Final Session Summary: H o w to Evolve Exception Handling in A d a Chair: Alan Burns Rapporteur: Alexander Romanovsky 1. Agenda The agenda of the final session included: ¢ a discussion on improving the ways of using exception handling in modern application development; ¢ a brief analysis of the following two topics: exception handling in systems with concurrency and object-orientation; exception handling in distributed systems; ¢ a wrap-up to discuss possible Adz Interpretations that can be developed further in the nearest future. do not usually use exception handling for fault tolerance but use it mainly for debugging. After a brief discussion we all agreed that dealing with abnormal events usually means tolerating faults. For example, switching modes, degradation, safe stop are typical examples of fault tolerance. Many workshop participants have agreed that it is bad practice to use exception handling for debugging (this can be only explained by the absence of good debuggers) but not to use it for tolerating faults. It is clear to us that it is not practically possible to rid software of all possible faults (including software and hardware ones, operators' mistakes, etc.) while developing complex modem applications. Tullio Vardanega gave a brief
/lp/association-for-computing-machinery/how-to-evolve-exception-handling-in-ada-j7qjmpMVUV