Pitfalls of Jason Concurrency

Abstract

Jason is a well-known programming language for multiagent systems where fine-grained concurrency primitives allow a highly-concurrent efficient execution. However, typical concurrency errors such as race conditions are hard to avoid. In this chapter, we analyze a number of such potential pitfalls of the Jason concurrency model, and, describe both how such risks can be mitigated in Jason itself, as well as discussing the alternatives implemented in eJason, an experimental extension of Jason with support for distribution and fault tolerance. In some cases, we propose changes in the standard Jason semantics.