Dependence Testing: Extending Data Flow Testing with Control Dependence

This paper presents a new approach to structural testing, called dependence testing. First we propose dependence oriented coverage criteria that extend conventional data flow oriented coverage criteria with control dependence. This allows one to capture the full dependence information of a program or specification systematically. We then describe a model checking-based approach to test generation for dependence testing. It is shown that dependence oriented coverage criteria can be characterized in the temporal logics LTL and CTL. This enables one to use any LTL and CTL model checkers as test generators. Finally, we show that the temporal logic-based characterization can also be used for reducing the cost of dependence testing.