Infeasible paths in the context of data flow based testing criteria: identification, classification and prediction

infeasible paths constitute a bottleneck for the complete automation of software testing, one of the most expensive activities of software quality assurance. research efforts have been spent on infeasible paths, basically on three main approaches: prediction, classification and identification of infeasibility. this work reports the results of experiments on data flow based criteria and of studies aimed at the three approaches above. identification, classification, and prediction of infeasible paths are revisited in the context of data flow based criteria (potential uses criteria-pu). additionally, these aspects are also addressed in the scope of integration and object-oriented testing. implementation aspects of mechanisms and facilities to deal with infeasibility are presented taking into consideration poketool - a tool that supports the application of the potential uses criteria family. the results and ideas presented contribute to reduce the efforts spent during the testing activity concerning infeasible paths.