How to Use the Git Reset Command

Today, it's hard to imagine the work of a programmer or IT professional without version control. Among the various SCM tools, Git stands out, having quickly gained popularity and becoming the de facto standard in the world of version control systems. Git allows you to easily track project file changes, manage branches, collaborate, and centrally store code and other files. 

One of Git's strengths is its flexible ability to undo or remove changes. One such way to undo changes is with the git reset command, which supports three different modes. In this tutorial, we'll explore how to undo changes using git reset and its modes through practical examples.

Prerequisites

We'll focus on practical use cases of the git reset command, so it's necessary to have Git installed beforehand.

We'll use a Linux-based operating system for this tutorial, specifically Ubuntu 22.04. However, any Linux distribution will work, as Git is available in nearly all modern package managers.

In most distributions, Git comes pre-installed, though the version may not always be the latest. For Ubuntu-based systems, you can install Git from the official repository with the following commands:

add-apt-repository ppa:git-core/ppa && apt -y install git

For other Debian-based distributions (Debian, Linux Mint, Kali Linux, etc.), you can install Git using:

apt -y install git

For RHEL-based distributions (RedHat, CentOS, Fedora, Oracle Linux), the installation command will vary depending on the package manager:

For yum package manager:

yum -y install git

For dnf package manager:

dnf -y install git

After installation, verify the Git version:

git --version

What is git reset?

The git reset command is used to undo local changes. Technically speaking, git reset moves the HEAD pointer to a previous commit in the repository. HEAD is a pointer to the current branch and points to the latest commit in that branch.

The git reset command operates with three key elements: the working directory, the HEAD pointer, and the index. These elements are often referred to as "trees" in Git, as they are structured using nodes and pointers. We'll go into detail about each of these elements below.

It's worth noting that various Git-based web services like GitHub, GitLab, and Bitbucket offer the ability to undo actions through their web interface. However, they typically use a safer alternative, git revert, which preserves the entire project history, unlike git reset which can permanently remove commits.

The Working Directory

The working directory is where files are stored and tracked by Git. When you run the git reset command, Git knows which directory is being tracked because of a hidden .git folder created when you initialize a repository with git init.

Here's how the working directory works in practice:

1. Create a new directory and navigate into it:

mkdir new_project && cd new_project

2. Initialize a new Git repository:

git init

Once you initialize the repository, a hidden .git folder containing Git configuration files is created in the root directory.

The HEAD Pointer

HEAD points to the current branch and the latest commit in that branch. Every time you switch branches with git checkout, HEAD updates to point to the latest commit in the new branch.

Here's a practical example:

1. Create a new file:

touch new1.txt

2. Add the file to the repository:

git add new1.txt

3. Commit the file:

git commit -m "Initial commit"

4. To see where HEAD is pointing, use the git cat-file command:

git cat-file -p HEAD

Since there's only one commit, HEAD points to it.

Now, let's modify the file and add it again.

5. Modify the file:

echo "This is a test file" > new1.txt

6. Stage the file:

git add new1.txt

7. Commit the changes:

git commit -m "Added content to new1.txt"

8. Check the HEAD pointer again:

git cat-file -p HEAD

As you can see, HEAD now points to the new, latest commit.

The Index

The index (or "staging area") is where files go after being added with git add. Think of it as a pre-commit area. Files in the index are tracked by Git but not yet part of the actual commit. You can remove or modify files in the index before they are committed.

Create a new file:

touch new2.txt

Add it to the index:

git add new2.txt

Check the status:

git status

The file is now in the staging area but not yet committed.

 

Git Reset Modes

The git reset command supports three modes: soft, mixed, and hard.

Soft Mode

The soft mode undoes the last commit but keeps the changes in the index. This means that you can modify and recommit them.

Create a new file:

touch new3.txt

Add it to the index:

git add new3.txt

Commit the file:

git commit -m "Added new3.txt"

If we run git log now, that's what we'll see:

To undo the last commit:

git reset --soft HEAD~1

The commit is undone, but the file remains in the index.

Mixed Mode

The mixed mode is the default for git reset. It undoes the commit and resets the index, but leaves the working directory untouched.

Create three new files:

touch new{1..3}.txt

Add and commit them:

git add new1.txt new2.txt new3.txt
git commit -m "Added three files"

Now undo the commit:

git reset HEAD~1

The files remain, but the last commit is removed.

Hard Mode

The hard mode deletes the commit, resets the index, and removes the files from the working directory. This is the most destructive option.

Create and commit a file:

touch readme.md
git add readme.md
git commit -m "Added readme.md"

To remove the commit and the file:

git reset --hard HEAD~1

The file and the commit are permanently deleted.

Resetting to an Earlier Commit

You can also reset to a specific commit using its hash:

git reset --hard <commit-hash>

This will reset the repository to that specific commit.

Conclusion

In this tutorial, we explored the git reset command and its modes: soft, mixed, and hard. While git reset is a powerful tool for undoing local changes, it's essential to understand each mode's impact, especially the potential risks of using the hard mode to avoid irreversible data loss.