Entendendo a relação entre integração contínua e cobertura de testes: um estudo empírico

Abstract

The evolution of software development methodologies has enabled an increase in the delivering of new features and improvements. One of the best practices for increasing the delivery speed is continuous integration (CI). CI is a practice that motivates automating and integrating source code more often during software development. The adoption of CI helps developers to find integration issues faster. It is believed that the practice of CI helps the software to have fewer bugs throughout its lifecycle. One of the ways to find bugs is by performing software tests, and one of the most used metrics to ensure quality in software testing is test coverage. Therefore, it is believed that CI adoption and test coverage have a strong relationship. Previous studies have provided preliminary evidence for this relationship between CI and tests, however most of them do not demonstrate them empirically. This dissertation proposes an empirical study that aims to identify the relationship between CI adoption and test coverage through the analysis of several open source projects. We quantify coverage trend comparisons over time between projects that adopt (or do not ) CI. Our results suggest that CI projects have high test coverage rates and stability, while NOCI projects have low coverage rates and less potential for growth.