A Comparison of Some Reliable Test Data Generation Procedures

摘要

A set of mutants of a program P, M(P), is a finite subset of the set of all programs written in the language of P, and EM(P) is the set of programs in M(P) which are (functionally) equivalent to p. For a set of test data T, DM(P,T) is the set of programs in M(P) which give results differing from P on at least one point in T. As described elsewhere, it is possible to choose the function M so that ms (P,T) = 1 only if T demonstrates the correctness of P with high probability. This paper is a case study of four test data generation schemes. For a fixed program P, five sets of test data are generated and mutation scores are calculated using the FMS.2 mutation system. Since each set has a score less than one, the FMS.2 system is used to derive a set T such that ms(P,T)=1.