Efficient Mutation Testing in Configurable Systems

Abstract

Mutation testing is a technique to evaluate the quality of test cases by assessing their ability to detect faults. Mutants are modified versions of the original program that are generated automatically and should contain faults similar to those caused by developers' mistakes. For configurable systems, existing approaches propose mutation operators to produce faults that may only exist in some configurations. However, due to the number of possible configurations, generating and testing all mutants for each program is not feasible. To tackle this problem, we discuss to use static analysis and adopt the idea of T-wise testing to limit the number of mutants. In particular, we i) discuss dependencies that exist in configurable systems, ii) how we can use them to identify code to mutate, and iii) assess the expected outcome. Our preliminary results show that variability analysis can help to reduce the number of mutants and, thus, costs for testing.