Generating non-conspiratorial executions

Abstract

Avoiding conspiratorial executions is useful for debugging, model checking or refinement, and helps implement several wellknown problems in faulty environments; furthermore, avoiding non-equivalence robust executions prevents conflicting observations in a distributed setting from occurring. Our results prove that scheduling pairs of states and transitions in a strongly fair manner suf-fices to prevent conspiratorial executions; we then establish a formal connection between conspiracies and equivalence robustness; finally, we present a transformation scheme to implement our results and show how to build them into a well-known distributed scheduler. Previous results were applicable to a subset of systems only, just attempted to characterise potential conspiracies, or were tightly bound up with a particular interaction model.