Comment by Terr_

Yeah, fixup commits are the answer when it comes to responses to reviewers, who can easily see just what changed since the last time.

It's even better if there's some system that ensures authors remember to do the squashing, such as by prohibiting actions that would bring fixup commits into the main development branch.

Personally, I use `--fixup` even before making a PR, particularly if there's some work that I want to split into "refactoring prep" versus "the new feature."