An extensible metrics extraction environment for object-oriented programming languages

In this paper the enhancements to an existing production-quality compiler to facilitate the collection of complexity metrics as an integral part of the compilation and build process are described An object oriented programming language meta-model is presented. This meta-model is used in a relational database schema and enables a language neutral form of a compilation unit of source code or system of units to be stored We have used C language routines with embedded SQL to calculate a suite of 11 complexity metrics for each class stored in the database. We present the full implementation of one of these routines. The resulting metrics extraction tool is highly extensible, readily facilitating incorporation of new object oriented programming languages and new metrics.

1. Introduction

A software complexity metric is intended to provide an indicator of the quality of a software system design or a software system implementation. For example, a metric might attempt to indicate the degree of fault-proneness [1], [2], the impact of software changes in software maintenance [3], , [5] or the ease of change in program evolution [6].