Uma extensão de iStar para modelagem de requisitos para prevenção de code smells

Abstract

Software quality is one of the most important factors for the success of computational systems, and code smells are indicators of structural issues that can compromise this quality. Code smells are recurring patterns in code that point to potential design or implementation flaws, negatively affecting software maintenance and evolution. Despite the existence of practices and tools for their identification and correction, the prevention of code smells is often overlooked in the early stages of development, such as requirements engineering. Requirements engineering is a fundamental phase for defining and specifying the functionalities and desired characteristics of a system. Goal-oriented modeling languages, such as iStar, are widely used to capture the needs and motivations of stakeholders during this phase. iStar has been extended to address the specificities of various domains, supported by a structured process for creating extensions. Thus, this work proposes an iStar extension to represent requirements aimed at preventing code smells. The extension includes three main concepts: Quality Smell, Task Smell and Smell Metric. These concepts enable the systematic modeling of code smell prevention, contributing to the development of higher-quality systems with lower maintenance costs.