testar – scriptless testing through graphical user interface

• Published in: Software testing, verification & reliability Vol. 31; no. 3
• Main Authors: Vos, Tanja E. J., Aho, Pekka, Pastor Ricos, Fernando, Rodriguez‐Valdes, Olivia, Mulders, Ad
• Format: Journal Article
• Language: English
• Published: Chichester Wiley Subscription Services, Inc 01.05.2021
• Subjects: artificial intelligence; Graphical user interface; model inference; monkey testing; test automation; User interface
• ISSN: 0960-0833; 1099-1689
• DOI: 10.1002/stvr.1771

Summary Covering all the possible paths of the graphical user interface (GUI) with test scripts would take too much effort and result in serious maintenance issues. We propose complementing scripted testing with scriptless test automation using the open‐source testar tool. This paper gives a comprehensive overview of testar and its latest extensions together with the ongoing and future research. With this paper, we hope we can help and encourage other researchers to use testar for their GUI testing‐related research and pave the way for an international research agenda in GUI testing built upon stable and open‐source infrastructure. This article gives a comprehensive overview of TESTAR, an open source tool for scriptless graphical user interface (GUI) test automation, and presents the latest TESTAR extensions and industrial experiments. The novel extensions include, for example, active learning of state models (model inference) and support for Selenium WebDriver. The inferred state models enable more advanced action selection algorithms, for example, optimizing the GUI exploration with machine learning. Support for Selenium WebDriver improves TESTAR in recognizing widgets and properties of web applications.