Testing-for-trust: the genetic selection model applied to component qualification

This paper presents a method and a tool for building trustable OO components. The methodology is based on an integrated design and test approach for OO software components. It is particularly adapted to a design-by-contract approach, where the specification is systematically derived into executable assertions (invariant properties, pre/postconditions of methods). This method, based on test qualification using fault injection (mutation analysis), also leads to contract improvements. We consider a component as an organic set composed of a specification, a given implementation and its embedded test cases. The testing-for-trust approach, using the mutation analysis, checks the consistency between specification implementation and tests. It points out the tests' lack of efficiency but also the lack of precision of the contracts. As an advanced way of enhancing the test set, the application of a genetic algorithm is presented as complementary of mutation analysis for test qualification. In this paper, the genetic selection model, a "darwinian" analogy, is thus used for the problem of efficient test selection. The feasibility of component validation by mutation analysis and its usefulness for test generation are studied as well as the robustness of trustable and self-testable components into an infected environment.