workflows/eval: fix maintainer requests

[infinisil]

Fixes the problems after #366046, including:

• Avoiding requests from PR authors (therefore superseding eval/tag: remove pull request author from reviewer list #370186)
• Avoiding requests from people that aren't repo collaborators (which happens if you didn't accept the automated maintainer team invite)
• Avoiding requests from people that already left a review

This reuses part of a script that deals with requests from code owners introduced in #336261

Best reviewed commit-by-commit

Things done

• Tested codeowner reviews: Test code owner reviews Infinisil-s-Test-Organization/nixpkgs#39 (comment)
• Tested maintainer reviews: Test maintainer reviews Infinisil-s-Test-Organization/nixpkgs#38 (comment)
• Tested that org membership is not enough to be requested as reviewer: Test codeowner review for just org members Infinisil-s-Test-Organization/nixpkgs#40 (comment)

---

Add a 👍 reaction to pull requests you find important.

[SigmaSquadron]

Looks good to me! I left a couple of comments that can just be marked as resolved, they're questions and praise; not actionable issues. I can't find anything wrong with this.

[khaneliman]

LGTM, thanks

[GGG-KILLER]

@infinisil is the following error from this PR?
It doesn't seem to fail the build, but something we should take into account most likely:

This seems to arise from users not being contributors to the repo, however I think tit might be nice to omit the output as it might confuse people analyzing the logs: https://github.com/Infinisil-s-Test-Organization/nixpkgs/actions/runs/12592351001/job/35096822060#step:6:34

{"message":"Not Found","documentation_url":"https://docs.github.com/rest/collaborators/collaborators#check-if-a-user-is-a-repository-collaborator","status":"404"}gh: Not Found (HTTP 404)

[infinisil]

@GGG-KILLER I think it's better to show those errors, because it could also happen that it fails for a reason other than the user not being a collaborator, e.g. if GitHub is down. Without the error message we couldn't distinguish between the causes.

[infinisil]

I agree that it's not ideal though, I think this could be improved in a future PR.

[GGG-KILLER]

Code looks good and tests look fine as well. Thanks for quickly implementing a fix for this!

[nix-backports]

Backport failed for release-24.11, because it was unable to cherry-pick the commit(s).

Please cherry-pick the changes locally and resolve any conflicts.

git fetch origin release-24.11
git worktree add -d .worktree/backport-370456-to-release-24.11 origin/release-24.11
cd .worktree/backport-370456-to-release-24.11
git switch --create backport-370456-to-release-24.11
git cherry-pick -x 0371b7fb4b8d3f69f6cf7f1a7192a3c55b362006 0ebab0bccadbd38d9bc0aae3e53e9493afa9ec0b ab248be504bc38a3ea0f7409d31109b45e619224 077007a65827b7c8127458adf2c3661b91fc5545

[wolfgangwalther]

Backport is included in #370709.

[emilazy]

Seems to still be having issues: https://github.com/NixOS/nixpkgs/actions/runs/12605350098/job/35134091088?pr=370749

[infinisil]

@emilazy That's weird, by manually using the review request UI to check individually, I figured out that among the maintainers that would've been requested for that PR, only the user @pleshevskiy can't be requested for a review, even though they're a repo collaborator as part of the Nixpkgs maintainers team. The account looks "archived", but GitHub provides no such feature, and https://api.github.com/users/pleshevskiy doesn't show anything odd 🤔. I've also messaged pleshevskiy on Matrix to ask for help.

[JohnRTitor]

Can we just reuse the logic Ofborg uses for this?

[Moraxyc]

FYI: https://github.com/NixOS/nixpkgs/actions/runs/12609376751/job/35142923617?pr=361515

gh: Review cannot be requested from pull request author. (HTTP 422)
{"message":"Review cannot be requested from pull request author.","documentation_url":"[https://docs.github.com/rest/pulls/review-](https://docs.github.com/rest/pulls/review-requests#request-reviewers-for-a-pull-request)

[wolfgangwalther]

FYI: https://github.com/NixOS/nixpkgs/actions/runs/12609376751/job/35142923617?pr=361515

This should be fixed by #370824.

[wolfgangwalther]

I figured out that among the maintainers that would've been requested for that PR, only the user @pleshevskiy can't be requested for a review

ofborg seems to be able to figure that out, see the last PR for sonic-server in #320496.

r-ryantm mentioned pleshevskiy as a maintainer in the comment, but ofborg didn't request review from them.

[wolfgangwalther]

ofborg seems to be able to figure that out

Or maybe not, seems like ofborg just requests reviews 1-by-1 and is OK with failing some of them?

https://github.com/NixOS/ofborg/blob/1ef14d9c5b570b110d045795b4e80a37db2e1f0a/ofborg/src/tasks/eval/nixpkgs.rs#L553-L577

[wolfgangwalther]

Or maybe not, seems like ofborg just requests reviews 1-by-1 and is OK with failing some of them?

Created #370857 to do the 1-by-1 request.

[wolfgangwalther]

@emilazy That's weird, by manually using the review request UI to check individually, I figured out that among the maintainers that would've been requested for that PR, only the user @pleshevskiy can't be requested for a review, even though they're a repo collaborator as part of the Nixpkgs maintainers team. The account looks "archived", but GitHub provides no such feature, and https://api.github.com/users/pleshevskiy doesn't show anything odd 🤔. I've also messaged pleshevskiy on Matrix to ask for help.

Your assumption was correct, see the same PR with the fix applied:

https://github.com/NixOS/nixpkgs/actions/runs/12613829223/job/35152198458?pr=370749

It only fails for pleshevskiy.

(It also shows we need to apply a limit - we requested review from all 30 potential reviewers there)