Getting productive with git checkout

This post was originally posted on my blog.

---

As a developer, I use git on a daily basis. It has become an essential tool for my work. It’s great and I love it.

I also enjoy interacting with it directly in the Terminal. There’s something magical in writing a command and seeing it do something. I’ve tried using GitHub Desktop for some time but always came back to the Terminal.

The problem

There’s one command, though, that I used to like and hate at the same time. It’s git checkout.

I like it because of its simplicity and effectiveness. With it, I can create a new branch and switch to it all at once. Just like this:

git checkout -b new-feature

Beautiful.

But, in some cases, I hated it too because of the way it works. Using it as above creates a branch that is based off of the one I’m currently on and that isn’t something I always want.

Here’s a use case that I might have during my work.

Say I’m on a new-feature branch and have some changes, not related to the current task, I want to transfer to a new branch. Simply using git checkout -b new-awesome-feature will not help me because all commits made thus far will follow as well. In those scenarios my workflow was:

1. Stash everything.
2. Switch to the branch I want to base the new one off.
3. Use git checkout command to create a new branch and switch to it.
4. Pop the changes from the stash and continue working with them.

In code that looks like this:

git stash
git checkout master
git checkout -b new-awesome-feature
git stash pop

Definitely not the best workflow.

The solution

About a month ago I decided to see if I could streamline this process because I wasn’t quite happy with its current form. So I opened the git checkout docs and started to dig in. Fortunately, the solution was right there! Among the many options on that page, there’s one called <start_point>. By using it, I can specify a starting point for the new branch. It says that I can use “the name of a commit” but there’s actually more to it. Following the link to git branch reveals that ”[<start-point>] may be given as a branch name, a commit-id, or a tag”. This was exactly what I was looking for!

So now, whenever I want to transfer changes to a new branch that must have a different starting point than the current one, I use:

git checkout -b <branch-name> <start-point>

All four steps from above are down to just one! Pure enjoyment!

Ever since I found this, I’m at peace with the checkout command. I like it even more. Hope you do as well!

Comments

[Mesalazine]

You can also hard reset your new branch in case you forgot to provide starting point:

# when on <master>
git checkout -b new-awesome-feature
# <new-awesome-feature> is now at <master>
git reset --hard old-but-still-awesome-feature
# <new-awesome-feature> will now be at <old-but-still-awesome-feature>

[Dzhavat Ushev]

Oh, that‘s awesome! Will definitely use it. Thanks for the tip.

[nervewax]

Oh wow 🤯 yeah I'm going to use this all the time now, thanks @dzhavat

[Alex Parra]

Ah ah! Had this exact issue today.
Best!