Fault Classes and Error Detection in Specification Based Testing

摘要

Specification based testing relies upon methods for generating test cases frompredicates in a software specification. These methods derive various test conditions from logic expressions, with the aim of detecting different types of faults. Some authors have presented empirical results demonstrating their effectiveness of the test generation methods. This paper examines the conditions under which a particular fault class will cause an error for a given predicate. These conditions must be covered by a test set for the test set to guarantee detection of the particular fault class. By deriving the general conditions under which various fault classes cause an error, the authors show that there is a coverage hierarchy to fault classes. The fault hierarchy is then used to explain experiemntal results on fault based testing. This work is significant because it provides a method for comparing the effectiveness of test sets designed for various fault models.