wasmtime: don't propagate binaries for C API & library

[nekowinston]

Reduces the closure size if just the library/C API is used.

According to nix-tree, running on ./result-dev:
before:
NAR Size: 142.75 MiB | Closure Size: 233.31 MiB | Added Size: 233.31 MiB
after:
NAR Size: 142.92 MiB | Closure Size: 183.94 MiB | Added Size: 183.94 MiB

I've also noticed that there's a preexisting function (moveToOutput) doing what's done manually; it just doesn't seem to be documented anywhere.

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.

[nekowinston]

Since I've contributed the C API and done a bit of continuous troubleshooting/optimizing on it, I'd be glad to be added as a maintainer if the current maintainers are happy with that.

[ereslibre]

I'd be glad to be added as a maintainer if the current maintainers are happy with that.

Absolutely! Feel free to add yourself in this PR!

[ereslibre]

Thanks for the contribution, it's great to have smaller closures!

[ereslibre]

nixpkgs-review result

Generated using nixpkgs-review.

Command: nixpkgs-review pr 438391
Commit: 6c867d22dee9c36a93a1874697da46702fc99059

---

x86_64-linux

✅ 2 packages built:

• wasmtime
• wasmtime.dev

[ereslibre]

Approved automatically following the successful run of nixpkgs-review.

[ereslibre]

@NixOS/nixpkgs-merge-bot merge

[nixpkgs-merge-bot]

@ereslibre merge not permitted (#305350):
R-Ryantm Maintainer merge: pr author is not r-ryantm
CommitterPR: pr author is not committer

[nekowinston]

Oh, I just realized that if we turned building the static library into an opt-in, we could get rid of another 117M:
NAR Size: 31.58 MiB | Closure Size: 72.6 MiB | Added Size: 72.6 MiB

The reason I'm obsessing about this a bit is that #394870 would ask people to pull wasmtime.dev into their tree-sitter and thus neovim closure. I'd wager that the smaller that is, the higher the chances people will be okay with that being the default.

--- a/pkgs/by-name/wa/wasmtime/package.nix
+++ b/pkgs/by-name/wa/wasmtime/package.nix
@@ -6,6 +6,7 @@
   cmake,
   versionCheckHook,
   nix-update-script,
+  enableStatic ? stdenv.hostPlatform.isStatic,
 }:
 rustPlatform.buildRustPackage (finalAttrs: {
   pname = "wasmtime";
@@ -63,6 +64,9 @@ rustPlatform.buildRustPackage (finalAttrs: {
       # https://github.com/rust-lang/cargo/issues/9661
       cp -r target/${cargoShortTarget}/release/build/wasmtime-c-api-impl-*/out/include $dev/include
     ''
+    + lib.optionalString (!enableStatic) ''
+      rm $dev/lib/*.a
+    ''
     + lib.optionalString stdenv.hostPlatform.isDarwin ''
       install_name_tool -id \
         $dev/lib/libwasmtime.dylib \

There's some prior art for removing the static libraries like that in tree-sitter itself:

nixpkgs/pkgs/development/tools/parsing/tree-sitter/default.nix

Lines 210 to 214 in da5efcc

	postInstall = ''
	PREFIX=$out make install
	${lib.optionalString (!enableShared) "rm $out/lib/*.so{,.*}"}
	${lib.optionalString (!enableStatic) "rm $out/lib/*.a"}
	''

[ereslibre]

@nekowinston thanks for the context. I think it would make sense to change that as another PR, and I'd be fine approving that. I didn't have the real need to optimize that much, but I think it makes sense if it helps for the neovim case you mentioned.

[nekowinston]

Alright, I'll open a follow-up PR for that 🙂

[samestep]

Hey @nekowinston! I hit an issue that was caused by this PR, and wanted to check whether it's intentional or just a bug: wasmtime no longer appears on the PATH for nix-shell or nix develop. You can test this by making a flake and putting wasmtime in the dev shell, or more simply by just running the following command in the Nixpkgs repo:

nix-shell --pure -I nixpkgs=$PWD -p wasmtime --run "wasmtime --version"

As of 3f405f0 and later, that command fails and prints wasmtime: command not found, whereas on its parent commits 4a9c6a9 and 6c867d2, it succeeds and prints wasmtime 33.0.0.

If it was intentional to remove wasmtime from the PATH for the wasmtime package, is there a workaround I can use downstream? Or if not, do you happen to know what the right fix is?