PR statistics

[ccheungfb]

Problem:

Pullsbury gitboy currently takes the approach of an active prodding approach, but no means of reporting as far as I understand.

I'm hoping to get some statistics that can be used for a team, but also for an application/service health perspective. This is from a perspective of reporting and monitoring for quality of an app/service.

Stats I can currently think of:

Number of PR's open at the end of each day/week
Data on how long a PR is up on (can use this to measure min/max/average etc for week/month)
Number of comments/PR
Comment Response time (If a certain person commented asking for info, how long does it take to respond?)

Other ongoing ideas that would be useful:

How good are our reviews?

Currently working on a checklist for reviewers. It would be useful to understand what reviewer focusing on.
https://docs.google.com/document/d/1UqMTd8SwV_-ZOnvNWbn2cY-CAAvX84rOn67pMTxLYNM/edit?pli=1
Ex. With this info I can say, we've been focusing on repo guidelines, but not enough on OOD principles or reviewing test quality.

Since this is an open sourced project for you it probably this wouldn't go in

Change request severity

Based off review comments, would be good understand how sever the change request is. Is it a spelling mistake, or does it break functionality? Is it missing tests?
Ex. With this I can say provide feedback on how good X's initial PRs are. The amount of back and forth with reviewing is expensive.

How good are our commits?

As a reviewer, have a scale to determine how good you feel the PR is. If we're constantly committing stuff that is :( then it is a signal something is wrong.
Ex. For the past month, the repo has an end score of -6, which might mean the codebase health for the repo has decreased. Ideally the score should be positive

All the other ideas revolve around parsing comments and unfortunately will need more buy-in to get this working.