A Comparative Study of GUI Testing Approaches

• Autor(a) principal: Rui Emanuel Veigas de Carvalho
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/85357
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

Most of the modern software applications feature a Graphical User Interface (GUI), which turns the application easier to use, promoting higher productivity and better accessibility, and offering flexibility in how users perform tasks. However, due to GUI's complexity, the GUI testing process can be a time-consuming and intensive process. Therefore, automate the process as much as possible is indispensable to test any more evolved graphic user interface. There are some common automated GUI testing approaches, but while most of them require substantial manual efforts, others lack reusability or are only able to find specific types of errors. Many researchers state that a variety of techniques should be used. A new model-based testing approach, called Pattern- Based GUI Testing, was implemented in order to increase systematization, reusability and diminish the effort in modelling and testing. It is based on the concept of User Interface Test Patterns (UITP), which contain generic test strategies for testing common recurrent behavior (UI Patterns) on GUIs, and supported by the PBGT Tool which provides an integrated modeling and testing environment that supports the crafting of test models based on UI Test Patterns, using a GUI modeling DSL (PARADIGM). As a novel proposal, it is entirely relevant to submit it to systematized experiments and tests in order to assess its good performance/behavior and compare it with other techniques. Thus, this dissertation work mainly addresses PBGT's approach, aiming to compare it with different testing approaches/tools in what concerns to errors/fault detection, ease of use, and overall efforts required to test the application. To perform the experiments, mutations were introduced in each of three different web applications - iAddressBook, TaskFreak and Tudu - to cover a greater number of use cases, and each mutant was tested by each of the selected or developed testing tools which implement the considered approaches. Those approaches' benefits and problems are then conveniently described.