"Anything that one cannot measure cannot be controlled." Software metrics provide insight into the quality of software, informing the developers about the status of a project. Both what to measure and when to inform the developer of changes in code quality are important factors in a software metrics tool. Our work describes metrics that measure some aspects of software complexity. Complexities of a system arise from size, from the relationships between parts and between a part and the whole, and from changes or evolution. The metrics described in this paper concentrate on measuring complexities due to size and relationships. The work on Smalltalk contains seven metrics; on C++, eight. Differences are due to idiosyncrasies of the languages. We believe the best time to inform the developer of changes in code quality is at the time source code is edited. Our tools provide realtime metric feedback on the complexity of methods (Smalltalk or C++) whenever methods are edited. Chang...
Subroto Bhattacharya, Stephen G. Graham