0

On my local repo I have the main working tree full of many unrelated changes.

I want to split my messy index into many distinct tidy linked worktrees, so I can isolate changes by concern and handle them without getting insane or rely on memory.

What's the minimal amount of steps to redistribute chunks among different worktrees?

Example

$ git status
 M .env
 M .gitignore
 M Dockerfile
 M README.md
?? untracked.file
?? worktree/

I would

  • use git stash save Concern001 -p, to interactively select the changes I want to put into Concern001
  • to then git stash apply Concern001 into my worktree
    • (created via git worktree add worktree/Concern001).\

But this approach fails in the case I want to move untracked changes to another worktree. Trying git stash save Concern001 -p -u

Can't use --patch and --include-untracked or --all at the same time

Moreover, the presence of the worktree/ directory among the untracked files makes the stashing error prone; although I can fix this adding /worktree to .gitignore.

For now I tried:

  • git add -p to select the changes to isolate
  • git commit -m 'Concern001' => commit-SHA
  • cd worktree/Concern001
  • git cherry-pick <commit-SHA>
  • git reset HEAD^ to get again the changes in the worktree index
  • cd ../.. (back to main worktree)
  • git reset HEAD^..... but this will restore the changes I just moved away; and if I instead use git reset HEAD^ --hard then I would loose all the other temporary changes from the index.

This approach look verbose, cumbersome and error prone.

Is there a sequence of git commands to select specific changes and move them between worktrees, without having to do too many intermediate operations❓</h3>

Ideally I'd use git stash but doesn't cover all the use cases.

4

0 回答 0