I am working on branch feature-a, and I need to make a change to feature-b on an unrelated part of the source code. I forgot (or was too lazy) to make a secondary work tree, and edited some files that should be changed in the feature-b branch.

I could now make a worktree and rsync my uncommitted changes to the worktree. Or I could make my changes on the current branch and then git-cherry-pick/git-rebase them into the right place.

But really I just want to be able to make a commit on feature-b that reflects my edits, without actually changing the branch, switching out the contents of the main working tree, etc.

Is this possible with the Git CLI?

CodePudding user response：

Yes, that is possible. But you will have to use git low-level commands (a.k.a. git plumbing). The basic tasks would be:

• create new objects for the changed files (git hash-object -w)
• create new trees for the directories containing the changed files (git write-tree)
• create a new commit object referencing the tree for the root directory and the appropriate parent commit (git commit-tree)
• move the branch to that new commit
• push the branch

Note that for the last two steps you can use high-level git-commands, the first three are the tricky ones.

The example in chapter 10.2 does pretty much what you want.