Capture Software Design Rationale Based on an Integrated Analysis of Both Software Process and Product Quality Requirements from Multiple Perspectives

Abstract

The capture of design rationale in terms of the process and product quality requirements for a software system from different perspec tives in concurrent engineering poses two challenges (1) process and product quality requirements arising from different perspectives usually conflict with each other, and (2) both process and product quality requirements are often vague and imprecise Recent research into methods for handling software quality requirements has taken one of two approaches—quantitative and qualitative Quantitative approaches are based upon software metrics and specify requirements using predicate logic Qualitative approaches represent requirements qualitatively and decom pose the requirements into a hierarchy of subrequirements A unified approach for specifying and analyzing requirements quantitatively and qualitatively is also developed based on fuzzy logic However, none of these methods addresses design rationale capture based on integrated analysis of both product and process quality requirements from multiple perspectives in concurrent software development In this paper a formal framework is developed for an integrated analysis of software process and product quality requirements to support design rationale capture from multiple perspectives in concurrent software development It provides a top-down approach for decomposing vague, complex quality require ments based upon an ontological model of a perspective, and a bottom-up approach for analyzing inter-requirement relationships from multiple perspectives A feature-based approach for assessing the impact of design alternatives on both process and product quality requirements from multiple perspectives is developed by explicitly documenting the effects of design options on design features and the effects of design features on quality requirements The techniques described by the framework are illustrated using a distributed order processing system