A more automated way to run PR reviews

Hi! I’m just wondering if there’s anything in the community already that’s more automated than manually finding PRs to run nixpkgs-review on. I’ve had my first PR merged, and I saw how much time/effort it takes to review PRs.

Like an example system I’m envisioning is some simple dashboard (web or cli) that lists all the PRs which you can queue to review across machines (which gets run through ssh or something), then you can inspect the results and post them as a comment if they seem about right.

Doing it across machines is nice too, e.g. I’m planning to buy a RPi 5 to play around with aarch64 programs and possibly review nixpkgs PRs in aarch64.

Are there any other solutions already out there, for making reviewing easier? Just want to ask around before I try to reinvent the wheel.

Bot running nixpkgs-review on most nixpkgs PRs aka GitHub - rmcgibbo/r-rmcgibbo is the main ~bot here that I’m aware of.

Can’t comment on whether there’s anything like a dashboard since I havent run it or read through the source.

You only really need to run it when there could be regressions in other packages. This mostly applies to libraries because most leaf packages don’t qualify for this. Also passthru.tests can be used to build some key packages which is useful when targeting staging as building everything there is a bit wasteful.

1 Like