A quantitative framework for software restructuring

Many existing software systems can benefit from restructuring to reduce mai ntenance cost and improve reusability, Yet, intuition-based, ad hoc restruc turing can be difficult and expensive, and can even make software structure worse, We introduce a quantitative framework for software restructuring. I n the framework, restructuring decisions are guided by visualized design in formation and objective criteria. The design information can be extracted d irectly from code to restructure existing or legacy software. Criteria for comparing alternative design structures include measures of design-level co hesion and coupling. Restructuring is accomplished through a series of deco mposition and composition operations which increase the cohesion and/or dec rease the coupling of individual system components, An example and a case s tudy demonstrate the framework. The framework ensures that restructuring re sults in measurable improvements in design quality. Copyright (C) 1999 John Wiley & Sons, Ltd.