Translating APL, an empirical study

  • Authors:
  • M. A. Jenkins

  • Affiliations:
  • -

  • Venue:
  • APL '75 Proceedings of seventh international conference on APL
  • Year:
  • 1975

Quantified Score

Hi-index 0.00

Visualization

Abstract

The wide acceptance of APL\360 and other APL systems as computational tools indicates the usefulness of APL notation for expressing solutions to computational problems. This paper discusses techniques for translating APL to fixed object programs and describes an empirical study of the gains to be made in execution speed of APL programs by such translations compared to direct interpretive execution. The purpose is to determine how great the inefficiencies in interpreting APL are and how much can be recovered by using a translation (or compilation) approach. As a byproduct of this research we hope to produce an APL translator which complements an existing APL interpreter, in that it will be possible to generate an equivalent ALGOL program from an APL workspace with minimal effort. No attempt has been made to produce a hybrid system which will allow a mixture of interpretive and compiled code. Ashcroft [2] considers some of the implications of such system.