Author(s)

T. Andersson, K. Enholm & A. Torn

Abstract

Length-independent measure of software complexity T. Andersson, K. Enholm & A. Torn Abo Akademi University, Computer Science ABSTRACT A software complexity concept based on measurement theory is intro- duced. It is argued that a complexity measure conforming with our em- pirical understanding of complexity schould be independent of software length. In analogy with the formula for mass, volume and density the measure is defined by W(P) = L(P)c(P), where P is any piece of soft- ware, W(P) the effort in developing it, L(P] its length, and c(P) its com- plexity. The properties of the measure are discussed and it is found that some of the conflicts in desirable properties reported for other complexity measures are resolved. INTRODUCTION Software complexity is an attribute with many meanings that is used to characterize different entities. In many cases complexity is defined in such a way that it correlates strongly with the length of a program. We argue that complexity should

Keywords