gMutant: A gCov based Mutation Testing Analyser

Abstract

gMutant is a novel mutation analyser to measure the quality of test cases. It is based on gCov. The existing gCov outputs code coverage (line and branch). Here, we scale it up in order to produce an effective metric i.e., Mutation Score (%) which is useful to report the quality of test cases. The key idea is to generate one Mutant Meta Program that has all the mutants instrumented in it. Hence, for each test input, only the generated Mutant Meta Program gets executed, instead of executing all the mutants separately. This mitigates the computational cost challenge of Mutation Testing in producing Mutation score. The gMutant that has been proposed is a generic tool which can be plugged in with any tester. It needs test cases as well as a program to generate a score (%) and a descriptive report. In our experimentation, we used Bounded Model Checker and Fuzzer to generate test cases. The run time of the gMutant is very advantageous in contrast to the traditional method of computing mutation score. It gives scores, in 4.03 (s) for CBMC and 4.01 (s) for AFL, which proves its efficiency.