A prototype tool for generating and executing test cases from UML-based interface behavior descriptions

Abstract

We present the Configurable Advanced Verification of Software (CAVS), a prototype tool that automatically generates an efficient set of test cases from descriptions of required interface behavior based on Unified Modeling Language (UML) artifacts. The input to CAVS consists of enhanced UML activity diagrams that define the required behavior of a software component as observed on its interfaces. CAVS automatically analyzes activity diagrams and uses the choice-relation framework to partition the input space. Partitions are selected to define a reduced set of abstract test cases by calculating a covering array. Concrete test cases are generated by using boundary value analysis to select specific values for each partition in the covering array. The test cases are then executed using cloud computing technologies. Lastly, expected outputs are automatically verified.