Domesticating imperative constructs so that they can live in a functional world

Abstract

Many problems in program analysis and transformation are simplified in the context of pure functional languages. The desire for efficiency and notational flexibility often leads, however, production versions of functional languages to include ad-hoc imperative extensions that defeat program analysis. This paper shows how a hierarchy of restricted imperative constructs can be introduced into a functional language. These constructs preserve a local impurity property where the detrimental effects of non-functional constructs are limited to the functions in which they appear. The properties of the language are investigated in the context of a formal semantics.