Generalizing AOP for Aspect-Oriented Testing

In profiling program execution to assess the adequacy of a test set, a challenge is to select the code to be included in the coverage assessment. Current mechanisms for doing this are coarse-grained, assume traditional concepts of modularity, require tedious and error-prone manual selection, and leave the tester’s intent implicit in the input provided to the testing tools. The aspect-oriented constructs of languages such as AspectJ promise to help ease and extend our ability to select the code to be included in a test adequacy criterion and to document the tester’s intent within the source code itself. Our contribution is a language-centric approach to automated test adequacy analysis that we call concern coverage. We claim that our approach enables explicit, precise, abstract, and machine-readable representation of the tester’s intent and that it can ease testing by eliminating the need for manual selection and explicit maintenance of test adequacy criteria.