A Simplification of a Conversation Design Scheme Using Petri Nets.

The conversation scheme is a promising way of developing fault-tolerant software and several mechanisms for its implementation have been proposed. In an earlier paper Tyrrell and Holding used Petri nets to design conversations. Their procedure, although correct, appears to be more complex than necessary. In this correspondence, a simplified transition identification method is proposed. Using a robot arm control program we show that the corresponding Petri net graph is simpler than the one proposed by Tyrrell and Holding, but the communication state change table is the same. It is also shown that these two methods are equivalent. [ABSTRACT FROM AUTHOR]