Extending Halstead's software science for a more precise measure of APL

Abstract

Software engineering is a growing part of computer science. Software engineering attempts to put measurements and metrics into the area of software programming. One of these metrics, Halstead's software science [1], measures a program's difficulty, a language level, and in turn, a programmer's productivity. Software science uses an extensive measure as opposed to the intensive measure of most other software engineering techniques. This is accomplished by counting functions and operands, and applying several formulae modified from thermodynamics. Even though software science attempts to place an objective measurement on these parameters, it is not meant to be a definitive decision tool, but rather a general guideline. This paper will explore the possible extension of Halstead's theory to include operators, as defined in APL, in the counting methodology and software science formulae.