Developing Reliable UI Test Automation in Agile Development
Automated UI tests typically suffer from a number of problems: They are brittle, hard to maintain, hard to read, often test only one browser, and are frequently slow. These problems are amplified in agile environments which often have less up-front planning for these tests, contributing to a number of issues affecting their reliability and maintainability. Michael Durrant says that writing reliable, automated UI tests in agile environments requires a development style that emphasizes a standards-based approach and the application of specific practices such as selecting elements, JavaScript timing, working with AJAX calls, multiple browsers, and particularly intermittent test failures. Because intermittent failures are the bane of automated UI testing, Michael shares tests, code, and approaches to help address them. For his example environment, Michael uses Ruby with Capybara, RSpec, and Selenium testing against multiple browsers from Chrome to Firefox, Safari, and Internet Explorer. In addition, Michael shows how automated UI tests relate to unit, integrated, performance, and exploratory testing in the Four Quadrants of Agile Testing and in the testing pyramid.
After more than twenty years as a software developer, Michael Durrant recognized his true passion for quality engineering and for the past five years has worked in several Boston-based small- to medium-sized startups including Zipcar and Everquote, where he currently works as the principal quality engineer. Michael is a passionate Stack Overflow’er, including the Quality Assurance Stack Exchange site. He frequently relaxes with Unix and Linux, too. Michael has spent the past six years working with the Ruby on Rails framework and has occupied himself recently with writing reliable, automated multibrowser tests.