Post #2695911
2026-04-17 18:20 UTC
@isaiah@mastodon.social I usually switch between branches within one directory, but sometimes that is impractical. For example, when I was doing major work on QEMU, I’d have one checkout for the big feature and another for smaller things. A switch would usually trigger a significant rebuild, which took an annoyingly long time.
For me, branches stay around for as long as they are unique. If its HEAD is just a commit that’s an ancestor of another branch, it can be deleted or replaced by a tag.
Replies (1)
-
@pmdj@mstdn.social 2026-04-17 18:26
@isaiah@mastodon.social Overall, I have no hard rules, my handling of this varies pragmatically from project to project. One consistent habit is rebasing on the main branch very regularly. I hate dealing with merge conflicts, and I hate nonlinear histories when tracing regressions. So even in projects with a policy to use merge commits, I will try to make sure the “loops” created by my merges will be as short as possible. I don’t use merge commits at all if I get to choose.