Evolution Of The Static Test Analysis Technique

Evolutionary Testing is an emerging technology for automatically generating test data with good quality, and it has been successfully applied in engeneering practice. It converts the testing data generation into an overall search problem, and use the the robust ability to search the optimum in the global space of Evolutionary Algorithm, such as Genetic Algorithm, to search the test data. However, in structural testing, the search process of evolutionary testing will be hindered in the presence of inner state problem, consequently causing the failure of test data generation.In Object-Oriented Program (OOP), the presence of inner state problem is that to reach a test goal needs to call a series of methods. On the one hand, to change the data member need to execute the corresponding function member;on the other hand, the encapsulation makes that the test to non-public procedure only could be done by excuting corresponding public method interfaces. In this paper, in order to describe the inner state program, under the requirement of the dynamic search in evolutionary testing, branch dependence analysis, a new method of program static analysis, is presented. The author discusses and partitions the branch dependence relations, defines the branch dependence graph, designs it's construction algorithm. This paper also analyses the call chain and append the information of control calling brach information to the branch dependence graph. This analyse method is based on the branchand breaks through conventional program dependence analyse methods which are based on the sentence. The branch dependence graph has less information which is more fit on software testing espacialy evolutionary testing, and the algorithmic complexity decreases as it's analysis granularity largens.