A study on test cases generation for object-oriented programs based on UML state diagram.

Software testing expenses are estimated to be between 20% and 50% of total development costs. Software testers need methodologies and tools to facilitate the testing portion of the development cycle. State-based testing is one of the most recommended techniques for testing object-oriented programs. Data flow testing is a code-based testing technique, which uses the data flow analysis in a program to guide the selection of test cases. Both state-based testing and data flow testing have their disadvantages. State-based testing does not analyze the program code, and thus could miss the detection of data members that do not define the states of the object class. Selecting data flow test cases from data members for testing classes is difficult and expensive. To overcome their weakness, a hybrid class test model is proposed, which contains both the information from specification about the state change of object instances of the Class Under Test (CUT) and the information from the source code about the definition and use of the data members in the CUT. With such an uniformed architecture, we can obtain automated tools to generate test cases for state-based testing and perform data flow testing at the same time. The combination of the two techniques is essential in improving our testing environment, and thus contributes to the enhancement of the reliability of software products. The proposed hybrid testing strategy can be used in both software design stage and software implementation stage. A Standard-based UML information exchange format, XMI is used to describe UML design specification in the hybrid testing strategy, to bridge the software designer and software tester. No matter what kind of CASE tools designer use, as long as it is saved as XMI format, the testing tool can easily understand design specification from different design tools. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2001 .Y36. Source: Masters Abstracts International, Volume: 40-03, page: 0730. Adviser: Xiaojun Chen. Thesis (M.Sc.)--University of Windsor (Canada), 2001