Enriching Component Interfaces with Checkable Dependence Specifications

Component middleware frameworks such as the CORBA Component Model (CCM) and Enterprise Java Beans (EJB) are increasingly being applied to address challenges involved in building large-scale distributed systems. We seek to provide foundations and tools for modeldriven development of such systems in which architectural models serve as highly leveragable abstractions that form a substrate into which a variety of forms of essential behavioral properties are woven. Our vision emphasizes a synergistic integration of analysis and analysis clients at multiple levels of abstraction in a system including both the architectural model (component interfaces and connections) level and the source code (component implementation) level. Dependence and flow properties have proven to be useful behavioral abstractions that can be leveraged in a variety of ways at both the architectural model level and code level. However, previous works provide no mechanism for connecting architectural-level dependences to codelevel dependences, e. g., for the purpose of guaranteeing soundness of model-level specifications by checking that dependences present in implementations conform to model-level specifications – a task that is especially challenging when working with sophisticated component frameworks like CCM. In this paper, we propose a layered approach to enriching component interfaces specifications to include a variety of forms dependence and flow information. We have implemented conformance checks between layers of component specifications using a flexible Java dependence analysis framework, and we report on our experiences applying the framework to a collection of CCM component specifications and implementations. ∗This work was supported in part by the U.S. Army Research Office (DAAD190110564), DARPA/IXO’s PCES program (AFRL Contract F33615-00-C-3044), NSF (CCR-0306607, CCF-0429149, CCF-0444167), and Lockheed Martin. †Manhattan KS, 66506, USA. {rvprasad, jung, hatcliff}@cis.ksu.edu ‡Lincoln, NE, USA. [email protected]