PIs: Mary Jean Harrold, Alessandro Orso, Gregg Rothermel
When object-oriented and component-based software systems are developed, they must be integration tested. Integration testing methods proposed to date for such systems, however, suffer from three limitations: they fail to consider system evolution, they fail to accommodate heterogeneous engineering models and sources of nformation, and they provide little or no empirical investigation or understanding of the factors influencing the testing effort. This research addresses these limitations. The initial objective of the work is to design and conduct families of experiments to compare the cost-benefits of existing techniques for integration testing applied to evolving object-oriented and component-based software, and evaluate the potential tradeoffs among those techniques. The results of these studies will guide the development of evolution-aware integration testing techniques, capable of accommodating heterogenous sources of testing information, and responding to the wide range of engineering models associated with object oriented and component-based software. By making experimental data, infrastructure, and new experimental methodologies available to the research community, the proposed work will enable scientific progress on the issues investigated to accelerate. By providing methodologies for testing this important class of software systems, the work will reduce the risks these systems present to society.
