Pull Requests and Code Review

Duration: 8 min

Try it in Google Colab:

What is a Pull Request?

A Pull Request (PR) is a way to propose changes to a repository. It allows others to review your code before merging it into the main branch.

Creating a Pull Request on GitHub

First, push your branch to GitHub:

# Create a local branch
git checkout -b feature/add-validation

# Make changes
echo 'def validate_input(x):' >> model.py
echo '    return x is not None' >> model.py

# Commit changes
git add model.py
git commit -m 'Add input validation'

# Push to GitHub
git push origin feature/add-validation

Enumerating objects: 3, done.
Counting objects: 100% (3/3), done.
Writing objects: 100% (3/3), 250 bytes, done.
To github.com:username/my-ai-project.git
 * [new branch]      feature/add-validation -> feature/add-validation

Opening a Pull Request

On GitHub, you'll see a prompt to create a PR:

1. Click "Compare & pull request"
2. Add a title: "Add input validation"
3. Add a description explaining your changes
4. Click "Create pull request"

Code Review Process

Once your PR is created, team members can:

• Review your code
• Comment on specific lines
• Request changes if needed
• Approve when satisfied

Example review comment:

This function looks good, but consider adding error handling for None values.

Responding to Review Comments

# Make the requested changes
echo 'def validate_input(x):' > model.py
echo '    if x is None:' >> model.py
echo '        raise ValueError("Input cannot be None")' >> model.py
echo '    return True' >> model.py

# Commit and push
git add model.py
git commit -m 'Add error handling for None values'
git push origin feature/add-validation

The PR automatically updates with your new commits!

Merging a Pull Request

Once approved, merge the PR:

1. Click "Merge pull request" on GitHub
2. Choose merge strategy (usually "Create a merge commit")
3. Click "Confirm merge"
4. Delete the branch (optional but recommended)

Best Practices for PRs

💡 Tip: Keep PRs focused on one feature. If you're adding validation AND preprocessing, make two separate PRs.