Connectors conveying Software Architecture Evolution
Résumé
This paper presents our work on component-based software evolution. More precisely, we highlight how connectors enhance component-based software architecture evolution. Indeed, connectors, by their intermediary's position between components, can play an important role for automating the propagation of evolution impacts, while preserving the architecture coherence. For that, we outline connectors' characteristics that we consider as principal for aims of evolution management. We validate and exploit these characteristics on our evolution model SAEV (Software Architecture Evolution Model). To achieve this, we need to enrich the concept of connector with information about the degree of correlation and of dependency existing between components. For that, we propose to characterize the connector concept with semantic properties, namely: exclusivity/sharing, dependency/independency, predominance/non predominance, cardinality and reverse cardinality. These properties improve the semantic of connectors, increase their expressiveness and therefore enhance the evolution impact propagation process. We use them to improve SAEV.