Great job on this, been meaning to do this but got sidetracked.

I think this merge was way too quick. We didn't go through a round of review on this. This introduces new Bash code against our policy. Revert in #456391, @infinisil please open it again, so that we can properly discuss it.

We should encourage quick merges, but yeah 4 hours is perhaps a bit too quick :P

But @wolfgangwalther, the policy states

New CI features need to be introduced in JavaScript, not Bash.

This is for all intents and purposes not a new CI feature. All the scripts in the PR exist already and are just extended and refactored. I don't mind writing new JS, but I do mind rewriting working bash in JS.

the policy states

New CI features need to be introduced in JavaScript, not Bash.

This is for all intents and purposes not a new CI feature. All the scripts in the PR exist already and are just extended and refactored. I don't mind writing new JS, but I do mind rewriting working bash in JS.

The discussion in #453985 (comment) explicitly mentioned such a case, so yes "significantly expanding scope of an existing bash script" is considered a feature. Now, I have not looked at the actual changes here at all, the timing alone was reason enough for a revert. But at first glance they looked like a significant review effort, which was another big reason against introducing new bash code given in #453985 (comment).

I certainly don't want to push rewriting the reviewer scripts in JavaScript onto you. I have that on my list, and will do it as part of a new feature I have in mind. But this might not play with your timeline.

Edit: To be clear - I suggest you re-open the existing PR right now, because I assume we can discuss details about the actual logic, the nix code etc., all without even looking at the bash code right now. This might not mean we immediately merge, but we will certainly be able to progress this topic further.