Combinatorial Test Generation for Software Product Lines Using Minimum Invalid Tuples

A software product line is a set of software systems that share some common features. Several recent works have been reported that apply combinatorial testing, a very effective testing strategy, to software product lines. A unique challenge in these efforts is dealing with a potentially large number of constraints among different features. In this paper, we propose a novel constraint-handling strategy that uses minimum invalid tuples (MITs) as an alternative to traditional constraint solvers. Our approach systematically derives all MITs from a software product line, and uses them to quickly determine the validity of a test configuration during test generation. We implemented a test generation research tool called LOOKUP that integrates the proposed constraint-handling strategy with a general test generation algorithm called IPOG-C. Experimental results show that LOOKUP performs considerably better than two existing test generation tools in terms of test size and execution time.