Distributed software testing with specification

A specification-based testing method aimed primarily at detecting errors of distributed systems is presented. This specification technique emphasizes the communication behaviors and the synchronization properties of distributed processes. The functions of behavior specification for the testing method serve two purposes. First, it provides the communication-oriented high-level view of the system, which dictates the method of test case generation. Five test criteria peculiar to distributed software are considered. Test data can be derived from the source code based on the symbolic execution technique applied to each test case. Second, the specification serves as a correct reference model, which helps analyze the communication events of the system execution. Currently, this testing method, as an important support to the INTEGRAL verification and validation system, has been largely realized in a SUN workstation environment