Abstract:

The issue of random testing of the program in the aspect of comparing the effectiveness of random testing and partitioning testing was considered in this paper. However, unlike other works on this subject, the term partition does not refer directly to a specific subset of the all program testing cases, but to a fragment of the program, which is a specific set of paths, leading from the beginning of the program to its end, where each path belongs to exactly one of the selected subsets of paths. A formal model of the program under the testing is presented. This model is based on the notion of a control flow graph. The relationship between faults and failures is characterizes by means so called the characteristic matrix of the program under testing. The main result of the paper is obtaining formulae for the mean value of the number of faults encountered during the random testing and program partition testing . The formulation of the single-criterion optimization task which allows the determination of a testing strategy that maximizes the expected value of the number of program faults is also formulated. Considerations are illustrated with a simple numerical example.