ABSTRACT
Software testing is not a new area. Many texts have been written in this area and several methodologies have been developed. However, the idea of testing software architecture (SA) is comparatively new and requires a more rigorous effort. Testers must not only have good development skills but also be knowledgeable in formal language, graph theory, and algorithms (Whittaker, 2000). Software testing is usually approached in four phases:
1. Modeling the software’s environment 2. Selecting test scenarios 3. Running and evaluating test scenarios 4. Measuring the testing process
This serves as a partition of the entire process of testing, similar to the STEP model given by Eickelmann & Richardson (1996) and Torkar (2005). There is a plethora of books on software testing since the first text by Myers (1978) that address tough testing issues, but the area of software architecture testing has not resulted in a mature methodology that is stable. Research is continuing in the current area. From code-level testing, the testing area has grown to include modelbased testing, Unified Modeling Language (UML) as a means to support the modeling, to development of the Software Architecture Analysis Method (SAAM) framework (Kazman et al., 1994). However, the transition has not been smooth and appears as two separate classes of methodologies. The former is focused toward code level testing and coverage analysis, while frameworks like SAAM are focused toward the entire evaluation and effectiveness of any particular SA. This chapter summarizes the various efforts that have been put in the recent years in these two disparate classes, and argues that DEVS M&S provides an integral framework that helps align these two fields in coherence. Of the four-part process mentioned above, selecting test scenarios appears to be the most time-consuming, rigorous, and well attended in the literature. Test execution is assumed to be simpler until DEVS M&S provides a mathematical framework to conduct test-model execution in a formalized manner.
