基于执行剖面过滤的分割测试

Software Partition Testing Based on Execution Profiles Filtering

提出了一种基于执行剖面过滤的分割测试方法.该方法通过从大量的执行操作中过滤出包含容易引起错误输出的特殊执行操作的子集,并对子集的执行操作进行回放和检验,从而可以发现软件潜在的错误.从而使引起错误输出的输入元素集中分割在某些子域内,提高了发现错误的概率.实验结果表明,该方法分割错误元素的集中度以及命中错误的概率较高,相同条件下其效果要优于随机测试.

This paper proposed a new partition testing method based on execution profiles filtering. It startd with a large universe of executions filtered to find subsets, which consisted of unusual failure-causing execution behaviors. These executions were then replayed and checked according to requirements to find faults. This method increased the probability of finding faults because most failure-causing inputs were partitioned in certain subsets. Experiment results indicated that this method had relatively high fault-partition concentration and fault-finding probability, and its effectiveness was better than random testing in the same conditions.