HOTL: Hypotheses and observations testing logic

Abstract

"To ensure the conformance of an implementation under test (IUT) with respect to a specification requires, in general, the application of an infinite number of tests. In order to use finite test suites, most testing methodologies add some feasible hypotheses about the behavior of the IUT. Since these methodologies are designed for considering a fixed set of hypotheses, they usually do not have the capability of dealing with other testing scenarios where the set of assumed hypotheses varies. In this paper we propose a logic to infer whether a set of observations (i.e., results of test applications) allows one to claim that the IUT conforms to the specification if a specific set of hypotheses (taken from a repertory of hypotheses) is assumed. We show the soundness and completeness of our logic with respect to a general notion of conformance.