A dynamic size measure for object oriented software

Many static metrics exist in literature for object oriented (OO) software quantifying various static aspects of complexity of its design or source code. However, there is a growing need to employ dynamic measures as they are obtained during the execution of code and thus can reflect actual run-time situations. This has led to various dynamic measures being defined in the past few years focussing on dimensions like coupling, cohesion etc. However, dynamic measures on size have been rarely addressed. In this paper, we propose a dynamic measure for size of OO software at system level which takes into account the number of objects created during the execution. The proposed measure is then theoretically validated using two popular theoretical frameworks. A dynamic analyzer code is developed using AspectJ, an aspect oriented programming extension for Java, which facilitates the collection of metric data. An empirical study consisting of descriptive statistics, Pearson correlation analysis and principal component analysis is carried out on ten sample Java programs to compare the proposed measure with several existing static measures. The study indicates that the proposed measure can serve as a useful alternative to measure size of OO software. Further, the proposed measure is correlated with maintainability of OO software as an external quality attribute. The results indicate that the proposed measure has significant positive relationship with maintainability.