MEASURES OF TESTABILITY AS A BASIS FOR QUALITY ASSURANCE

Program testing is the most used technique for analytical quality assurance. A lot of time and effort is devoted to this task during the software lifecycle, and it would be useful to have a means for estimating this testing effort. Such estimates could be used, on one hand, for guiding construction and, on the other, to help organise the development process and testing. Thus the effort needed for testing is an important quality attribute of a program; we call it its testability. We argue that a relevant program characteristic contributing to testability is the number of test cases needed for satisfying a given test strategy. We show how this can be measured for glass (white) box testing strategies based on control flow. In this case, we can use structural measures defined on control flowgraphs which can be derived from the source code. In doing so, we bring together two well researched areas of software engineering: testing strategies and structural metrication.