Test-aware combinatorial interaction testing

Combinatorial interaction testing (CIT) approaches system- atically sample a given configuration space and select a set of configurations, in which each valid t-way option setting combination appears at least once. A battery of test cases are then executed in the selected configurations. Exist- ing CIT approaches, however, do not provide a system- atic way of handling test-specific inter-option constraints. Improper handling of such constraints, on the other hand, causes masking effects, which in turn causes testers to de- velop false confidence in their test processes, believing them have tested certain option setting combinations, when they in fact have not. In this work, to avoid the harmful conse- quences of masking effects caused by improper handling of test-specific constraints, we compute t-way test-aware cov- ering arrays. A t-way test-aware covering array is not just a set of configurations as is the case in traditional covering arrays, but a set of configurations, each of which is asso- ciated with a set of test cases. We furthermore present a set of empirical studies conducted by using two widely-used highly-configurable software systems as our subject applica- tions, demonstrating that test-specific constraints are likely to occur in practice and the proposed approach is a promis- ing and effective way of handling them.