Comparative analysis of two algorithms of combinatorial techniques for testing SPLs — An exploratory study in six SPLs

摘要

Software Product Line (SPL) is an approach for developing a family of software systems that has differences among them - SPL variabilities. The number of variability and number of products make the exhaustive tests in SPLs. The generation of valid and representative subset of tests for all possible product configurations is an open challenge. This paper presents a comparative analysis of two algorithms of combinatorial techniques for testing SPLs. Those algorithms were used in six SPLs and the results were evaluated by four metrics. The main contribution of this evaluation was to find the best solution generated by algorithms using metrics which allowing identify the quality of results. Those results confirm that the building of a set of test for SPLs requires thinking about new strategies, because some test methods in SPL consider that features are independent, but their interactions are possible and influence in the product construction and testing.