MODEL ORIENTED CAPTURE ANALYSIS (MOCA)
William E. Howden, R. T. Kim, P. P. Tran Cse · 2002
Model-oriented testing requires that a program model be covered during testing. Often a model is not available, either because other kinds of specifications were used or because no specification is available. In the MOCA paradigm, as in extreme programming, tests are used as a specification for a system. MOCA carries this further, using them to construct a model of the kind that is associated with model-based testing. A MOCA model summarizes the tests that have been performed, provides a high level functional specification, and can be used to suggest addition kinds of tests to perform. The basic ideas of the MOCA approach are described, along with a prototype MOCA tool that builds models from Rational Robot traces. Additional features of the MOCA tool include wizards that suggest additional kinds of tests that might be carried out, either because the model seems incomplete, or because it has features that suggest the possible occurrence of commonly occurring kinds of faults. A theory of behavioral modeling is introduced that formalizes the concepts of model abstraction and model equivalence. It introduces the idea of a context-sensitive state model. Additional concepts include that of the contextfree-completion of a set of traces, which is used analyze the soundness of a modelsynthesizing procedure.