Collaborating with a remote repository


Teaching: 25 min
Exercises: 10 min
  • How do I update my local repository with changes from the remote?

  • How can I collaborate using Git?

  • Understand how to pull changes from remote repository

  • Understand how to resolve merge conflicts

Pulling changes from a remote repository

Having a remote repository means we can share it and collaborate with others (or even just continue to work alone but from multiple locations). We’ve seen how to clone the whole repo, so next we’ll look at how to update our local repo with just the latest changes on the remote.

We were in the laptop_paper directory at the end of the last episode, having pushed one commit to the remote. Let’s now change directory to the other repository paper, and git pull the commit from the remote.

$ cd ../paper
$ git pull origin master

We can now view the contents of and check the log to confirm we have the latest commit from the remote:

$ git log -2

Still in the paper directory, let’s add a figures section to, commit the file and push these changes to GitHub:

$ nano		# Add figures section
$ git add
$ git commit -m "Add figures"
$ git push

Now let’s change directory to our other repository and fetch the commits from our remote repository,

$ cd ../laptop_paper		# Switch to the other directory
$ git fetch

git fetch doesn’t change any of the local branches, it just gets information about what commits are on the remote branches.

We can visualise the remote branches in the same way as we did for local branches, so let’s draw a network graph before going any further:

git log --graph --all --decorate --oneline
* 7c239c3 (origin/master, origin/HEAD) Add figures
* 0cc2a2d (HEAD -> master) Discuss results
* 3011ee0 Describe methodology
*   6420699 Merge branch 'simulations'
| * 7138785 (origin/simulations) Add simulations
| * e695fa8 Change title and add coauthor
* | e950911 Include aircraft in title
* 0b28b0a Explain motivation for research
* 7cacba8 Cite previous work in introduction
* 56781f4 Cite PCASP paper
* 5033467 Start the introduction
* e08262e Add title and author

As expected, we see that the origin/master branch is ahead of our local master branch by one commit — note that the history hasn’t diverged, rather our local branch is missing the most recent commit on origin/master.

We can now see what the differences are by doing,

$ git diff origin/master

which compares our master branch with the origin/master branch which is the name of the master branch in origin which is the alias for our cloned repository, the one on GitHub.

We can then merge these changes into our current repository, but given the history hasn’t diverged, we don’t get a merge commit — instead we get a fast-forward merge.

$ git merge origin/master
Updating 0cc2a2d..7c239c3
Fast-forward | 4 ++++
 1 file changed, 4 insertions(+)

If we look at the network graph again, all that has changed is that master now points to the same commit as origin/master.

git log --graph --all --decorate --oneline -4
* 7c239c3 (HEAD -> master, origin/master, origin/HEAD) Add figures
* 0cc2a2d Discuss results
* 3011ee0 Describe methodology
*   6420699 Merge branch 'simulations'

We can inspect the file to confirm that we have our changes.

$ cat

So we have now used two slightly different methods to get the latest changes from the remote repo. You may already have guessed that git pull is a shorthand for git fetch followed by git merge.

Fetch vs pull

If git pull is a shortcut for git fetch followed by git merge then, why would you ever want to do these steps separately?

Well, depending on what the commits on the remote branch contain, you might want to abandon your local commits before merging (e.g. your local commits duplicate the changes on the remote), rebase your local branch to avoid a merge commit, or something else.

Fetching first lets you inspect the changes before deciding what you want to do with them.

Let’s write the conclusions:

$ nano		# Write Conclusions
$ git add
$ git commit -m "Write Conclusions"
$ git push origin master
$ cd ../paper			# Switch back to the paper directory
$ git pull origin master	# Get changes from remote repository

This is the same scenario as before, so we get another fast-forward merge.

We can check that we have our changes:

$ cat
$ git log

Conflicts and how to resolve them

Let’s continue to pretend that our two local repositories are hosted on two different machines. You should still be in the original paper folder. Add an affiliation for each author. Then push these changes to our remote repository:

$ nano		# Add author affiliations
$ git add
$ git commit -m "Add author affiliations"
$ git push origin master

Now let us suppose, at a later date, we use our other repository (on the laptop) and we want to change the order of the authors.

The remote branch origin/master is now ahead of our local master branch on the laptop, because we haven’t yet updated our local branch using git pull.

$ cd ../laptop_paper		# Switch directory to other copy of our repository
$ nano		# Change order of the authors
$ git add
$ git commit -m "Change the first author"
$ git push origin master
 ! [rejected]	     master -> master (fetch first)
error: failed to push some refs to '<USERNAME>/paper.git'
hint: Updates were rejected because the remote contains work that you do
hint: not have locally. This is usually caused by another repository pushing
hint: to the same ref. You may want to first integrate the remote changes
hint: (e.g., 'git pull ...') before pushing again.
hint: See the 'Note about fast-forwards' in 'git push --help' for details.

Our push fails, as we’ve not yet pulled down our changes from our remote repository. Before pushing we should always pull, so let’s do that…

$ git pull origin master

and we get:

CONFLICT (content): Merge conflict in
Automatic merge failed; fix conflicts and then commit the result.

As we saw earlier, with the fetch and merge, git pull pulls down changes from the repository and tries to merge them. It does this on a file-by-file basis, merging files line by line. We get a conflict if a file has changes that affect the same lines and those changes can’t be seamlessly merged. We had this situation before in the branching episode when we merged a feature branch into master. If we look at the status,

$ git status

we can see that our file is listed as Unmerged and if we look at, we see something like:

<<<<<<< HEAD
G Capes, J Smith
J Smith, G Capes
>>>>>>> 1b55fe7f23a6411f99bf573bfb287937ecb647fc

The mark-up shows us the parts of the file causing the conflict and the versions they come from. We now need to manually edit the file to resolve the conflict. Just like we did when we had to deal with the conflict when we were merging the branches.

We edit the file. Then commit our changes. Now, if we push

$ nano		# Edit file to resolve merge conflict
$ git add		# Stage the file
$ git commit			# Commit to mark the conflict as resolved
$ git push origin master

… all goes well. If we now go to GitHub and click on the “Overview” tab we can see where our repository diverged and came together again.

This is where version control proves itself better than DropBox or GoogleDrive, this ability to merge text files line-by-line and highlight the conflicts between them, so no work is ever lost.

We’ll finish by pulling these changes into other copy of the repo, so both copies are up to date:

$ cd ../paper			# Switch to 'paper' directory
$ git pull origin master	# Merge remote branch into local

Collaborating on a remote repository

In this exercise you should work with a partner or a group of three. One of you should give access to your remote repository on GitHub to the others (by selecting Settings tab -> Access -> Collaborators). The invited person should then check their email to accept the invitation.

Now those of you who are added as collaborators should clone the repository of the first person on your machines. (make sure that you don’t clone into a directory that is already a repository!)

Each of you should now make some changes to the files in the repository e.g. fix a typo, add a file containing supplementary material. Commit the changes and then push them back to the remote repository. Remember to pull changes before you push.

Creating branches and sharing them in the remote repository

Working with the same remote repository, each of you should create a new branch locally and push it back to the remote repo.

Each person should use a different name for their local branch. The following commands assume your new branch is called my_branch, and your partner’s branch is called their_branch — you should substitute the name of your new branch and your partner’s new branch.

$ git switch -c my_branch		# Create and switch to a new branch.
					# Substitute your local branch name for 'my_branch'.

Now create/edit a file (e.g. fix a typo, add supplementary material etc), and then commit your changes.

$ git push origin my_branch		# Push your new branch to remote repo.

The other person should check out local copies of the branches created by others (so eventually everybody should have the same number of branches as the remote repository).

To fetch new branches from the remote repository (into your local .git database):

$ git fetch origin
Counting objects: 3, done.  remote:
Compressing objects: 100% (3/3), done.
remote: Total 3 (delta 0), reused 2 (delta 0) Unpacking objects: 100% (3/3), done.
9e1705a..640210a master -> origin/master
* [new branch] their_branch -> origin/their_branch

Your local repository should now contain all the branches from the remote repository, but the fetch command doesn’t actually update your local branches.

The next step is to check out a new branch locally to track the new remote branch.

$ git switch their_branch
Branch their_branch set up to track remote branch their_branch from origin.
Switched to a new branch 'their_branch'

Key Points

  • git pull merges remote changes into local branch of repository