TinyCORP
Tetsuo Kamina, Tomoyuki Aotani · 2019
Current trend of seamless connections between computing systems and their surrounding environments requires software to be more reactive and adaptable, and reactive programming (RP) and context-oriented programming (COP) have been studied to directly support reactive behavior and dynamic adaptation. Sometimes reactive behavior and dynamic adaptation interact with each other. One issue of such interactions is how to avoid a loop of reactive behavior and dynamic adaptation when there are mutually recursive dependencies between them. This paper proposes TinyCORP, a core calculus for context-oriented reactive programming that is designed in a main-stream, general-purpose language setting. This calculus is expressive enough to represent both features of signals (i.e., time-varying values in RP) and layer-based partial methods in COP, and their interactions including the ability to specify the mutually recursive dependencies between dynamic adaptation and reactive behavior. We also demonstrate that the computation in TinyCORP do not result in the loop of reactive behavior and dynamic adaptation.