Coroutines and networks of parallel processes

Overview & Synopsis

The concept of coroutine or process is useful in a large class of applications, usually involving incremental generation or transformation of data. We present a language based on a clear semantics of process interaction, which facilitates well-structured programming of dynamically evolving networks of processes. These networks exhibit the same input /output behavior whether they are executed sequentially or in parallel. Sample program proofs are used to illustrate the benefits of the language's simple denotational semantics. The language serves also to clarify the relationships between coroutines, call-by-need, dynamic data structures, and parallel computation.