Improvements of and Extensions to FSMWeb: Testing Mobile Apps
Abstract
A mobile application is a software program that runs on mobile device. In 2017, 178.1 billion mobile apps downloaded and the number is expected to grow to 258.2 billion app downloads in 2022 [19]. The number of app downloads poses a challenge for mobile application testers to find the right approach to test apps. This dissertation extends the FSMWeb approach for testing web applications [50] to test mobile applications (FSMApp). During the process of analyzing FSMWeb how it could be extended to test Mobile Apps, a number of shortcomings were detected which we improved upon. We discuss these first. We present an approach to generate black-box tests to test fail-safe behavior for web applications. We apply the approach to a large commercial web application. The approach uses a functional (behavioral) model to generate tests. It then determines at which states in the execution of behavioral test failures can occur and what mitigation requirements need to be tested. Mitigation requirements are used to build mitigation models for each failure type. From those mitigation models failure mitigation tests are generated.
Next, this dissertation provides an approach for selective black-box model-based fail-safe regression testing for web applications. It classifies existing tests and test requirements as reusable, retestable, and obsolete. Removing reusable test requirements reduces test requirements between 49% to 65% in the case study. The approach also uses partial regeneration for new tests wherever possible. Third, we present the new FSMApp approach to test mobile applications and compare the approach with several other approaches [88, 37]. A number of case studies explore applicability, scalability, effectiveness, and efficiency of FSMApp with other approaches. Future work makes suggestion on how to improve test generation and execution efficiency with FSMApp.