Action-State Testing—A Model for Test Design Automation

Model-based testing (MBT) is essential in software testing, offering automation, comprehensive coverage, and defect prevention. It uses abstract models to automatically design and generate test cases, representing the expected system behaviour, including states, transitions, inputs, and outputs. This paper explores the action-state testing modelling technique, originally introduced by the authors in Forgács and Kovács (2000). In this approach, a model step comprises an action (input), one or more responses (outputs), and an optional state. The steps can be arranged sequentially, or they may be forked and joined. Sequential steps appear within the same test case. Forked steps are distributed across different test cases. The joined steps also belong to separate test cases. In addition, the graphical model can be constructed using a text editor. This paper builds upon the concept by establishing its theoretical foundation. We demonstrate how the action-state model eliminates the need for guard conditions and coding, maintains a concise and manageable structure, and seamlessly incorporates outputs, ultimately enhancing testing efficiency. Additionally, we provide guidelines for adding new states and empirically validate the benefits of action-state testing over alternative techniques, achieving a 100% defect detection percentage (DDP). This paper marks the first instalment of the author’s Test Design Trilogy, dedicated to refining and unifying various test design techniques.