Greenify: A Game with the Purpose of Test Data Generation for Unit Testing
Abstract
One of the most important, but tedious and costly tasks of software testing process is test data generation. Several methods for automating this task have been presented, yet due to their practical drawbacks, test data generation is still widely performed by humans in industry. In our previous work, we employed the notion of Game With A Purpose (GWAP) and introduced Rings as a GWAP to reduce time and costs of human-based test data generation and increase its appeal to engage even nontechnical people. In this paper, we propose a new game, called Greenify, with the purpose of test data generation so that it solves the main issues of Rings. The environment of this game is built based on a program’s control flow graph. To evaluate the proposed approach, we designed several game levels based on six different C++ programs and gave them to volunteering players. The results show that in comparison to both conventional human-based approach and Rings, Greenify generates test data with less rime for all feasible paths of the given benchmark programs. In addition, Greenify identifies the smaller set of likely infeasible paths.
Origin | Files produced by the author(s) |
---|