compilers/ghc/common-hadrian: expand flags for structuredAttrs

[SFrijters]

Otherwise we lose all flags apart from the first.

This results in a __structuredAttrs = true; build trying to find xelatex while it is not available.

Things done

• Built on platform:
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• Tested, as applicable:
  • NixOS tests in nixos/tests.
  • Package tests at passthru.tests.
  • Tests in lib/tests or pkgs/test for functions and "core" functionality.
• Ran nixpkgs-review on this PR. See nixpkgs-review usage.
• Tested basic functionality of all binary files, usually in ./result/bin/.
• Nixpkgs Release Notes
  • Package update: when the change is major or breaking.
• NixOS Release Notes
  • Module addition: when adding a new NixOS module.
  • Module update: when the change is significant.
• Fits CONTRIBUTING.md, pkgs/README.md, maintainers/README.md and other READMEs.

---

Add a 👍 reaction to pull requests you find important.

[sternenseemann]

Commit message should probably read haskell.compiler: … or something if you mention hadrianFlags it should be clear enough what it refers to.

[SFrijters]

I based the commit message prefix on a previous commit that touched the same file, but I've updated this one to something better.

[sternenseemann]

We need those to be space split.

Suggested change

	hadrian "''${hadrianFlags[@]}" "''${hadrianFlagsArray[@]}" binary-dist-dir
	hadrian ''${hadrianFlags[@]} "''${hadrianFlagsArray[@]}" binary-dist-dir

[SFrijters]

Done, Stage1 still successfully starts building for __structuredAttrs.

[wolfgangwalther]

One of the main selling points of structuredAttrs is the "native" support for spaces - so this change seems very much backwards to me.

Instead, we should add the quotes back and then enable __structuredAttrs = true for this file. This will give us the correct behavior consistently.

[SFrijters]

That would be my preferred solution also, but then I think we should also revisit #470435 for consistency, where setting __structuredAttrs = true was not welcomed.

[SFrijters]

Also, we'll run into the same discussion for generic-builder.nix in haskell-modules.

[wolfgangwalther]

That would be my preferred solution also, but then I think we should also revisit #470435 for consistency, where setting __structuredAttrs = true was not welcomed.

The difference in that PR was, that the code would behave correctly both with and without structuredAttrs. The "consistent" pattern here is: Only add _structuredAttrs = true when it actually makes a difference. It does here, because we can't easily treat both cases. Ofc it's possible, but we'd have to add the same code as in many of the setup-hooks with concatTo etc. - but that'd be overkill for here, where we actually control the derivation. Thus, __structuredAttrs = true is the preferred solution here, from my perspective.

[wolfgangwalther]

Also, we'll run into the same discussion for generic-builder.nix in haskell-modules.

We will, but since the builder is used for many more different packages, it might make sense to just implement it in a more robust way, that supports both cases, similar to the setup hooks I mentioned earlier.

[SFrijters]

To be replaced by #475546 .