Experiences evaluating the effectiveness of JML-JUnit testing

Abstract

This paper reports on the issues the authors encountered while evaluating the JML-JUnit unit testing strategy. Given a predefined set of parameter values, JML-JUnit can automatically provide unit tests for Java programs that have specifications. We present a mutation testing experiment that evaluates the effectiveness of this testing strategy, and the lessons learned from doing this experiment. We conclude that a benchmark will enable the testing research community to meaningfully assess testing approaches.