Contextual Constraints – Breaking The “To Constrain Or Not To Constrain” Dilemma

Abstract

Software architectures are developed incrementally, and the development involves experimenting with a variety of components, interaction patterns and styles. Composition, substitution and refinement are three important operations that are performed during these design experiments. ADLs provide constructs to express composition, substitution and refinement. It is important to be able to specify that “X is composed with Y,” but the fact “why X is composed with Y” is much more important and valuable, for they serve as the (only) guide line for the evolution of the system. Such design decisions are captured as constraints in the architecture specifications. Every design decision, specified as a constraint, is tightly bound to its context, and is often applicable only in that context. To be able to specify the context of a constraint at the finer granularity of operations, it should be possible to specify the operational context in which the constraint is applicable. In this paper we present contextual constraints, as supported by our ADL TriSL, which permit tagging of constraints with operational contexts. TriSL blends the notion of type conformance with contextual constraints to ensure that the constraints are checked when (and only when) the corresponding design operations are performed.