ytmdesktop: build from source

[TomaSajt]

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• 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/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 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
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[cjshearer]

LGTM - glad to see building packages with yarn-berry got better over this past year.

[TomaSajt]

Well, I don't know what it was like before, but as you may have seen, we aren't getting the same hash on linux and darwin...

[cjshearer]

Could it be some platform dependent download? What if you specify env.ELECTRON_SKIP_BINARY_DOWNLOAD = "1";, like in this package.

It seems to be common for electron packages that use yarn.

[TomaSajt]

I'm 99% sure that won't help, that env var is only used when running electron's rebuild script

---

We could maybe just keep the dual hash like this, and once it gets cached by hydra, we can easily compare the two.

[cjshearer]

I don't think it's a big deal. If you want, you could see if env.CI = 1; affects the installation behavior in any platform-dependent way.

I recall there being a lot of things I had to disable when I first tried getting this to build on nix from source.

[flokli]

There now is a yarn-berry fetcher, see #399404.

[TomaSajt]

@yuyuyureka could you help with figuring out why the config hook is failing?

[yuyuyureka]

The fix for that problem is now included in #401631

Can I push a version that works on your branch?

has been addressed

[TomaSajt]

I see, you added the dependency that would have been installed with npx.

I think the better solution would be to just patch that script out.
Or, since it doesn't seem like any of them are required, just disabling all scripts would work fine too.

[yuyuyureka]

A short explanation for the changes I pushed:

The dependency fastify-socket.io@npm:5.1.0 runs a npx only-allow [...] in its postInstall script. This is evil, because only-allow is not specified as a dependency of fastify-socket.io, so it is not present in the cache and npx will go and download the latest version of only-allow, which is obviously not reproducible.

So I added a patch to the package.json and yarn.lock to add only-allow as a dependency, which means npx will use the existing installed version at that point.
We should consider making an upstream PR to add only-allow to the dependencies in https://github.com/ducktors/fastify-socket.io/.

Next, I ran into the issue that the only-allow script would not run, because it has a shebang with #!/usr/bin/env which does not exist in the sandbox. I split the "yarn install" in yarnBerryConfigHook in two parts, and added a patchShebangs in between, similar to how npmConfigHook does it.

I did not consider skipping the build scripts entirely, because I don't know if that impacts the functionality of the app now or in the future. This might be an option for you if you know the application well and can confirm that it works without them.

[TomaSajt]

Thanks for figuring this out :)

I think I will go with the YARN_ENABLE_SCRIPTS solution for now.
If you think the patchShebangs modification is still a change worth making, please do it in another PR.

[teutat3s]

nixpkgs-review result

Generated using nixpkgs-review.

Command: nixpkgs-review pr 397639

---

x86_64-linux

✅ 1 package built:

• ytmdesktop

[TomaSajt]

Currently broken on Darwin, don't merge yet.

[TomaSajt]

Whhyyyyy.....
It's NixOS/nix#12698

I opened #303307

[TomaSajt]

It builds fine on ofborg's darwin, because it doesn't have the sandbox enabled.
Will merge soon, if there are no more objections.