Hybridizing Evolutionary Testing with the Chaining Approach

Fitness functions derived for certain white-box test goals can cause problems for Evolutionary Testing (ET), due to a lack of sufficient guidance to the required test data. Often this is because the search does not take into account data dependencies within the program, and the fact that some special intermediate statement (or statements) needs to have been executed in order for the target structure to be feasible. This paper proposes a solution which combines ET with the Chaining Approach. The Chaining Approach is a simple method which probes the data dependencies inherent to the test goal. By incorporating this facility into ET, the search can be directed into potentially promising, unexplored areas of the test object’s input domain. Encouraging results were obtained with the hybrid approach for seven programs known to originally cause problems for ET.