Executable semantic descriptions

Abstract

Abstract We describe how the denotational semantics of a programming language can be executed directly, if it is expressed in a suitable functional programming language such as ML. We also apply Mosses' idea of ‘semantic algebras’ to construct semantic descriptions that are significantly more modular and understandable than usual. The possibility of executing a language's semantic description directly supports a methodology of language design that we advocate: express the design as a formal language description, and use this to test and refine the design, before becoming committed to constructing a compiler.