When coding, I usually commit in small steps to keep the history clear. But sometimes, after finishing a whole branch and preparing to open a pull request, I realize there is a small error or need to add code to a previous commit. If I keep adding that change to the latest commit, the history will be fragmented: a feature/bugfix that should be encapsulated in one commit is now scattered across many commits.
My favorite solution: git rebase -i to edit old commits and consolidate changes back to their rightful place.
Step-by-Step Guide
1. Identify the commit you want to edit
Suppose you want to edit a commit that is among the n most recent commits. Run the command:
git rebase -i HEAD~n
Replace n with the number of the commit you want to review (e.g., HEAD~5 to open the 5 most recent commits).
2. Select the commit to edit
The editor opens and lists the commits. By default, each commit is marked with pick.
- Change
picktoediton the commit you want to edit (onvim, pressito enter insert mode, then make the change). - Save and exit the editor (on
vim, pressEsc, then type:wqand pressEnter).
3. Add changes to the old commit
Git will stop at that commit so you can edit the code.
- Edit the file as desired
- Stage changes (for example:
git add .) - Then amend the commit:
git commit --amend --no-edit
The --no-edit flag helps keep the old commit message intact, or you can use -m "new message" to change it.
4. Continue rebase
After editing the commit, run: git rebase --continue
Git will reapply the commits after the one you just edited. If there are no conflicts, you’re done, but if there are, you resolve and continue with git rebase –continue.
Conclusion
With this workflow, you can:
Keep your commit history neat and clear.
Group related changes into the right commit.
Make it easier for reviewers (and yourself in the future) to read the code.
Next time you need to edit an old commit, try git rebase -i. It takes a little effort at first, but it will make your Git history much “cleaner”.