Software Testing Based on Formal Specification

It is well accepted that formal specifications can be useful bases for software testing; some pioneering papers come back to the seventies. Given a specification and a system under test, any testing activity is, explicitly or not, based on a satisfaction relation (often called conformance relation): does the system under test satisfy the specification? Tests are derived from the specification on the basis of the satisfaction relation, and often on the basis of some additional knowledge of the system under test and of its operational environment called testability hypothesis. The verdict about the success or not of a test execution depends on the observations that can be made on the system under test, and it is based on the satisfaction relation. This course presents a generic framework for developing testing methods based on formal specifications, and its specialization to several formalisms: Finite State Machine, Algebraic Specifications, Input-Output Transition Systems and Transition Systems with Priorities. After presenting some motivation and context, Section 2 illustrates the principles sketched above in the case of algebraic specifications, for which they were developed originally. Section 3 goes back in time and slightly reformulates the work on testing based on FSM in this new framework. Section 4 presents Input-Output labelled Transition systems (IOTS), the associated ioco conformance relation, defined originally in [318], and extended to JOTS with data types in [209]. Section 5 discusses problems that arise when introducing some notion of priority between inputs and outputs, extending IOTS models and the ioco conformance relation.