Quality guidelines = Designer metrics

In spite of the significant body of traditional source code metrics, there research on has been a general failure to produce conclusive evidence as to their effectiveness for measuring software quality. We describe and recommend a potentially much more powerful and sensitive quality assessment alternative, software quali~ guidelines. Software quality guidelines are presented as designer metrics , that is, user-defined rules or constraints relating to measurable features of a program\u27s structure, semantics, and syntax that affect its quality. To provide a methodology for designing, applying, and validating software quality guidelines, we recommend and briefly summarize IEEE standard 1061. This standard gives a process for constructing and implementing a software quality metrics framework that can be tailor-made to meet quality requirements for a particular project andlor organization. Our paper then demonstrates how software quality guidelines fit within the IEEE framework and gives an example illustrating how userdefined guidelines can be applied to evaluate or assess the quality of an Ada source unit. This guideline-based assessment of quality is then compared with an analysis based on traditional McCabe and Halstead metrics. Finally, we introduce a tool being developed by SAIC, called AdaReVuTM, as an effective mechanism for implementing and applying user-defined quality guidelines for Ada source code