Update bindings to be compatible with Nix 2.30, and package with Nix

[cottand]

Brings lib up-to-date so it is compatible with the new C API since Nix 2.24 (now in Nix 2.30!), which now uses the new typdef nix_value.

In order to do this, I also implemented a couple things on the side

• Replaced the dev-env with a proper nix build (you can still nix develop .)
• Fixed freeing some CStrings (still many more to go)

Because of the new API are a few non-backwards compatible changes:

• The Go API for registering PrimOps has changed
  • You are now able to return an error from Go, which will get correctly handled on the Nix side
  • The return value is now owned by Nix, rather than allocated during the Go callback. So instead of returning a *Value, you should now call SetInt(), SetString(), etc on a *Value that is provided to you (see C example or Go tests)
• InitPlugins is gone - I could not get this to work, it seems Nix is not providing it anymore even though it is in the C API header files (although it is not in the docs)
• GetSetting and SetSetting are often returning key error, even though the API is unchanged. I am not sure this is a problem in this library, but could not find much in the Nix changelog either.

I am not a CGO FFI expert, so there may be incorrect assumptions!

[cottand]

I do also plan to follow up and add the new Flake C API too to my fork, if this PR goes well I will also upstream that here

[marcusramberg]

I checked out your branch and tried building, but I'm seeing this test failure:

❯ nix build .                                                                                 gonix-env
error: builder for '/nix/store/39yvgsq4szy5xbkrpy6zypxrpdfwqap6-gonix.drv' failed with exit code 1;
       last 25 log lines:
       > Running phase: checkPhase
       > --- FAIL: TestExampleRegisterGlobalPrimOp (0.00s)
       > panic: runtime error: invalid memory address or nil pointer dereference [recovered]
       >       panic: runtime error: invalid memory address or nil pointer dereference
       > [signal SIGSEGV: segmentation violation code=0x1 addr=0x8 pc=0x51a6d0]
       >
       > goroutine 20 [running]:
       > testing.tRunner.func1.2({0x547740, 0x6a0c60})
       >       /nix/store/rn5bndqvz2y3ayc1j0fcchdf2hcbmx18-go-1.24.5/share/go/src/testing/testing.go:1734 +0x1ac
       > testing.tRunner.func1()
       >     /nix/store/rn5bndqvz2y3ayc1j0fcchdf2hcbmx18-go-1.24.5/share/go/src/testing/testing.go:1737 +0x334
       > panic({0x547740?, 0x6a0c60?})
       >       /nix/store/rn5bndqvz2y3ayc1j0fcchdf2hcbmx18-go-1.24.5/share/go/src/runtime/panic.go:792 +0x124
       > github.com/farcaller/gonix.(*Store).NewState.func2(...)
       >        /build/p1qk150r1175cv2fsmhy86zjv4cmy6r2-source/state.go:36
       > github.com/farcaller/gonix.(*Store).NewState(0x0, {0x0, 0x0, 0x4000075f28?})
       >       /build/p1qk150r1175cv2fsmhy86zjv4cmy6r2-source/state.go:36 +0x130
       > github.com/farcaller/gonix_test.TestExampleRegisterGlobalPrimOp(0x4000102e00)
       >       /build/p1qk150r1175cv2fsmhy86zjv4cmy6r2-source/primop_test.go:32 +0x100
       > testing.tRunner(0x4000102e00, 0x57df38)
       >       /nix/store/rn5bndqvz2y3ayc1j0fcchdf2hcbmx18-go-1.24.5/share/go/src/testing/testing.go:1792 +0xe4
       > created by testing.(*T).Run in goroutine 1
       >   /nix/store/rn5bndqvz2y3ayc1j0fcchdf2hcbmx18-go-1.24.5/share/go/src/testing/testing.go:1851 +0x374
       > FAIL github.com/farcaller/gonix      0.045s
       > FAIL
       For full logs, run 'nix log /nix/store/39yvgsq4szy5xbkrpy6zypxrpdfwqap6-gonix.drv'.

(this is on aarch64-linux if that matters)

[cottand]

thanks for checking! I had built on ARM darwin only, and on linux calling nix_store_open inside a Nix build (which is the case during tests) seems to be a problem

I have amended the test to print the error properly, and also added doCheck = false to the nix build, so that we don't try to run gonix tests during nix build

[marcusramberg]

Btw, seems 2.31 is going to be the next stable version, any reason we're not targetting that? NixOS/nixpkgs#440681

[marcusramberg]

This probably shouldn't have been included in the commit.

[cottand]

absolutely, my bad.

I also included it in the gitignore

[marcusramberg]

Btw, seems 2.31 is going to be the next stable version, any reason we're not targetting that? NixOS/nixpkgs#440681

Oh I see, it's not in 25.05. Maybe we could get it backported..

[cottand]

I can target 2.31 by using https://github.com/NixOS/nix as a flake input rather than nixpkgs.nix if you'd like, we'd be adding a new input to the flake is all.

It also doesn't seem like there are significant C API changes between .30 and .31