A qualitative analysis of code clones and object oriented runtime complexity based on method access points

In this paper, we present a new object oriented complexity metric based on runtime method access points. Software engineering metrics have traditionally indicated the level of quality present in a software system. However, the analysis and measurement of quality has long been captured at compile time, rendering useful results, although potentially incomplete, since all source code is considered in metric computation, versus the subset of code that actually executes. In this study, we examine the runtime behavior of our proposed metric on an open source software package, Rhino 1.7R4. We compute and validate our metric by correlating it with code clones and bug data. Code clones are considered to make software more complex and harder to maintain. When cloned, a code fragment with an error quickly transforms into two (or more) errors, both of which can affect the software system in unique ways. Thus a larger number of code clones is generally considered to indicate poorer software quality. For this reason, we consider that clones function as an external quality factor, in addition to bugs, for metric validation.