Avoiding useless mutants

摘要: Mutation testing is a program-transformation technique that injects artificial bugs to check whether the existing test suite can detect them. However, costs of using mutation are usually high, hindering its use in industry. Useless mutants (equivalent and duplicated) contribute increase costs. Previous research has focused mainly on detecting useless only after they generated compiled. In this paper, we introduce strategy help developers with deriving rules avoid generation mutants. To our strategy, pass as input set programs. For each program, also need passing As output, yields candidates. After manually confirming classified by "useless" indeed useless, derive their thus decrease best knowledge, 37 new right before generation. We then implement subset these MUJAVA tool. Since have been derived based small Java programs, take version embedded execute it industrial-scale projects. Our reduced number almost 13% average. results promising because (i) generation; (ii) identifying more case complex programs; (iii) derived.