cli: include untracked remote branches in default immutable_heads()

[yuja]

I used to use "remote_branches() & ~mine()" to exclude "their" branches from the default log, and I don't think that's uncommon requirement. Suppose untracked branches are usually read-only, it's probably okay to make them immutable by default.

Checklist

If applicable:

• I have updated CHANGELOG.md
• I have updated the documentation (README.md, docs/, demos/)
• I have updated the config schema (cli/src/config-schema.json)
• I have added tests to cover my changes

[ilyagr]

I think this is a good idea. Nice!

It seems like this will affect the default log revset automatically? See my other comment. Ultimately, I think the new behavior would be quite sensible.

If so, the worst I imagine happening is that people will be confused by the change to jj log default output, and maybe we should document in more places that you need to jj track branches for them to show up in the default jj log.

There is still a chance there is some other problem neither of us sees, but I think this would be nice to test out on anyone using the master branch.

Still, let's wait for a day before merging, in case others have comments (for example, if anyone wants us to wait longer).

[ilyagr]

Also, I'll cc the discussion on local(), #3529 . I haven't been following that conversation carefully, but on one hand it is discussing similar issues, and on the other hand if we do come up with some local() revset it might be appropriate to add something based on ~local() to the immutable heads.

[scott2000]

I'm definitely in favor of this change. The main motivation for my PR for adding untracked_remote_branches() was so that I could add this to my immutable_heads() (mainly for the way it cleans up the log), and I imagine it would be a reasonable default for most people.

I also think it makes sense semantically for these commits to be immutable. Rewriting commits on untracked branches generally doesn't seem like a good idea since the remote branch will remain at the old commit even after it is rewritten, which probably isn't what a user would intend.

[martinvonz]

Good idea