A Model For Specifying Rationale Using An Architecture Description Language

Besides structural and behavioural properties, rationale plays a crucial role in defining the architecture of a software system. However, unlike other architectural features, rationale often remains unspecified and inaccessible to tools. Existing approaches for recording rationale are not widely adopted. This paper proposes a simple model for capturing rationales as part of an architecture specification and attaching them to elements in the architecture. The bi-directional links between rationales and elements enable forward and backward traceability. We describe a textual architecture description language named Grasp that implements this model, and illustrate its capabilities using an example.