Changing local directory to be git's main instead of detached head?

[davetron5000]

I have workflow scripts that rely on git status locally knowing I'm on main, which is not the default case with jj (I'm on a detached head). I don't know what jj is actually doing under the covers, so am I safe to git checkout main, do what I need to do and then jj new to get back to working?

Basically, I want to "turn off" whatever jj is doing so I can run these automations as if I was still using Git, then "turn on" jj to keep using it.

[apoelstra]

You probably want to run jj new main which is equivalent to git checkout main.

[scott2000]

Assuming you don't have any changes in your working copy (@), it should be good to just do git checkout main.

Otherwise, it might be better to do jj new main first to get to an empty working copy in the right place, then git checkout main to tell Git to use the branch instead of detached HEAD.

[davetron5000]

Cool - is there some docs that explain what is happening under the covers that results in being a detached head state from git's point of view?

[scott2000]

I'm not sure if there's docs on this behavior, but basically whenever you switch to a new commit using jj in a colocated repo, it also updates the Git HEAD to point to first_parent(@) to ensure that Git sees the same state. But since jj doesn't have any concept of a current branch, it just points the Git HEAD to the commit ID itself, which is what Git calls a "detached HEAD".

When importing changes from the Git repo, jj doesn't care whether HEAD is detached or not, since it just evaluates it to a commit ID either way. So if you're already on a branch from git checkout, you'll remain on that branch until you switch to a different commit using jj I believe, at which point it'll go back to detached HEAD.