How are you managing code reviews in the AI era?
We’ve noticed that most of our tickets are getting stuck at the code review stage, mainly because PRs have become so large that they take much longer to review!
I’m curious to hear how you’re handling this in your team!
Commit
b14ef7a
... with AI. And a shitload of tests.
Yup, tests are another issue we’re facing. That shitload of tests takes a crazy amount of time to review, and sometimes even the ticket assignee can’t properly review them. So we’re struggling to ensure that the test are asserting that product behaves as expected.
Honestly, there are 2 ways around this:
- throwing more agents at it to review (I don't recommend this)
- splitting up massive PRs into smaller PRs (this is what I recommend)
Generally speaking, I don't like huge PRs. Enforce breaking them up into PRs that are manageable in size. If they must go in together, stack the PRs: https://docs.github.com/en/pull-requests/how-tos/stacked-pull-requests

GitHub DocsStacked pull requests - GitHub DocsUse stacked pull requests to break large code changes into a chain of smaller, dependent pull requests you can review and merge independently.
commit b14ef7a