Reducing the Cost of Mutation Testing with the Use of Primitive Arcs Concept

Mutation testing is a testing criterion used to measure the quality of a test suite. In mutation, a test suite is executed against the set of mutants of a given program under testing. A score is computed to measure the adequacy of the test suite in detecting faults. Although powerful, mutation testing has two major drawbacks: the high-computational cost to generate and execute the set of generated mutants and the existence of equivalent mutants. In this paper, we present a preliminary experimental study to investigate the use of control-flow information, aiming to reduce the number of mutants. For this study, only a subset of mutants, defined by its location, is executed. Such location is determined by the set of primitive arcs of a given program under testing. Next, it is analyzed the relationship between minimal mutants and primitive arcs. Results indicate that the approach reduces the number of mutants and equivalent mutants and, in most cases, still maintains a high mutation score concerning full mutation. Moreover, the results also indicate that there is a concentration of minimal mutants on the nodes related to primitive arcs. Finally, we compare the effectiveness of our strategy over random mutant sampling.