Reverse engineering of test cases for selective regression testing

A problem arises in the maintenance of large systems when the links between the specification based test cases and the code components they test are lost. It is no longer possible to perform selective regression testing because it is not known which test cases to run when a particular component is corrected or altered. To regain that knowledge, it is necessary to restore the links between test cases and code. We describe two complementary approaches to achieving that goal of reverse engineering - one by static analysis and the other by dynamic analysis. The objective in both cases is to determine which test cases exercise which methods in which components in order to enable selective regression testing.