Mitigando a carga cognitiva em testes unitários: uma abordagem focada em processos de trabalho

Abstract

Software quality is a central aspect of software engineering projects, as it influences reliability, maintainability, and user satisfaction. In this context, unit tests verify the behavior of small units of code and enable early fault identification. However, their adoption may be hindered by factors that make the practice costly or, at times, neglected. Among the elements that may compromise unit testing is cognitive load, understood as the mental effort required to understand and perform tasks. This load tends to increase in the absence of planning, in the presence of redundant processes, gaps in automation, uncertainties about validation criteria, and pressure for quantitative metrics. In such scenarios, delays, rework, and lower adherence to desired practices may occur. This dissertation aims to propose strategies and recommendations to mitigate the cognitive load faced by developers during unit testing and to identify characteristics that favor their adaptation to different development contexts. The research adopts a qualitative approach based on Grounded Theory, focusing on work-process aspects, drawing on semi-structured interviews with developers. As an outcome of the analysis, a set of recommendations is presented and organized into four stages of the testing process — conception, writing, execution, and validation — formulated as suggestions for flexible application. The proposals aim to make criteria more explicit, support the standardization of decisions, and guide the required automation, providing a starting point for local adjustments according to each team’s context.