Abstract
This paper is intended to solve a particular problem related to the refinement of a shared sequential buffer into a parallel collection of buffers arising from a study on the IBM CICS project. Using the notion of cooperating refinement we show that the two systems are equivalent from the users' points of view (except with respect to efficiency). This is achieved by constructing an interleaving for each possible sequence of commands which access the buffer. The induction used in the proof is non-standard, and makes the problem harder than it would at first seem. Further we show that the interleaving cannot be done ‘on the fly’, showing that in some other sense, the parallel collection is indeed superior, as intuition suggests.
References
IBM Corporation.CICS/ESA General Information. GC33-0155, 1985.
C. A. R. Hoare.Communicating Sequential Processes. Prentice-Hall International, London, 1985.
J. Jacob. Refinement of shared systems. InThe Theory and Practice of Refinement: Approaches to the Formal Development of Large-Scale Software Systems, pages 27–36. ed. John McDermid, Butterworths, London, UK, 1989.
B. M. Yelavich.Customer Information Control System—an evolving facility. IBMSJ24 2/3, 1985
Author information
Authors and Affiliations
Additional information
This author was supported by the Esprit REX project
Rights and permissions
About this article
Cite this article
Kay, A., Lupton, P. Sequential to parallel buffer refinement. Formal Aspects of Computing 4, 487–492 (1992). https://doi.org/10.1007/BF01211395
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1007/BF01211395