTING.md.

[pennae]

this would be much better done in flake.nix by setting nix.registry.nixpkgs.flake = nixpkgs, and nixPath also needs set appropriately

[MatthewCroughan]

@pennae That doesn't solve anything. If you run nix shell nixpkgs#foo from a live CD, it will grab the latest nixpkgs master, unless this PR is merged. What you're talking about only takes effect after the generation you're suggesting, is activated.

[pennae]

ah, mistakenly thought this piece also goes into the installed system. that'll also obviate the nixPath change then since it'll already be set correctly. :)

[bbjubjub2494]

I support the idea. Does not work on my machine:

error: attribute 'outPath' missing

       at /nix/store/1ygz8a75ds7rwfpixmwwcr67pj0czxmh-nixpkgs-patched-a63021a/nixos/modules/services/misc/nix-daemon.nix:436:26:

          435|                   type = "path";
          436|                   path = config.flake.outPath;
             |                          ^
          437|                 } // filterAttrs
(use '--show-trace' to show detailed location information)

outputs typically doesn't return an outPath.

For the record I'm building a custom iso but I don't think that makes a difference.

[MatthewCroughan]

@lourkeur does the commit I just pushed, fix anything? This would have been much easier if #167947 wasn't reverted.

[bbjubjub2494]

@MatthewCroughan nope. Same error.

I'd love to help better but I'm out of my depth with the flakes magic. I would like this to be possible other than setting to to a path directly...

nix.registry.nixpkgs.to = {
  type = "path";
  path = pkgs.path;
};

[ck3d]

LGTM, but please squash the commits

[bbjubjub2494]

All good afaiac