Validating Programs in Concurrent ML

Abstract

Interactive and embedded applications often require an ability to deal with concurrent operations. In this chapter we give an overview of one way to extend a functional language with concurrency primitives. Unfortunately, the presence of concurrency may make programs harder to reason about because the flow of control is less perspicuous (as is the case when imperative features are present). This calls for the use of automatic techniques to extract the flow of control in a more readable and succinct form so as to validate the overall structure of communication taking place in the program. In this chapter, we show how type and effect systems can be extended with a notion of causal behaviour (in the manner of process algebras) that makes this possible.