Testing for imperfect integration of legacy software components

Abstract

In the manufacturing domain, few new distributed systems are built ground-up; most contain wrapped legacy components. While the legacy components themselves are already well-tested, imperfect integration can introduce subtle faults that are outside the prime target area of generic integration and system tests. One might postulate that focused testing for integration faults could improve the yield of detected faults when used as part of a balanced integration and system test effort. We define such a testing strategy and describe a trial application to a prototype control system. The results suggest that focused testing does not add significant value over traditional black-box testing.