Action systems in incremental and aspect-oriented modeling

Abstract.

Action systems were first introduced as an execution model that made it possible to model distributed systems at a high level of abstraction and to refine these models into implementation descriptions. This paper describes how these initial ideas have developed into a formally based specification and design method for reactive systems, and discusses some related research. In this approach, TLA is adopted as programming logic, and action systems are extended with various language facilities. Systems are modeled as closed systems, which makes effective interface refinement possible. The associated design method is based on refinement by superposition, and the resulting specifications are layered structures, in which aspect-orientation is supported by the possibility to compose independent refinements of common base systems. A new kind of analysis of observations of concurrent executions shows that TLA-based fairness assumptions can be used also in high-level abstractions of distributed systems.