A Component Calculus for Modeling the Olan Configuration Language

Components are to becoming a key concept for the next generation of software architectures because of their impact on effective software reuse, real interoperability and integration. The Olan project [18] investigates a “programming in the large” approach that allows the specification of applications as hierarchies of interacting components, glued together through connectors, which are agents specialized in communication and coordination tasks. However we face the difficulty of defining an operational semantics able to reflect the diversity of execution models involved in real applications: the coordination of active as well as passive entities, the need for various communication schemes (synchronous and asynchronous) and the encoding of complex interconnection graphs make the use of available theories difficult. Indeed, existing process calculi (as CSP[6] and especially CCS[9], -calculus [11]) offer the required abstractions such as encapsulation and process equivalences, but they rely on the fundamental assumption that agents are active, i.e autonomously able to initiate communication. Processes are also the basic structural units of these approaches. However, components, viewed as software pieces with explicit interfaces, require a notion of passive composition that allows, for instance, several components to be traversed by a same process. We introduce in this paper a calculus, named ICCS, which extends the Milner’s CCS calculus with (1) an operator for passive composition, and (2) selective interactions. While preserving the powerful theory of process equivalences established for CCS, this calculus provides an operational definition of passive components