The Research And Application Of Software Testability Metric Based On Semantic

A program's testability is a prediction of its ability to hide faults when the program is black-box-tested with inputs selected randomly from a particular input distribution. Testability is an important criterion to evaluate whether the software is easy to test or not, and it can also instruct the depth of testing and insure the testing resources are allocated reasonably. The semantic information of software does a great contribution to software testability. By computing the semantic information, we can judge the testability is good or not in some degree. This paper establishes the relations between testability and semantic information based on researching a lot of literature materials.First, this paper introduces the testability and the factors which affect the testability, and then analyses and compares some techniques existed about testability metric. At this basis, a new testability metric method which is based on the size of semantic fault is raised. In succession, on the base of researching the metric of hardware testability and combining with the characteristics of software testability, this paper describes two new attributes of software testability. They are about whether fault is easy to detect and locate. They are called respectively detecting ability and location ability of fault. After that, the pivotal techniques about semantic used to compute the new attributes are introduced. Aiming at one of the techniques―the dependence relation between input and output, this paper makes up the defect of coarse disposal to pointer and improves the precision.In the end, this paper describes a prototype system which is about the computing of software semantic information. On the one hand, this system can produce the semantic difference between the new and the old version of the same software. On the other hand, this system can extract all the statements which are semantic correlative with the variable in a statement. Using the two results produced by the system, we can compute the new attribute of testability.