Make getExe (writeScript x y) work

[roberth]

Description of changes

Broken, because path is missing:

nix-repl> lib.getExe (writeScript "hi" "hello")
"/nix/store/q8ybdl7z2v0sz0r6y9pkxb1bjk5xf0zd-hi/bin/hi"

After PR:

nix-repl> lib.getExe (writeScript "hi" "hello")
"/nix/store/q8ybdl7z2v0sz0r6y9pkxb1bjk5xf0zd-hi"

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/)
• 22.05 Release Notes (or backporting 21.11 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.

[Artturin]

this wouldn't match the behavior of nix run.

what if we instead check if meta.mainProgram is an absolute path and use that if so

that'll need a change in
https://github.com/NixOS/nix/blob/b1223e1b621ed4ad11562f0ef65c65d1c78c5e4b/src/nix/app.cc#L93-L112

opinions?

[roberth]

So it looks like nix run would need to change in any case.

what if we instead check if meta.mainProgram is an absolute path and use that if so

I think there's an unwritten rule that meta should not contain store references.
Also the default executable location is normal data that we use in normal expressions, not metadata.

opinions?

Another way to approach this, is to make writeScript return a writeScriptBin package with outPath set to the main executable. Single-file store paths are a forward compatibility mistake anyway, so maybe we should be thinking in this direction? Then nix run could check whether outPath is a store path with subpath, and if so, run that.

[roberth]

It might be preferable for writeScript to return a writeScriptBin x y + "/bin/${x}".
Unfortunately overrideAttrs doesn't seem capable of resetting outPath based on outPath, so that overrideAttrs would keep working as is. We could either add finalDerivationOutputs next to finalPackage, but my preferred option would be to factor out a mkPackage function that would allows us to construct the right behavior without much of the legacy and without the complexity of the finalPackage solution. I'll save that for when we have RFC 92 Dynamic Derivations though.