next up previous
Next: 2.6 Validation of Formal Up: 2 Position Previous: 2.4 Models of Components

2.5 Componentology

So what has all this to do with building distributed systems from components? It seems reasonable to suppose that many important distributed systems in the future will be built using a client-server paradigm. Especially where the behaviour of the system is to evolve dynamically over time. Having methods to describe the behaviour of such systems formally gives us the ability both to explore alternative designs and to validate chosen designs to ensure that they have the behaviour which we expect. The pi-calculus affords one such method of formal description.

Describing components as processes, and studying how combinations of components behave, provides a model which is complementary to the object-oriented paradigm. Like the object-oriented approach we can specify the individual behaviour of components and then, abstracting from that, describe the composite behaviour of systems built from those components. Such a scheme of components and rules of combination we refer to as a componentology. Within the componentology we can design components and systems. We can formulate descriptions and ask questions. And by various means we can furnish answers to those questions.



Peter Henderson
Sep. 12 1997