Abstract

At the present time, even for safety-critical applications, it is very difficult, if not impossible, to produce a software that is "completely error-free". One of the important issues associated with this realistic situation is how to minimise the number of errors in a given software. Effective testing of software using trusted CASE tools is one possible strategy. This thesis discusses the development of a prototype CASE tool, called Apollo, that automates some of the "tedious, complex and error-prone" manual activities that are associated with the unit testing of software modules. The input to Apollo is a design specification document where the design is specified using a tabular notation. This specification is sufficiently detailed to enable execution by a machine. The tool generates a set of test cases and the anticipated test outcome for each test case by executing the tabular specification. Tabular specification is considered as a "practical" formal method, since it is a method that software developers can easily learn and apply without much mathematical background. The tabular design specification is parsed and test cases are generated based on the boundary value analysis. The anticipated test outcome for each test case is generated by executing the parsed design specification. The proposed methodology is applied to a hypothetical case study for unit testing of software related to nuclear industry. This application replaces some of the manual generation of test cases and anticipated test outcome thereby reduces the cost of software testing