Reliable and Maintainable Webdriver Tests
summary
Built a suite of hundreds of WebDriver end-to-end tests that run in under ten minutes by parallelizing tests (six to seven per node), stubbing remote services and serving assets locally to remove network flakiness, and adding a quarantine JUnit rule that captures HAR logs, JS console output and screenshots when tests fail nondeterministically. Tests use explicit wait utilities and assertion wrappers, adopt the page object pattern and screenplay pattern for composability, run against BrowserStack for cross‑browser checks, and double as production monitoring while tolerating rare driver or graphics failures.
AI-written summary · about
chapters (13)
- 0:00Reliable end-to-end testing
- 1:53Introduction
- 3:41Why end-to-end tests matter
- 8:20Journey with automated testing
- 13:39Prototype for end-to-end tests
- 16:10Page object pattern
- 19:29Diagnostics and reliability
- 23:51Reducing flakiness
- 26:44Tests as production monitoring
- 30:16Performance tips for suites
- 34:36Screenplay pattern
- 38:38Key takeaways
- 41:18Q&A
related talks
30mLessons from testing browsers using WebDriver
53mModern Web Testing: Going Beyond Selenium
37mAutomated Performance Testing With WebDriver
01h 14mContinuous Testing
46mThe Road to Reliable Automated Tests
30mModern Web Testing: Going Beyond Selenium
36mModern Web Testing: Going Beyond Selenium
55mModern Web Testing: Going Beyond Selenium