Verifying Properties of Coordination by Well-Structured Transition Systems

ReSpecT Nets: Towards an Analysis Methodology for ReSpecT Specifications

“…Section 5 provides some results and sketches some research directions toward the analysis of ReSpecT nets, focusing on the safety properties that can be expressed in terms of control state reachability, also known as covering [12,23]. Most notably, we show that ReSpecT specifications that do not exploit the no r (test for absence) primitive are encoded into Petri nets where transitions featuring inhibitor arcs can be simply removed without altering covering properties.…”

“…This problem can be seen in connection with the framework of well-structured transitions systems [12]: inhibitor arcs make the Petri nets formalism lose its wellstructure-in a sense, its monotonic behaviour with respect to operational semantics and inclusion of markings. As common in these situations, sufficient conditions for decidability are still worth investigating [23,3,2]. In particular, the ReSpecT programming practice [8] shows that a number of interesting and useful specifications do conserve an intrinsic well-structure, and might then in principle be analysed by a formal tool for effectively proving properties of interest.…”

“…This means that all the exiting tools for Petri nets (with inhibitor arcs) can be directly exploited to verify properties-such as safety and liveness-of ReSpecT specifications [14,5]. A particularly relevant safety property in the context of coordination is covering [23], stating that a given submarking will never be reached, which translates into the fact that a given (unsafe) configuration of tuples will never occur in the tuple centre.…”

“…By using for instance the prototype system described in [6] one could verify whether unsafe configurations are never reached from a given initial state [23]. For instance, consider the initial state go12 | c3, meaning that a client is releasing locks 1 and 2.…”