Abstract
The problem of measurement in software engineering has been addressed by many authors, and one of the most common questions is, “Can we learn from measurement in physics and can we transform this to software engineering measurement?” (Zuse, 1997, 1998). An engineering measure, from the physical point of view, is relevant if it can quantify the object under measurement, since qualitative measures usually are considered too coarse. The problem arises when we shift this concept in software engineering, where nearly all the possible measurements are qualitative ones. All of the characteristics defined by ISO9126 (1991) are qualitative and not directly related to precise physical or tangible phenomena. On the other hand, a wide skepticism of using numerical values is diffused, because it can be hard to give the requested semantic to the number. Metrics, that are the measures performed on code, can be split into two categories: metrics for software complexity/size measurement and effort estimation, and metrics for qualitative characteristics evaluation. In the first part of the chapter, the former type of metrics is discussed, while in the second part, a general overview of quality in use by metrics will be performed. Apart from this general distinction, a more accurate taxonomy of the software metrics is reported in order to classify them on the basis of the applicability field in which they can be adopted.