Error Handling Based on Software Agent Technology

In order to build complex software systems that are highly reliable, it is essential that the software conform to a rigid structure that bounds the types of errors that can occur at each point in the software. IAI has spent nearly a decade developing complex software systems based on autonomous software agents. In these systems, each agent has its own thread of control, and interacts with other agents via message passing. This structure is particularly well suited to advanced error detecting and error handling methods because errors associated with the interaction between agents can be separated from the errors related to each individual agent. Additionally, any errors caused by side effects or by data intended to be used by more than one agent must occur through the content of explicit messages. We have developed a rigid structure for these agents and within that structure, have developed a tool that proves the correctness of the design against a set of important errors. However, at this point the tool has no services to assist in debugging, and no services to support run time error checking or error handling. The work herein proposed will develop these features for our agent-based software CASE tool.