Quasi-Dominators and Random Selection in Mutation Testing

Mutation Testing is a powerful approach to bug detecting and assessing code quality; however, software developers may be reluctant to embrace the technique due to the monstrous quantity of redundant mutants it generates. In spite of their large numbers, redundant mutants are relatively innocuous. Recent research indicates that redundant mutants affect a test engineer's work effort only slightly, whereas equivalent mutants have a direct linear impact. Moreover, the time invested analyzing equivalent mutants produces no unit tests. Dominator mutants seek to address the redundancy problem, but using them requires first identifying and killing all nonequivalent mutants, essentially the same work required by random selection, in order to identify subsumption relationships. This paper introduces the notion of quasi-dominator mutants, which together with dominator mutants are readily encountered by random selection, and provides new insight into why random mutant selection performs so well.