Measuring Behavioral Software Design Patterns

Abstract

Large software systems can be made reliable by design patterns reuse. But there is a lack of ways to assure implementation correctness of the patterns themselves. We propose run-time measurement as a systematic verification approach for software design patterns. It is shown that standard state probabilities have a stable asymptotic behavior, allowing safe identification of standard patterns. The software measuring device ¿ called "patternometer" ¿ is non-intrusive and generic. This is possible due to an aspect-oriented implementation. Case studies ¿ as Observer and Mediator - demonstrate the approach.