Generating Test Data from a UML Activity Using the AMPL Interface for Constraint Solvers

摘要

Testing is one of the most wide-spread means for quality assurance. Modelling and automated test design are two means to improve effectivity and efficiency of testing. In this paper, we present a method to generate test data from UML activity diagrams and OCL constraints by combining symbolic execution and state-of-the-art constraint solvers. Our corresponding prototype implementation is integrated in the existing test generator ParTeG and generates C++ unit tests. Our key improvement is the transparent use of multiple industry strength solvers through a common interface; this allows the user to choose between an expressive constraint language or highly optimised test data generation. We use infeasible path elimination to improve the performance of test generation and boundary value analysis to improve the quality of the generated test data. We provide an industrial case study and measure the performance of our tool using different solvers in several scenarios.