|
It is an important to The universe and its reflection in the ideas of man have wonderfully complex structures. Our ability to comprehed this complexity and perceive an underlying simplicitry is intimately bound with our ability to symbolize and communciate our experience. The scientist has been free to extend and invent language whenever old forms became unwieldy or indadequate to express his ideas. His readers however have faced the double task of learning his new language and the new structures they describe. There has therefore arise a natural control: a work of elaborate linguistic inventiveness and meager results will not be widely read. As the computer scientist represents and manipulates information within a machine, he is simulating to some extent his own mental processes. He must, if he is to make substantial progress, have linguistic constructs capable of communicating arbitrarily complicated information structures and processes to his machine. One might expect the balance between linguistic elaboration and achieved results to be operable. Unfortunately, the computer scientist, before he can obtain his results, must successfully teach his language to one particularly recalcitrant reader: the computer itself. This teaching task, called compiler writing, has been formidable. Consequently, the computing community has assembled, under the banner of standardization, a considerable movement for the acceptance of a few committee-defined languages for the statement of all computer processes. The twin ideals of a common language for programmers and the immediate interchanibility of programs among machines have largely failed to materialzie. The main reason for the failure is that programmers, like all scientists before them, have never been wholy satisfied with their heritage of lingustic constructs. We hold that that demand for a fixed standard programming language is the antithesis of a desire for progress in computer science. That the major responsibility for computer language design should rest with the language user will be our central theme. in "An Apporach to Computer Language Design", W. M. McKeeman, PhD Thesis, Stanford University (1966) republished in "Compiler Construction, An Advanced Course", Goos & Hartmanis (ed), Springer Verlag (1974)
|