Supporting Concern-Based Regression Testing and Prioritization in a Model-Driven Environment

Abstract

Traditional regression testing and prioritization approaches are bottom-up (or white-box). They rely on the analysis of the impact of changes in source code artifacts, identifying corresponding parts of software to retest. While effective in minimizing the amount of testing required to validate code changes, they do not leverage on specification-level design and requirements concerns that motivated these changes. Model-based testing approaches support a top-down (or black box) testing approach, where design and requirements models are used in support of test generation. They augment code-based approaches with the ability to test from a higher-level design and requirements perspective. In this paper, we present a model-based regression testing and prioritization approach that efficiently selects test cases for regression testing based on different concerns. It relies on traceability links between models, test cases and code artifacts, together with user-defined properties associated to model elements. In particular we describe how to support concern-based regression testing and prioritization using TDE/UML, an extensible model-based testing environment.