Measuring software structure has many appealing applications, including maintenance activities and system re-engineering. Giving meaning to software measurements is often difficult because it is not always easy to relate a software component to a number.

This paper characterizes well-known class inheritance metrics, with inheritance being a fundamental relation in object-oriented programming languages. This characterization is based on the seminal framework provided by Briand et al. [1]. Krishna and Joshi innovate by introducing the notions of viewpoint and projection, two essential ingredients for the characterization. Their analysis reveals that classical metrics, such as number of inherited attributes and number of inherited methods, are size metrics. They find that other metrics, such as depth of inheritance of a class, follow length properties. As a consequence, inheritance metrics do not satisfy the symmetry property.

Even though this paper is well written and easy to read, solid knowledge of metrics will help readers to better understand details, such as the way it instantiates the framework of Briand et al. This paper presents a deep analysis, making it worth reading.