Evaluating TESTAR's effectiveness through code coverage

Testing is of paramount importance in assuring the quality of software products. Nevertheless, it is not easy to judge which techniques or tools are the most effective. A commonly used surrogate metric to evaluate the effectiveness of testing tools is code coverage, which has been widely used for unit and integration testing. However, for GUI testing approaches, this metric has not been sufficiently investigated. To fill this gap, we run experiments with the TESTAR tool, a scriptless testing tool that automatically generates test cases at the Graphical User Interface (GUI) level. In the experiment, we analyze and compare the obtained code coverage when using four different action selection mechanisms (ASMs) in TESTAR that are used to test three SUTs.