Avaliação de refatoração de software com programação orientada a aspectos usando métricas

Abstract

Code scattering and code tangling aect software development in many ways, including poor traceability of requirements, low productivity, poor overall quality and increased eorts for software maintenance. Aspect-oriented programming (AOP) has emerged as a proposal to address these issues through the modularization of crosscutting concerns to aspects. Few studies on empirical evaluation of the benets of aspect-oriented paradigm were published. Results presented in these studies are frequently subjective, and some studies are non-conclusive. In addition, most of these studies are based on the implementation of only one or two crosscutting concerns into aspects, and the evaluation is based on few software metrics. In this work, the evaluation of AOP implementation through software metrics is proposed. The main idea is to implement crosscutting concerns as aspects, with focus on those that were not given properly attention in the literature. Four types of crosscutting concerns are implemented as aspects, the evaluation of POA was based through large quantity of metrics and also was performed a verication of the impact of refactoring on quality attributes. Two softwares were used as case studies. From metrics and experiences during the implementation phase, quantitative and qualitative analyzes were produced. The benets and detriments of refactoring with POA were reported for developers to assess whether POA presents the best solution. As conclusion of this work, refactoring software with POA is not indicated for all types of crosscutting concerns. The crosscutting concern tracing is indicated as a case in which POA is extremely relevant.