Reactive SystemsOrganizationToolsMethodology

Methodology

Topic:
behavioural specifications; relation between specifications and programs
Coordinator:
Michel Bidoit (Cachan) [previously Marie-Claude Gaudel (Orsay), Andrzej Tarlecki (Warsaw) ]
Mailing List:
cofi-methodology
Notes Series:
M
Case Studies Pages:
http://www.pst.informatik.uni-muenchen.de/~baumeist/CoFI/index.html, mirror

The original motivation for work on algebraic specification was to enable the stepwise development of correct software systems from specifications with verified refinement steps. Casl provides good support for the production of specifications both of the problem to be solved and of components of the solution, but it does not incorporate a specific notion of refinement. Architectural specifications go some way towards relating different stages of development but they do not provide the full answer.

Other methodological issues concern the `endpoints' of the software development process: how the original specification is obtained in the first place (requirements engineering), and how the transition is made from Casl to a given programming language. Finally, the usual issues in programming methodology are relevant here, for instance: verification versus testing; software reuse and specification reuse; software reverse engineering; software evolution.

Casl has been designed to accommodate multiple methodologies. Various existing methodologies and styles of use of algebraic specifications have been considered during the design of Casl to avoid unnecessary difficulties for users who are accustomed to a certain way of doing things. For example, the methodology in [50] is being adapted to Casl.

The major task at the moment is the production of a user's guide for Casl. Moreover, various case studies are to be coordinated within this task group.


CoFI : CoFI -- Version:  -- November 29, 2004.
Comments to pdmosses@brics.dk

Reactive SystemsOrganizationToolsMethodology