Lesson 33 · Creating a branch

Creating a branch

A new branch is just a new ref pointing at a commit. Today you create one, and confirm it starts life pointing at the same commit as the branch you are on, without moving HEAD.

The goal

Create a branch ref at the current commit, leaving HEAD unchanged.

Start here - the target

TO DO

Scenario: A new branch points at the current commit

Giventhe repository on main at 9bae7f0a71d91d794187cf9de39901bfbfbfc7b6, with HEAD as ref: refs/heads/main

WhenBranch(dev) is called

Thenrefs/heads/dev is created holding 9bae7f0a71d91d794187cf9de39901bfbfbfc7b6, the same id as refs/heads/main

AndHEAD is unchanged, still ref: refs/heads/main, so creating a branch does not switch to it

Background

Creating a branch is one of Git’s cheapest operations because a branch is one small file. Branch("dev") resolves the current commit and writes it to refs/heads/dev. Now two branch labels, main and dev, point at the very same commit 9bae7f0a.... No objects are copied; both are just names for the same tip.

Crucially, creating a branch does not move HEAD onto it: you stay on main. This separation matters, because the branch you are on (HEAD) and the branches that merely exist are different things. To start adding commits to dev you must first switch to it, which retargets HEAD, and that is the last piece of this chapter.

Make it work

// point a new branch ref at wherever HEAD currently resolves
func (r *Repo) Branch(name string) error {
  id, _ := r.Resolve("HEAD")
  return r.UpdateRef("refs/heads/"+name, id)
}

CheckpointDONE

You can create branches. Commit and stop here.