Hacker News Reader

Comment by vaylian | Hacker News Reader

One of the golden rules of git is: Don't rewrite public history (for example by rebasing and force-pushing already published commits). The article hints at immutable and mutable commits (https://isaaccorbrey.com/notes/jujutsu-megamerges-for-fun-an...) but I am unsure about how fundamental this protection is. I don't want to ruin my co-worker's day by accidentally rewriting public history.

View on HN

Don't rewrite shared public history

It's ok to force-push a branch that only you have worked on (and even in the case of others working on the same branch it can be fine as long as you communicate with them)

I agree. This is a valid exception.

Rewriting immutable commits is only "fundermental" if it's enforced upstream. You can always break the rule locally.

That said, jj will warn you if you try to edit an immutable commits, and you can configure what it considers immutable.

There's a configurable setting for which changes are marked as immutable. The default works perfectly for my workflow (pull-only from upstream, rewrite and push freely to my fork). other workflows may presumably need to tweak it

https://www.jj-vcs.dev/latest/config/#set-of-immutable-commi...

Jujutsu is immutable for public changes by default.

You can force changes with a ‘—ignore-inmutable’ flag.

More like a custom which sometimes has good reasons to be broken than a golden rule. A branch in Git is nothing but a pointer to a commit, anything more than that is social agreement that can differ in various contexts.

Reply on news.ycombinator.com