摘要：This paper considers the test case generation for distributed software (a test case contains one or more test sequences). Applying the single finite state machine (FSM) test approach to distributed software, we will suffer from some problems: 1) the state combinatorial explosion problem; 2) some unexecutable test cases may be generated; 3) some fault may be masked and cannot be isolated accurately. This paper proposed a new test case generation method based on the FSM net model. Instead of testing the global transitions of product machine, the generated test cases are used to verify the local transitions. We discuss the detailed methods of verifying the outputs and the tail states of the local transitions. Moreover, we prove that if all the local transitions are right, the transition structure of the distributed software is right. The tests are generated on the local transition structure of components, so we will not meet the state combinatorial explosion problem. All the outputs of the local transitions are checked, so the fault isolation results may be more accurate.
关键词：distributed software;finite state machine net;output identifying sequence;extended unique input/output sequence;test case generation
