Git Reset

git reset is the most direct and powerful way to undo changes in Git. It allows you to move your current branch pointer backward to a previous state, affecting one or more of the repository's internal "layers."

1. The Three Layers of Git

To understand Reset, you must understand the three places Git stores code:

1. HEAD: The pointer to the last commit on your current branch.
2. Index (Staging Area): Where files are prepared for the next commit.
3. Working Directory: The actual files you see and edit in your IDE.

2. Reset Modes Comparison

Soft Reset (--soft)

Best for: Undoing a commit but keeping the work staged so you can commit again (e.g., to fix a commit message or combine commits).

git reset --soft HEAD~1

Mixed Reset (--mixed) [Default]

Best for: Unstaging files you accidentally added with git add, while keeping your actual code changes safely in your working directory.

git reset HEAD filename.txt

Hard Reset (--hard)

Best for: "Wiping the slate clean" and returning to a known good state. Use with extreme caution: this will delete all unstaged and uncommitted work permenently.

git reset --hard <commit_hash>

3. Before and After (Visualizing History)

When you perform a reset, the branch pointer moves. The commits that were "after" the reset point don't disappear immediately—they become Orphaned Commits.

• They won't show up in git log.
• They still exist in the Git database temporarily.
• You can find them using git reflog if you need to recover them.

4. Common Workflows

Unstage a specific file

git reset <filename>

Discard all local changes (Nuclear Option)

git reset --hard HEAD

[!CAUTION] Safety on Shared Branches Never use git reset --hard on commits that have already been pushed to a shared repository. This "rewrites history" and will cause massive conflicts for your teammates. Use git revert instead for public history.