Software partitioning for effective automated unit testing

Abstract

A key problem for effective unit testing is the dificulty of partitioning large software systems into appropriate units that can be tested in isolation. We present an approach that identifies control and data inter-dependencies between software components using static program analysis, and divides the source code into units where highly-intertwined components are grouped together. Those units can then be tested in isolation using automated test generation techniques and tools, such as dynamic software model checkers. We discuss preliminary experimental results showing that automatic software partitioning can significantly increase test coverage without generating too many false alarms caused by unrealistic inputs being injected at interfaces between units.