I keep finding git behaviours the same way: set up the case, run the command, then check the state I actually cared about instead of trusting the success message. Guides pair rerere.enabled with rerere.autoupdate when the second one removes the checkpoint the first one earns you. GitHub reports mergeable while the reviewer is still reading stale bytes. --autosquash prints Successfully rebased and updated and leaves a fixup! commit sitting in history.
Copy the config. The reasoning is why you would leave one of them off.
This started as a comment on Sylwia Laskowska's git list. She asked me to turn it into a post. The first four items are from ordinary rebasing. The rest I measured this week on a stacked-PR repo.
1. rerere does not learn from git add alone
git config rerere.enabled true
