Handling Constraints for Search Based Software Test Data Generation

A major issue in software testing is the automatic generation of the inputs to be applied to the programme under test. To solve this problem, a number of approaches based on search methods have been developed in the last few years, offering promising results for adequacy criteria like, for instance, branch coverage. We devise branch coverage as the satisfaction of a number of constraints. This allows to formulate the test data generation as a constrained optimisation problem or as a constraint satisfaction problem. Then, we can see that many of the generators so far have followed the same particular approach. Furthermore, this constraint-handling point of view overcomes this limitation and opens the door to new designs and search strategies that, to the best of our knowledge, have not been considered yet. As a case study, we develop test data generators employing different penalty objective functions or multiobjective optimisation. The results of the conducted preliminary experiments suggest these generators can improve the performance of classical approaches.