An assessment of the number of paths needed for control flow testing
Nicos Malevris · 1997
Generating test data does require a great amount of effort. The estimation of this effort beforehand is difficult and unclear when generating test data for exercising certain program structural characteristics. Methods that try to generate test data based on a minimum number of paths fail to estimate a lower bound in advance. When some of the paths they generate are infeasible, this situation is worsened. Thus, the generation of the test paths is important due to the presence of infeasible paths. Knowing the effort involved in deriving the appropriate test paths is a very useful exercise. This effort is linked directly to the estimation of the effort involved in achieving a high level of reliability. In this paper methods that do suggest a lower bound and an experimental upper bound are presented. Their ability of circumventing the incidence of infeasible paths is discussed and the lower and upper bounds of the cost and effort for test data generation analysed.