A framework for parallel unit testings: work in progress

Abstract

The power of parallel computing needs to be exploited on software testing because of challenges that remain today, such improving test effectiveness and automating test input generation. There has been research on applying parallel computing to software testing, but there are still areas where the power of parallel computers can be exploited. Another challenge which has remained due to the lack of tool support is the oracle problem. We want to be relieved of tedious and manual processes in software testing. This paper proposes a framework for parallel unit testing. To alleviate the oracle problem, we will use the Java Modeling Language as the test oracle. We will take advantage of the power of parallel computing and apply it to random unit testing of Java classes to overcome some of today's challenges. We believe that random testing can be used to help us achieve this goal because it is effective and also cost-effective. We will show that we can overcome these challenges by generating and executing more test cases in the same amount of time, and by implementing data diversity. Our framework will be extensible to support additional programming languages and technique diversity.