Property based coverage criterion
Fadi A. Zaraket, Wes Masri · 2009
Coverage metrics answer the question of whether we adequately checked a given software artifact. For example, statement coverage metrics measure how many and how often lines of code were executed. Path coverage metrics measure the frequency of execution of interleaving branches of code. In recent years, researchers have introduced several effective static analysis techniques for checking software artifacts. Consequently, more and more developers started embedding properties in code. Also, some techniques and tools emerged that automatically infer system properties where they do not explicitly exist. We hypothesize that it is often more effective to evaluate test suites based on their coverage of system properties than than of structural program elements. In this paper, we present a novel coverage criterion and metrics that evaluate test cases with respect to their coverage of properties, and measure the completeness of the properties themselves.