A Finite Domain-Testing Strategy for Computer Program Testing.
Abstract
Program testing continues to be a practical approach to software validation, however the strategies currently being used lack a solid analytical foundation. The goal of this research is to analyze the program testing process, develop a strategy to maximize its effectiveness, and identify its limitations. A program can be viewed as a complex mapping from a N-dimensional space of input variables to an M-dimensional space of output variables. In the testing process the correctness of the program over a domain of this input space is inferred from its observed correctness on a small set of 'well-chosen' test values for that domain. The Domain Testing Strategy is used to determine the necessary set of test values and is shown to be successful for all types of errors except a small subclass called 'Missing Path Errors of Reduced Dimensionality'. The domain testing Strategy is developed for both continuous and discrete input spaces and for both linear and nonlinear predicates. The domain testing strategy is developed for both continuous and discrete input spaces and for both linear and nonlinear predicates. The only completely effective testing strategy is an exhaustive test which is totally impractical. The domain testing strategy offers a major reduction in the high cost of computer program testing with a minimal loss of testing effectiveness. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1977
- Accession Number
- ADA054771
Entities
People
- Edward I. Cohen
- Lee J. White
Organizations
- Ohio State University