Someone's bot is creating multiple repeated issues for failing packages

Development
1

See:

https://github.com/NixOS/nixpkgs/issues/186871
https://github.com/NixOS/nixpkgs/issues/186872
https://github.com/NixOS/nixpkgs/issues/187043
https://github.com/NixOS/nixpkgs/issues/187045

It’s not helpful to flood the issues tracker with such things. We will be aware that something is broken, and it will be fixed eventually.

2 Likes
2

I assume Nixpkgs's current development workflow is not sustainable provides some context on why.

1 Like
3

Indeed, this is GitHub - samuela/nixpkgs-upkeep: Auto-updates and CI 🤖 for Nixpkgs!

Apologies for the duplicates in this case… I was off-the-grid for the past week and did not have the opportunity to triage. nixpkgs-upkeep has some delicate methods for detecting and preventing duplicate issues. I’ll have to debug further why they didn’t apply here. Currently my guess is:

  1. `python310Packages.nbconvert` build failure on x86_64-linux as of `a093c35b` · Issue #186871 · NixOS/nixpkgs · GitHub and `python310Packages.nbconvert` build failure on x86_64-linux as of `a093c35b` · Issue #186872 · NixOS/nixpkgs · GitHub were race condition collisions and both submitted posts at the same time. Same thing with `python310Packages.nbconvert` build failure on x86_64-linux as of `bdf1b57c` · Issue #187043 · NixOS/nixpkgs · GitHub and `python310Packages.nbconvert` build failure on x86_64-linux as of `bdf1b57c` · Issue #187045 · NixOS/nixpkgs · GitHub.
  2. The two latest issues were not flagged as duplicates of the first two issues since the nix store hashes of the dependencies had changed which caused the failure logs to differ.
4 Likes
4

Check out mitigate race conflicts between jobs when creating issues · samuela/nixpkgs-upkeep@d478e5c · GitHub which should fix issue 1 and ignore /nix/store/<hash>'s when looking at logs · samuela/nixpkgs-upkeep@d814770 · GitHub which should fix issue 2.

1 Like
5

Can you please stop automatically creating issues in nixpkgs and instead use one of your own repositories? They are not only sometimes duplicated with themselves but also the root cause is often duplicated and a minimum amount of triaging is missing. If the issue can be solved in less than 5 minutes there shouldn’t be an issue for it and instead a PR that fixes it straight away. Otherwise we eventually drown in untriaged issues.

4 Likes