Selecting Test Cases for Object Oriented Programs

Testing is used to assess the correctness of a program by running the program against a set of test data. The main problem in testing is that only a tiny fraction of the possible inputs of a program can be used to test it. Test selection techniques choose the test cases using models of the program and the bug distribution. Many different test selection techniques have been proposed. Each test selection technique has its own strengths and weaknesses. No one technique can be used by itself. Rather, an effective testing methodology must use several different techniques together. A framework is needed to relate the strengths and weaknesses of different test selection techniques, and understand how to combine different ideas and techniques. There is a real need to understand the relation between testing techniques at different levels of abstraction, the unit, integration and system levels. The unit level has the greatest ability to control the execution and observe faults of the unit. However, it does not give information about the correct behaviour of the system as a whole. The system level can give this information, but because it has little ability to control the execution, it leaves much code untested. Another area of need is in developing new testing techniques for object-oriented programs. Object-oriented programs are structured differently than non-objectoriented programs, so they have different program models, and different distributions of bugs. Much work is being done to extend traditional testing techniques to object-oriented programs. Object-oriented programs have a simple relationship between behaviour of units and the system: units are objects, and the system is composed of communicating objects. The focus of my work is to take advantage of the object-oriented structure to relate the coverage of different levels of testing. The goal is to obtain better quality of testing overall.