S. Nakajima, S. Takada, and N. Doi (Japan)
software testing, coverage criteria, spanning sets
Executing all possible test cases during software testing is not realistic due to the number of test cases, meaning we need to select test cases. The quality of the selected test cases (as well as their adequacy) is often measured with coverage, i.e., how much of the program has been tested. Many research on test case selection are based on maxi mizing coverage while keeping the number of test cases to a minimum. However, these research often focus too much on coverage and not enough on detecting faults, which is the primary goal of testing. This paper proposes a method to maximize coverage and faults detected, while keeping the test cases as small as possible. We use spanning sets to maximize coverage and focus on boundary values in condi tional statements to increase the number of detected faults. We show details of our approach as well as an evaluation. By incorporating the boundary values, the number of de tected faults rose 1.41 times when compared with one not using boundary values.
Important Links:
Go Back
