Git merge

Git Merge

Merging is Git's way of putting a forked history back together again. The git merge command lets you take the independent lines of development created by git branch and integrate them into a single branch.

Note that all of the commands presented below merge into the current branch. The current branch will be updated to reflect the merge, but the target branch will be completely unaffected. Again, this means that git merge is often used in conjunction with git checkout for selecting the current branch and git branch -d for deleting the obsolete target branch.

Usage

git merge <branch>

Merge the specified branch into the current branch. Git will determine the merge algorithm automatically (discussed below).

git merge --no-ff <branch>

Merge the specified branch into the current branch, but always generate a merge commit (even if it was a fast-forward merge). This is useful for documenting all merges that occur in your repository.

Discussion

Once you’ve finished developing a feature in an isolated branch, it's important to be able to get it back into the main code base. Depending on the structure of your repository, Git has several distinct algorithms to accomplish this: a fast-forward merge or a 3-way merge.

A fast-forward merge can occur when there is a linear path from the current branch tip to the target branch. Instead of “actually” merging the branches, all Git has to do to integrate the histories is move (i.e., “fast forward”) the current branch tip up to the target branch tip. This effectively combines the histories, since all of the commits reachable from the target branch are now available through the current one. For example, a fast forward merge of some-feature into master would look something like the following:

Git Tutorial: Fast-forward merge

However, a fast-forward merge is not possible if the branches have diverged. When there is not a linear path to the target branch, Git has no choice but to combine them via a 3-way merge. 3-way merges use a dedicated commit to tie together the two histories. The nomenclature comes from the fact that Git uses three commits to generate the merge commit: the two branch tips and their common ancestor.

Git Tutorial: Three way merge

While you can use either of these merge strategies, many developers like to use fast-forward merges (facilitated through rebasing) for small features or bug fixes, while reserving 3-way merges for the integration of longer-running features. In the latter case, the resulting merge commit serves as a symbolic joining of the two branches.

Resolving Conflicts

If the two branches you're trying to merge both changed the same part of the same file, Git won't be able to figure out which version to use. When such a situation occurs, it stops right before the merge commit so that you can resolve the conflicts manually.

The great part of Git's merging process is that it uses the familiar edit/stage/commit workflow to resolve merge conflicts. When you encounter a merge conflict, running the git status command shows you which files need to be resolved. For example, if both branches modified the same section of hello.py, you would see something like the following:

# On branch master
# Unmerged paths:
# (use "git add/rm ..." as appropriate to mark resolution)
#
# both modified: hello.py
#

Then, you can go in and fix up the merge to your liking. When you're ready to finish the merge, all you have to do is run git add on the conflicted file(s) to tell Git they're resolved. Then, you run a normal git commit to generate the merge commit. It’s the exact same process as committing an ordinary snapshot, which means it’s easy for normal developers to manage their own merges.

Note that merge conflicts will only occur in the event of a 3-way merge. It’s not possible to have conflicting changes in a fast-forward merge.

Example

Fast-Forward Merge

Our first example demonstrates a fast-forward merge. The code below creates a new branch, adds two commits to it, then integrates it into the main line with a fast-forward merge.

# Start a new feature
git checkout -b new-feature master
# Edit some files
git add <file>
git commit -m "Start a feature"
# Edit some files
git add <file>
git commit -m "Finish a feature"
# Merge in the new-feature branch
git checkout master
git merge new-feature
git branch -d new-feature

This is a common workflow for short-lived topic branches that are used more as an isolated development than an organizational tool for longer-running features.

Also note that Git should not complain about the git branch -d, since new-feature is now accessible from the master branch.

3-Way Merge

The next example is very similar, but requires a 3-way merge because master progresses while the feature is in-progress. This is a common scenario for large features or when several developers are working on a project simultaneously.

# Start a new feature
git checkout -b new-feature master
# Edit some files
git add <file>
git commit -m "Start a feature"
# Edit some files
git add <file>
git commit -m "Finish a feature"
# Develop the master branch
git checkout master
# Edit some files
git add <file>
git commit -m "Make some super-stable changes to master"
# Merge in the new-feature branch
git merge new-feature
git branch -d new-feature

Note that it’s impossible for Git to perform a fast-forward merge, as there is no way to move master up to new-feature without backtracking.

For most workflows, new-feature would be a much larger feature that took a long time to develop, which would be why new commits would appear on master in the meantime. If your feature branch was actually as small as the one in the above example, you would probably be better off rebasing it onto master and doing a fast-forward merge. This prevents superfluous merge commits from cluttering up the project history.

Ready to learn Git?

Try this interactive tutorial.

Get started now