In Praise of APL (1977)

Summary

In Praise of APL: A Language for Lyrical Programming Professor Alan J. Perlis Yale University Many reasons can be given for teaching one or more aspects of computer science (defined as the study of the set of phenomena arising around and because of the computer) to all university students. Probably every reader of this note supports some of these reasons. Let me list the few I find most important: (1) to understand and to be able to compose algorithms; (2) to understand how computers are organized and constructed; (3) to develop fluency in (at least) one programming language; (4) to appreciate the inevitability of controlling complexity through the design of systems; (5) to appreciate the devotion of computer scientists to their subject and the exterior consequences (to the student as citizen) of the science’s development. Even though computer science deals with symbolic objects whose nature we study mathematically, it cannot be taught as an orderly development arising from a few fundamental ideas whose existence the student has already observed intuitively during his maturation, such as gravitation and electricity. It is during this first computer course that the student awakes to the possibilities and consequences of computation. They arise most usefully and in greatest profusion during his writing of programs. He must program and program and program! He must learn how to state precisely in a programming language what he perceives about the nature of symbolic systems. I know of no better way to expedite this awakening than by programming. But what should the student program? and in what language? I do not place much emphasis on heavy use of other people’s programs, packages if you will, that perform real services such as statistical packages, data management systems, linear equations solvers, etc. While it is wise to use standard programs when they match one’s needs, it is more important to master self-expression during this initial contact. Available time is a li...