pnpm.fetchDeps: multiple workspaces; better pnpmInstallFlags

[pyrox0]

Adds support for installing multiple PNPM workspace packages' dependencies with the pnpmWorkspaces variable, and adds pnpmInstallFlags to other parts of PNPM configuration. Also adds appropriate documentation.

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/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 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.

[pyrox0]

Forgot to push the commits with the related package changes. I expect a good ofborg eval, can't check locally atm sorry.

[gepbird]

Ofborg failed because you forgot to change pnpmWorkspace in pnpmDeps.

Also would you mind splitting up your first commit? It contains many formatting changes that makes the diff a bit hard to read.

Edit: eval fails for some other reason too

[gepbird]

Aside from the those and some minor formatting issues that we don't have time for, this LGTM! Tested the 2 affected packages on x86_64-linux. I'm going to sleep and may not be available before this should be merged, I'd appreciate if someone could run a nixpkgs-review on this.

[pyrox0]

Ofborg failed because you forgot to change pnpmWorkspace in pnpmDeps.

Also would you mind splitting up your first commit? It contains many formatting changes that makes the diff a bit hard to read.

Edit: eval fails for some other reason too

The formatting changes come from the throwIf call being added, should I do a commit with just that and then a reformat commit after?

[emilazy]

My wisdom is to do assert lib.throwIf … true;, which formats nicely and is a somewhat pleasing “statement”. Also good for asserts.

[pyrox0]

My wisdom is to do assert lib.throwIf … true;, which formats nicely and is a somewhat pleasing “statement”. Also good for asserts.

Fixed, a sanity check would be nice though just to double check the code actually makes sense. Evals properly on my machine, though.

[gepbird]

Thanks!

Result of nixpkgs-review pr 350751 run on x86_64-linux 1

1 package built:

• nixpkgs-manual

[gepbird]

Sorry for the early approval, packages fail for me even though ofborg and nixpkgs-review passed.
For example when trying to build astro-language-server the assert triggers.

[gepbird]

If we are talking about formatting here are my nits I held back due to the time pressure. We have ~10 hours to merge this with the pnpmWorkspace deprecation.

[Scrumplex]

If we are talking about formatting here are my nits I held back due to the time pressure

I am sure we will get this merged by then. There are three committers involved/pinged in this thread ^^

[Scrumplex]

LGTM

[gepbird]

Changes LGTM! astro-language-server, bash-language-server and another pnpm package like heroic builds.

I started nixpkgs-review (on a different machine) and I'm getting 52 rebuilds compared to the previous 1 rebuild.

[doronbehar]

Looks good overall, small comments left. Haven't read other comments in the thread.

[gepbird]

nixpkgs-review looks good, I won't wait out the electron build.

[gepbird]

Throws an eval time error when pnpmWorkspace is both in a package derivation and in pnpm.fetchDeps, and a build time error when pnpmWorkspace is only in the derivation, as expected. Those 3 packages build, LGTM.

[GetPsyched]

This should have been flagged in open PRs that use pnpmWorkspace; #340035 was just merged and this will likely break in master. Raised #352830 to fix it.

[Scrumplex]

This should have been flagged in open PRs that use pnpmWorkspace; #340035 was just merged and this will likely break in master. Raised #352830 to fix it.

We don't really have tooling to find PRs that utilize pnpm.fetchDeps/pnpm.configHook currently. We tried our best previously with a change to pnpm store hashes but even missed a few PRs there

[gepbird]

We really need a tool to search PR diffs. GitHub seems to only search in comments:

Also by running nixpkgs-review before merging, these issues could be avoided. However I understand committers have to deal with many PRs and don't have time for this.

[doronbehar]

As for the surprising hash change you experienced, I just wanted to report that I checked:

nix build -Lf. {astro,bash}-language-server.pnpmDeps --rebuild --keep-going

And they didn't report a change of hash.. Not sure why that happened to you. It could be that in a manner unrelated to the workspace change that package's pnpmDeps is not very reproducible.

[gepbird]

@doronbehar pnpmDeps hashes were changed for most of the packages, see #349093 (comment)