Question

I have a slightly outdated experimental branch, that was pushed aside due to time constraints. I need to pick it up and move it to the HEAD of the project.

Now the problem is that the patches in the branch are actually sort of big with huge amount of collisions, so I would like to break the commits in the branch into single file commits, so I can more easily apply them one by one.

Is there some automatic command, that would do this for me, or do I need to do this manually?

Answer 1

You can do it with:

git checkout -b experiment master^^^^
git rev-list master^^^..master | 
  while read rev; 
  do 
    git diff --name-only $rev^..$rev | 
      while read file; 
      do 
        git checkout $rev -- $file;
        git add $file; 
        git commit -C $rev;
      done; 
  done

please note the number of ^s in each statement

this assumes that you want to expand the range master^^^^..master and it is linear history or you don’t mind losing merges. It also assumes you want to keep the same commit message for the files affected each time.