 |
Random Search versus Uniform Sampling, A Counter-intuitive Result
1 : Laboratoire des Sciences du Numérique de Nantes
Institut National de Recherche en Informatique et en Automatique, Centre National de la Recherche Scientifique : UMR6004, IMT Atlantique, Nantes Université - École Centrale de Nantes, Nantes université - UFR des Sciences et des Techniques, Centre National de la Recherche Scientifique
With the increase in software size and complexity comes the need of good automatically generated test suites. For example, the Linux kernel has thousands of features (e.g. options, libraries, ...). A commonly used quality measure of a test suite is the t-wise coverage. It aims at analysing how many interactions of t features are tested.
We show here that a random search, i.e. a search strategy that chooses randomly a feature, and add it to a configuration or not, outperforms a uniform sampler for the task of t-wise coverage on feature models.
