ITR/SY(CISE): Compositional Connectors

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

NSF ITR Proposal #0113810 Compositional Connectors David Garlan For systems composed from independently developed parts, specialized forms of interaction are often needed to bridge component mismatches or to achieve extra-functional properties (e.g., security, reliability), making the design and implementation of these interaction mechanisms a critical issue. Unfortunately, system developers have few options: they must live with available, but often inadequate, generic support for interaction (such as RPC), or they must handcraft specialized mechanisms at great cost. This research investigates a new approach whereby interaction mechanisms are constructed compositionally. Specifically, basic connectors (such as RPC or data streams) can be augmented with selected adaptations or enhancements to produce a more complex connector that meets the system requirements. This work will investigate the hypothesis that (a) there exists a collection of general-purpose transformations that can be applied to connectors to produce increasingly rich forms of interaction, and (b) tools can be built to automate the application of these transformations to existing interaction mechanisms. If successful, this research will reduce the cost of component integration by partially automating the production of complex interaction mechanisms, improve the quality of systems by making it simple to add security and dependability features, and develop new foundations for specifying and reasoning about complex interaction mechanisms.