Supporting the testing of scientific frameworks with software product line engineering: a proposed approach

Testing scientific software involves dealing with special challenges like missing test oracle and different possible sources of a problem. When testing scientific frameworks, additionally a large variety of mathematical algorithms and possible applications for the framework has to be handled. We propose to use concepts of software product line engineering to handle this variability. The contribution of this paper is a two-step process for reengineering a variability model out of a framework for scientific software. This process is explained with a real case study. Furthermore, we sketch how the variability model can be used to systematically derive system test applications for the framework.