AN EVALUATION OF SOFTWARE-DESIGN USING THE DEMETER TOOL

The purpose of this study was to investigate the provision of measures suitable for evaluating software designs. These would allow some degree of predictability in estimating the quality of a coded software product. Experiments were conducted to test hypotheses concerning the evaluation of software design metrics related to code complexity. Three medium-sized software projects were used. Metrics from the design stage of these projects were collected by using a software package called DEMETER. This tool was developed specifically to provide the researcher with a usable method of collecting data for calculating design metrics. Data analysis was performed to identify relationships between the measures of design quality and software coded complexity represented by control flow as well as data complexity. The results indicated that reducing the number of interconnections between software units, together with an increase in the relationships of the elements within a module (controlled by the flow of global data), improves the resultant software.