How Git Checkout's Previous Branch Shortcut Works Under the Hood

One piece of Git shorthand I use all the time is git checkout -. Much like cd -, it references the previous item in your history. In the case of cd, it will change your current directory to the previous one you were in. So,

$ echo $PWD
$ cd code/git-req
$ echo $PWD
$ cd -
$ echo $PWD

Git has its own version of this:

$ git branch --show-current
$ git checkout my-new-feature
Switched to branch 'my-new-feature'
Your branch is up to date with 'origin/my-new-feature'.
$ git checkout -
Switched to branch 'master'
Your branch is up to date with 'origin/master'.

Handy! But how does it work? I poked through the files in .git (and even watched the filesystem for changes, but nothing jumped out).

The good news is that Git is open source. So, let's go to the source.

The Git repository is big. Git does a lot, it makes sense. How do I learn where to look? Well, I have bit of prior knowledge. I'm the author of a Git extension named git-req. The hyphen in that name? It is a nifty way to inject commands into the git namespace. So, if I wanted to invoke it, I run git req (no hyphen)! Git's codebase is similarly laid out, with each entry point represented by its own file.

├── add.c
├── ...
├── branch.c
├── ...
├── checkout.c
├── ...
├── clone.c
├── ...
├── commit.c
└── ...

You can see some greatest hits there. What's interesting to us is checkout.c - that's the command we're looking for!

Now, where to look in that file? By convention, most C programs use main as an entrypoint. Let's search for that.

checkout_main()? Check.

Now, how does that method receive arguments? Visually scanning the function surfaces this command:

 * Extract branch name from command line arguments, so
 * all that is left is pathspecs.

I think we're getting warmer. A few lines down, there's a reference to parse_branchname_arg(). Since the hyphen that we're looking for is a functional stand-in for a branch name, that sounds like a logical place to look next. Sure enough, at the top of the function, there's the following expression:

if (!strcmp(arg, "-"))
    arg = "@{-1}";

Bam. If the argument is a hyphen, then replace it with @{-1}.... wat. Let's check the docs:

Branch to checkout; if it refers to a branch (i.e., a name that, when prepended with "refs/heads/", is a valid ref), then that branch is checked out. Otherwise, if it refers to a valid commit, your HEAD becomes "detached" and you are no longer on any branch (see below for details).
You can use the @{-N} syntax to refer to the N-th last branch/commit checked out using "git checkout" operation. You may also specify - which is synonymous to @{-1}.

... I guess I should've started there. But, this raises another question. How does Git store this history? What is the terminology for this new syntax?

a few minutes of grepping later...

This is a syntax for something called a reflog. What's that? Sorry to disappoint you, but I'm not diving into the code without checking the docs first. git-reflog is what we're looking for.

Reference logs, or "reflogs", record when the tips of branches and other references were updated in the local repository. Reflogs are useful in various Git commands, to specify the old value of a reference. For example, HEAD@{2} means "where HEAD used to be two moves ago", master@{one.week.ago} means "where master used to point to one week ago in this local repository", and so on. See gitrevisions for more details.

So, this data is stored in the reflog, and the syntax is a gitrevision.

The construct @{-<n>} means the th branch/commit checked out before the current one.

There we have it. To recap, git checkout - is the same as git checkout @{-1} This more verbose gitrevision syntax unlocks the ability to jump further back in your history.