Store StructuredAttrs directly in Derivation

[Ericson2314]

Instead of parsing a structured attrs at some later point, we parsed it right away when parsing the A-Term format, and likewise serialize it to __json = <JSON dump> when serializing a derivation to A-Term.

The JSON format can directly contain the JSON structured attrs without so encoding it, so we just do that.

Motivation

Depends on #13273

Context

---

Add 👍 to pull requests you find important.

The Nix maintainer team uses a GitHub project board to schedule and track reviews.

[Ericson2314]

Thanks to talking to @puckipedia, I became satisfied that the other way of changing the abstract syntax (pair of env and optional structured attrs) wasn't "too lax" because there already was a way within the language to mix arbitrary env vars with structured attrs.

This PR is now updated to work that way, and is passing all the tests, but let's look at #13273 first.

[dpulls]

🎉 All dependencies have been resolved !

[Ericson2314]

Notes from today's Nix team meeting:

• What about perf of parsing the structured attrs eagerly?

  • @Mic92 can check, perhaps

• What about derivations that no longer parse during GC?

  This branch in

  nix/src/libstore/gc.cc

  Lines 754 to 760 in 20226c8

  	if (gcKeepDerivations && path->isDerivation()) {
  	for (auto & [name, maybeOutPath] : queryPartialDerivationOutputMap(*path))
  	if (maybeOutPath &&
  	isValidPath(*maybeOutPath) &&
  	queryPathInfo(*maybeOutPath)->deriver == *path)
  	enqueue(*maybeOutPath);
  	}

  does parse derivations, to see if they point to outputs we should keep.

  We should recover if we fail to parse a derivation, and have a test for this. We also need to decide what we want to do in that case: delete or keep the derivation that doesn't parse?

• @edolstra what are the use-cases / ultimate purpose you have for doing this?

  @Ericson2314: Sure

  • Better derivation JSON format (or other new formats):

    That is useful for A very restricted recursive nix socket in the sandbox #8602 and dynamic derivations

  • Maybe better on-disk format too

    For input-addressed outputs, ATerm has to stick around forever at least as part of the path calculation. But for the newer experimental derivations types we have the opportunity to do a clean break.

[tomberek]

Needs discussion to address potential perf and laziness implications.

[nixos-discourse]

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/2025-05-28-nix-team-meeting-minutes-229/65205/1

[nixos-discourse]

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/2025-05-04-nix-team-meeting-minutes-230/65206/1