Steps on the Road to Component Evolvability

We have recently developed a calculus for dynamically evolvable aggregations of components. The calculus extends CCS with primitives for describing components and their evolvability capabilities. Central to these novel primitives is a restricted form of higher-order communication of processes involved in update operations. The origins of our calculus for components can indeed be traced back to our own previous work on expressiveness and decidability results for core higher-order process calculi. Here we overview these previous works, and discuss the motivations and design decisions that led us from higher-order process calculi to calculi for component evolvability. Introduction. The deployment of applications by the aggregation of elementary blocks (modules, components, Web services, ...) is a long-standing principle in software engineering. Our interest is in the correctness of aggregations of components which are subject to evolvability and adaptation concerns. The term “component” is used here in a broad sense, as it refers to elementary blocks such as Web services in cloud computing scenarios, but also to analogous concepts in different settings, such as services in service-oriented computing or long-running processes in workflow management. To this end, we have recently defined E , a process calculus equipped with primitives for describing components and their evolvability. Using E as a basis, we have studied the decidability of verification problems associated to the correctness of aggregations of components [2]. In this short paper, we present E and discuss the origins and motivations that led to its definition. In particular, we elaborate on the relationship between the notion of component evolvability in E and higher-order process calculi. Steps Towards Specification Languages. Higher-order process calculi are calculi in which processes can be passed around in communications. Higher-order (or processpassing) concurrency is often presented as an alternative paradigm to the first-order (or name-passing) concurrency of the π-calculus [8] for the description of mobile systems. As in the λ-calculus, higher-order process calculi involve term instantiation: a computational step results in the instantiation of a variable with a term, which is copied as many times as there are occurrences of the variable. The basic operators of these calculi are usually those of CCS [7]: parallel composition, input and output prefix, and restriction. Replication and recursion can be encoded. Proposals of higher-order process calculi include the higher-order π-calculus [10], Homer [5], and Kell [11]. ? Supported by the EU integrated project HATS, the Fondation de Coopération Scientifique Digiteo Triangle de la Physique, and FCT / MCTES (CMU-PT/NGN44-2009-12) INTERFACES