A new method of software structural testing by path analysis
Hiroshi Masuyama, Tetsuo Ichimori · Systems and Computers in Japan · 1986
Abstract This paper discusses the structural software testing method by path analysis. The problem of the method is discussed, and a coverage measure is proposed. The problem in the structural testing is discussed first. If the evaluation is made based on the past coverage measures, the quality of the software tends to be overestimated. A problem occurs of testing the cost when the number of steps for determining the test data increases with the size of the software. Also, another problem occurs wherein if a new coverage measure is introduced so that the quality is not overestimated, additional test cost is produced. From such a viewpoint, this paper proposes a method which can essentially avoid these problems by using the minimum number of test paths. An algorithm for determining the test paths is presented which is independent of the earlier method. Then it is shown that the number of test paths determined by that algorithm is the minimum. The total distance of the test paths is also discussed. Finally, the computational complexity in determining the test paths is discussed.