Design and code metrics through a DIANA-based tool

Abstract

This paper discusses a metrics approach for analyzing software designs which helps designers engineer quality into the design product. These metrics gauge project quality as well as design complexity at all times during the design phase. The metrics are developed from primitive design metrics which are predictive, objective and automatable. The architectural design metrics used are comprised of terms related to the amount of data flowing through the module and the number of paths through the module. A detailed design metrics component takes into account the structure and complexity of a module. To automate the calculation of the design metrics in the Rational environment, DIANA (Descriptive Intermediate Attributed Notation for Ada) was utilized. Provided in the environment are packages allowing for the traversal and retrieval of the DIANA structure. By combining the defined packages with customized packages, an Ada design metrics analysis tool was developed. This paper will discuss our design metrics and their automation at Magnavox. Empirical results will illustrate the metrics' success in identifying stress points in a software design and demonstrate their relationship to the quality of the resulting software.