Efficient Mutation Testing in Configurable Systems

Mutation operators for feature‐oriented software product lines

“…For instance, in the Prop4j SPL, we find no connected base and feature method, wherefore we can not apply the ARC and MRC operators. The results still show that our operators inject either structural or syntactical faults . In a large‐scale software product line that uses feature‐oriented programming in more situations, our operators should be able to create more suitable mutants.…”

“…Our previous results already indicate that we face redundant and equivalent mutants that can drastically increase the costs of mutation testing. In software product line engineering, this issue is arguably even more problematic, due to the variability that allows to instantiate multiple variants . A recent study indicates that variability‐aware mutation operators can easily result in almost 40% equivalent mutants .…”

“…The results we show in Section 5 illustrate the applicability of our proposed operators to resemble variability faults. Identifying challenges in mutation testing of software product lines. Applying mutation testing in variable systems results in new challenges, for example, increasing costs due to testing the same code structures multiple times . Still, testing variability is essential for software product lines, as feature interactions are a potential source for faults .…”

“…Due to the aforementioned problem, it is essential to improve the efficiency of mutation testing for software product lines by adapting cost reduction techniques (cf Section 2.2) . In the context of preprocessor‐based systems, we proposed static analysis and sampling to reduce the number of injected faults on variably code as well as the number of variants under test (do fewer approaches). For example, by using combinatorial interaction testing , variants can be sampled to cover each feature interaction only once.…”

“…In this article, we investigate mutation testing of composition‐based software product line—implemented with feature‐oriented programming —and corresponding challenges, building on our previous works on mutation testing of preprocessor‐based systems . We first propose a set of variability‐aware mutation operators that potentially introduce variability faults for feature‐oriented programming.…”