Adding multidimensionality to procedural programming languages

Abstract

One of the most serious shortcomings of multidimensional languages is their inability to collaborate with conventional programming languages and systems. Multidimensional languages are used to define (potentially infinite) streams, grids, cubes, and so on, concepts which resemble in nature the familiar imperative arrays. The main difference is that the former entities are lazy while the latter are generally eager. This paper proposes the embedding of multidimensional languages into conventional ones as a form of definitional lazy arrays. The paper describes the details of an implementation of the proposed idea, as well as the expressibility and the performance of the resulting system. The main advantage of the new approach is that multidimensional languages can now benefit from the advanced features that have been developed for conventional languages. Moreover, multidimensionality adds to conventional languages the idea of lazy arrays, which in many cases offer significant advantages compared to the classical imperative arrays. Copyright © 1999 John Wiley & Sons, Ltd.