Mutation-optimised subdomains for test data generation and program analysis

Software testing is an important part of the development process it consumes a large proportion of the labour resources required to produce a working program. Yet it is not usually possible to show that a program is completely free from faults. Instead, techniques are applied to assess the effectiveness of software testing; they provide confidence in its adequacy and act as a benchmark for its improvement. One such technique (mutation analysis) uses small changes in the program code to simulate actual faults. Mutation analysis has been shown to be more stringent than other testing techniques and a good predictor of the real fault-finding capability of a test suite. This thesis introduces new techniques for identifying, evolving and selecting input subdomains that can be sampled at random to produce efficient test suites which achieve a high level of mutation adequacy, and so are expected to be efficient at finding faults. Previous research into software testing has focussed on producing suites of individual test cases. This thesis represents the first attempt to optimise subdomains for each parameter to the program under test. The resulting subdomains can easily be comprehended by a human test engineer, so may be used to provide information about the software under test and design further highly efficient test suites.