Yunho Kim  Hyunwoo Kim  Woong-gyu Yang  Moonzoo Kim

10.5626/JOK.2018.45.4.342

Mutation Analysis program mutation mutation testing C program Software Testing

Mutation analysis generates mutants of a target program by applying syntactic changes to the source code and analyzes the difference of execution results of the mutants from those of the original program. For effective mutation analysis, mutant generation tools should be able to generate effective program mutants. For example, a mutant that is semantically equivalent to the original program or another mutant is not an effective mutant, because it does not generate an execution result different from that of the original program or another existing mutant. This paper presents a comparative study of two mutant generation tools for C programs, Proteum and Milu. To generate effective mutants effectively, we generated a canonical form of mutated expressions and removed duplicated mutants that have the same canonical form as that of other mutants. We applied Proteum and Milu to four Linux/Unix utilities in the SIR benchmark and showed that 48.7% and 46.4% of mutants generated by Proteum and Milu were effective mutants on average, respectively.