Distributed Exception Handling: Ideas, Lessons and Issues with Recent Exception Handling Systems

Abstract

AbstractException handling is an important feature of the tool-set that enables the building of fault-tolerant concurrent and distributed programs. While transactional distributed systems have been studied for a long time, distributed exception handling techniques are only now evolving towards maturity, especially within asynchronous multi agents or component based systems. In this paper, we review two recent proposals for distributed exceptions handling systems (DEHS), namely SaGE and Guardian, in the light of what the Erlang programming language brings to the table : native constructs for concurrency, distributedness and exception handling across processes. We expand on the merits and possible drawbacks of these systems. We advocate the need to introduce the notion of resumption, an often downplayed feature of EHSs, to modern day DEHS, in order to address the problem of multi agent systems facing the “real world”.KeywordsMulti Agent SystemException HandlingService ExecutionMethod InvocationCommon LispThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.