ESPRIT Basic Research Action 6309
FIDE II

The FIDE II Project

Work Area: Databases, Information Retrieval & Multimedia

CO-ORDINATOR

University of Glasgow
Lilybank Gardens 17
UK - GLASGOW

PARTNERS

Universitaet Hamburg, Germany
INRIA, France
Universita di Pisa, Italy
IEI-CNR, Italy
University of St Andrews, UK

CONTACT POINT

Prof. M. P. Atkinson
tel + 44/141 330 4359
fax + 44/141 330 4913
e-m: mpa@dcs.glasgow.ac.uk

Keywords persistent systems, databases, persistent languages, DBPLs

Start Date 24 July 1992
Status running

Abstract FIDE 2 will extend the work of the first FIDE project into fuller integration. Such integration will provide significant economy in the construction of sophisticated long-lived applications. This will be achieved by imposing consistency on the supporting technologies. Specifically, a consistent framework will house a consistent set of primitives supporting precise data description, persistent data management, concurrency, distribution and recovery. This will be supported via persistent object stores and used via persistent and database programming languages. Methodologies and a persistent programming environment will support persistent application system construction. The work builds on the results of FIDE 1 (3070)

AIMS

FIDE 2's research addresses the current critical situation in the construction, maintenance and operation of persistent application systems (PASs). The objective is to formulate a consistent framework for the support of the PAS construction and maintenance process, populate this with a consistent set of concepts, and show that this combination is not only sufficient for the construction and maintenance of PASs but also enables significant economies in their software construction and operation and permits the construction of more sophisticated PASs. FIDE 2 aims to encompass a wider range of the construction and operation activities than have been integrated hitherto and issues arising from longer-lived, larger-scale and more distributed systems than were considered in FIDE 1.

This objective will be met in three stages: 1) via a thorough analysis of the construction process of PASs and of the contributing concepts, technologies and components, and via a formulation of a consistent model for such applications, so that they may be constructed in the context of a much simpler, more coherent environment; 2) provide demonstrable prototype database programming language environments, database system environments and a prototype PAS workbench; 3) use these prototypes to demonstrate that a consistent and coherent Fully Integrated Data Environment (FIDE) can be achieved through technology and concepts integration, and that it yields a quantum leap in the productivity of PAS software engineers.

APPROACH AND METHODS

The FIDE 2 approach depends on the development of clearly specified concepts. These will be verified via discussion among the partners and via incorporation into experimental systems. These systems comprise database and persistent languages, persistent object stores and tool sets that enable their use. The consortium considers it essential to test the concepts in order to demonstrate that they are comprehensible to system builders and feasible to the engineer. All work in progress and final results are being published in the FIDE Research Report series, and the results will appear in the research literature.

PROGRESS AND RESULTS

The three main threads of the FIDE 2 work have all progressed significantly.

Within the programming language environment theme, important research topics were:

type systems:: foundations, algorithms, dependent types, late binding overloading, schema evolution and dynamic abstract witness types
bulk data types:: how to provide bulk types, notations for query languages and structuring of libraries
reflection
language design studies on:: more expressive data modelling mechanisms, modularisation and program composition mechanisms, and concurrency and transactions.

The database system environments work has concentrated on the comparison of the different Persistent Higher-Order Abstract Machines (PHOLAMs) used to implement the various FIDE languages and the prototyping of object stores architectures to explore areas such as performance, distribution and scalability.

The Persistent Application Systems (PAS) workbench strand has included the development of prototype workbench components and experiments in interoperability.

At the first review of the project in September 1993 a number of FIDE 2 prototypes were demonstrated. These were concerned with language development (Fibonacci); optimisation strategies and interoperability in the Tycoon environment; two different object store developments, and three prototype workbench components; a hyper-programming environment, a graphical interface for constructing Napier88 programs and a re-usable component retrieval system.

The second review of the project was held in St Andrews in September 1994 and a further series of prototypes were demonstrated. These demonstrations were: further work on Fibonacci; concurrency in Napier88; the continuing development of the two object stores (Eos and PIOS); development of the Tycoon environment demonstrating further aspects of integration and a Napier88 Programmer's Persistent Workbench.

POTENTIAL

In the short term, the primary influence of FIDE 2 will be on other research groups, who are expected to adopt its concepts and architectures in their explorations of extensions to database capabilities, programming language functionality and support system architectures. In the medium term, it will influence the development of widely used languages and the burgeoning research into persistent systems. Ultimately it will lead to new support system architectures modifying or replacing the operating systems of today. As the consistent and systematic semantics of concepts developed by FIDE 2 for building and managing long-lived systems gain adherence, the efficiency and achievements of the European software industry should be improved.

LATEST PUBLICATIONS

: Abiteboul S, Cluet S, and Milo T. Querying and Updating the File In: Proceedings Of The Nineteenth International Conference On Very Large Data Bases (Dublin, Ireland, 24th-27th August 1993), Morgan Kaufmann Publishers, San Mateo, CA (1993)
: Albano A, Bergamini R, Ghelli G, and Orsini R. An Introduction to the Database Programming Language Fibonacci In: Proc. Convegno Su Sistemi Evoluti Per Basi Di Data (Gizzeria Lido (CZ), Italy, 14th-16th June 1993) (1993)
: Connor R C H, Atkinson M P, Berman S, Cutts Q I, Kirby G N C, and Morrison R. The Joy of Sets In: Proceedings Of The Fourth International Workshop On Database Programming Languages (Manhattan, USA, 29th August-1st September 1993), Workshops in Computing, Springer-Verlag in collaboration with the British Computer Society (1993)
: Schmidt J W and Matthes F Lean. Languages and Models : Towards an Interoperable Kernel for Persistent Object Systems In: Proceedings Of The IEEE International Workshop On Research Issues In Data Engineering Ride (Vienna, Austria, April 1993), pp 2-17 (1993)
: Zezula P and Rabitti F. Object Store with Navigation Accelerator In: Information Systems 18(7) (1993)

INFORMATION DISSEMINATION ACTIVITIES

There are now more than 110 FIDE research reports published, about sixty of these relating to FIDE 2 work; many of these have also appeared in journals or been presented at conferences/workshops. FIDE technical reports are available as hard copy or electronic (some at least). A full list is available in bibtex format.

A two-day workshop was held in Pisa, September 1993, a similar one-day meeting was held in St Andrews, September 1994 and there will be a workshop associated with the final review in Glasgow on 31 August 1995.

FIDE-Related Publications

A good source of papers are the proceedings of the POS and DBPL workshops.

Other on-line copies of papers related to persistence can be found at the University of St Andrews.

THE UNIVERSITY OF GLASGOW FIDE GROUP

Susan Spence susan@dcs.gla.ac.uk

ESPRIT Basic Research Action 6309 FIDE II