A pragmatic means for measuring the complexity of source code ensembles

Most of the software metrics known and applied today are measured on a per file or even per function basis so that it is difficult to interpret their results for higher-order code ensembles such as components or whole systems. In order to overcome this weakness, we propose the hm-Index as a simple metric to condense the dependencies, i.e. the Fan-out, between source units in such code ensembles into a single number. As it is inspired by the h-Index in bibliometrics, it is based on a well-known procedure that already had significant impact in a different field. We expect the hm-Index to become a simple metric for comparing the code complexity of different components or systems in software engineering and present promising preliminary results from real-world systems confirming our assumption in this paper.