基于类间依赖集约简的类簇测试用例生成算法

Generating test cases of cluster-level based on classes dependencies reduction

针对类簇中类间依赖关系以及类间消息交互产生的错误而导致测试冗余和测试遗漏的问题,提出一种类间依赖集约简的类簇测试用例生成算法。形式化定义了类间继承依赖、聚集依赖和关联依赖这3种依赖关系,建立类簇测试模型。定义类簇模式和用户(领域专家)解释集,并在用户解释集中依据每一个类对象的若干实例广义相等建立等价类,得出用户解释集幂集的1个子集即匹配划分集。据匹配消去算法求出类的最大匹配集,证明该集合就是所求类的最大非平凡非依赖集;提出一种交互差组合约简算法,计算类的最大非平凡非依赖集的补集,约简得到类的最小交叠集,并证明其就是类的最小非平凡依赖集,即类的组合测试用例;对其中每一个类的组合元素进行组合测试,并与其余类交互测试,得出类的交互测试序列。

To resolve the problem of redundancy and pretermission in the cluster-level testing due to errors of classes dependency relationship and classes message interaction, an algorithm for generating test cases of cluster-level based on classes dependencies reduction was proposed. Three classes dependency relations, i.e., inheritance dependency, aggregation dependency and association dependency, were defined by using the method for formalization, and a test model for cluster-level was constructed. An equivalence relationship in user interpretation set was built by generalized equation among some objects that were instantiations of the corresponding class. An algorithm named matching elimination was presented to obtain the maximal matching set of one or more classes equal to the maximal independency set of the corresponding class. An approach named interaction difference combination reduction was presented for computing the complement set of the maximal independency set of a class to get the minimum overlap set of the corresponding class, whose equivalency with the minimum dependency set of the corresponding class was proved too, and combination test cases and the interaction test sequences among classes were obtained.