Product Measurement: Code

Abstract

Developing measures of software (code) quality has been a long-term challenge in computer science and software engineering. A survey of articles in the literature reveals that many metrics are available for measuring software. Some well documented metrics include Halstead’s Elements of Software Science (1977), McCabe’s Cyclomatic Number (1976) and Henry and Kafura’s “Software structure metrics based on information flow” (1981). A major criticism of using any one of these metrics in isolation is that it provides a snapshot of only one particular aspect of code quality. For example, McCabe’s Cyclomatic Complexity reports the number of paths through a program. While this particular characteristic is certainly related to program quality, it says nothing about the many other contributors to code quality, e.g. intermodule coupling, the use of structured programming constructs, code commenting, and so forth. Even if one employs a collection of existing metrics, each developed in relative isolation and narrowly focused on particular characteristics, drawing a conclusion from disparate measures is indeed difficult.