Leveraging Parameter and Resource Naming Conventions to Improve Test Suite Adherence to Persistent State Conditions (thesis)

Abstract

With the prevalence of web applications increasing daily in various aspects of modern life, the need for cost-e ective, e cient, and thorough web application testing is now greater than ever. One approach to web application testing is automatic test suite generation based on real user sessions. This approach is promising, but tends to leave a considerable amount of web application functionality untested. It remains overall ine ective due to most automatically generated test suites' lack of adherence to the persistent state of the application under test, or the parts of the application's state that are stored in a data store external to the application itself. This thesis explores the possibillity of leveraging the resource and parameter naming conventions of web applications to automatically predict the actions test cases will perform on data in persistent application state. I propose an algorithm that creates these predictions and another that leverages these predictions to improve test suites to more closely adhere to the conditions of data in persistent state. These improvements are achieved through optimization of request parameter selection. I perform experiments to determine the success of the two algorithms and I propose suggestions for improvements as well as suggestions for future work.