Entendendo a relação entre integração contínua e cobertura de código

Abstract

Continuous Integration (CI) is a widely adopted software engineering practice emphasizing frequent software integration through an automated build process. Although CI has been shown to detect errors earlier in the software life cycle, the relationship between CI and code coverage still needs to be clarified. Our work aims to fill this gap by investigating the quantitative and qualitative aspects of this relationship. In the quantitative study, we compare 30 CI projects and 30 projects that have never adopted CI (NOCI projects) to investigate whether CI is associated with higher code coverage rates. We analyzed 1,440 versions from different projects to identify trends in code coverage. Our findings reveal a positive association between CI and higher code coverage rates. Our qualitative study consisted of a survey and a document analysis. The survey revealed several significant findings, including a positive association between continuous integration (CI) and higher code coverage rates, indicating the value of CI in promoting testing practices. Additionally, our survey highlighted the importance of using code coverage during the authoring and review process, which can help identify potential issues early in the development cycle. The document analysis focused on coverage-related themes in pull request conversations in CI-enabled projects. Through this analysis, we discovered the main topics related to coverage usage during pull requests, which can provide valuable insights into how developers use coverage to improve code quality. This information can help guide the development of best practices for using coverage in CI-enabled projects, ultimately improving the quality and reliability of software products. Our work offers insights into the evolution of code coverage in CI, which can assist researchers and practitioners in adopting tools and practices for better monitoring, maintenance, and even enhancement of code coverage.