Automated Regression Testing with Splinter and Jupyter
By Jace Browning

This talk will start with a background on testing and quality assurance:

• Terminology
• Unit vs. Integration vs. System (end-to-end)

• Automated vs. Exploratory

• Goals of end-to-end testing

• Test production server configurations
• Test JavaScript engines in browsers
• Responsiveness/visual testing
• External API hookups

The next section will outline a layered approach for writing end-to-end tests with each step building on the last:

• Selenium basics
• Language bindings (Python examples)
• Sequential test scripts
• Adopting the “Page Object Model”
• Using a WebDriver API wrapper (Splinter examples)
• Helper functions to enable automated exploratory testing

Before the final section, there will be a demo of the previous examples run against a live website.

The final section will recommend an approach to building tests against your own site:

• Alternatives
• Dogfooding
• Beta testers

• Manual testers

• Make your site testable

• Name your forms and fields
• Accessible sites are testable sites
• Test mode for account signups
• Expose internal state when possible

• Allow “cheats” on your staging server

• Writing tests

• Start with under-tested, high-risk features
• Make your bot users identifiable for easy cleanup
• Run your tests every time you deploy
• Incorporate logging, screenshots, and reports