Tools for test case generation

Abstract

System vendors focus more and more on the quality of a system instead of increasing functionality. Testing is the most viable and widely used technique to improve several quality aspects, accompanying the entire development cycle of a product. Motivated by the success of model-based software development and verification approaches, model-based testing has recently drawn attention of both theory and practice. System development tools reflect this tendency in many ways, automatic model-based generation of test suites has incipiently found its way into practice. TestComposer and AutoLink are the dominating design tools in the SDL community. The UTP serves the need for test support within UML-based software development, and Microsoft's AsmL is another example for the effort major companies make to benefit from the existing theory. But whatever theory is chosen as a basis, none of them can belie the dominating problem of system complexity. Even simple behavioral models like FSMs or LTSs can generally not be specified or exploited exhaustively. In that sense testing is always a David vs. Goliath struggle, even when pragmatical approaches were chosen. Nevertheless it is worth the effort of improving the theory w.r.t. practicability. Furthermore there are system criteria which are not treated satisfactorily yet, like real-time constraints or symbolic data, e.g. infinite data domains. Although automatic testing is still in the fledgling stages it can already be exerted successfully to improve the quality of real world systems. Further research is needed to improve and ease its application. It is a promising field where formal methods find their way into practice. © Springer-Verlag Berlin Heidelberg 2005.