Techniques for selective revalidation

The authors address the question of how to decide which test cases to rerun after a modification. They emphasize that it is important that these tests be selected systematically, because executing an entire test suite to validate a few modifications can consume large amounts of time and computational resources and involve many people, and it is unreliable to exercise a system by selecting test cases intuitively or randomly. They develop a revalidation strategy that is based on an extension of the Fischer algorithm (see K.F. Fischer et al., Proc. Nat. Telecom. Conf., 1981, p.B6.3.1-B6.3.6). Fischer's revalidation technique is based on a zero-one integer programming model. The authors implement a prototype environment based on his methodology.>