Question

I created a “feature” branch off master and worked for a long while. I then fetched the latest master branch commits and rebased my “feature” branch commits on top of it. Then I merged “feature” into master. However, I forgot about the merge and continued committing into the “feature” branch. I want those commits in a new branch now, so I would now like to create another branch, “feature_2”, that is based on the commits on the “feature” branch since the last merge into master. Any suggestions?

Answer 1

Considering that:

• feature already references commits since last merge into master
• branches are just pointers

x--x--x (master)
       \
        y--y--y (feature)

, you can simply:

• git checkout feature
• git checkout -b feature2
• git branch -f feature master
  (provided no commits were made on master since the feature merge)

x--x--x (master, feature)
       \
        y--y--y (feature2)

All the commits from master are no longer referenced by feature (which is reset to where master is), but are now accessible through feature2 (which is where feature is before being reset to master)

---

The OP Chip Castle adds:

I have 3 other team members who have had their branches merged into master since then, so I don’t want to merge the same commits I have already merged.
I was hoping to give a SHA range only for the commits I need from feature to a new branch. Is there any way to do that or a better way?

So the situation is:

x--x--x--y'--y' (master, updated after a fetch from other repo)
       \
        y'--y'--y--y (feature, with y' being already merged, 
                      and y being the commits I need)

Then you can simply rebase feature on top of master: identical commits will be ignored.