Characterizing Support for a Third-party Library: A Study of External Pull Requests for npm packages

Third-party libraries play a key role in building contemporary software applications. Despite this, most libraries are open source that often rely on volunteer (usually unpaid and overworked) contributions for their sustainability. Our motivation is to understand the extent to which third-party libraries are supported by contributions in the form of Pull Requests (PR) from outside the project team (i.e., External PR). Concretely, we analyze 1,076,123 PRs to investigate the External PR prevalence, bots, and the PR characteristics. Our results show that external contributions are prevalent, with packages receiving a high rate of (median of 73.45%) External PR . Furthermore, contributors are also submitting a high proportion of External PR (median of 87.62%). Results indicate a statistical difference in the acceptance of PRs submitted by bots compared to abandoned or open PRs. Furthermore, comparing external and internal PR, we find that Internal PR are more likely to be accepted. Statistically, we find that submitted patches (i.e., commit and code metrics) submitted by Internal PR are higher than patches submitted by External PR. We find that the External PR and Internal PR both have the same content (i.e., introducing new features and fixing bugs). Differently, External PR have more PRs that relate to documentation content, while Internal PR relates to refactoring-related changes第三方库（third-party libraries）在当代软件开发中占据核心地位。尽管如此，绝大多数第三方库均为开源项目，其可持续发展往往依赖于志愿者（通常无报酬且工作负荷过载）的贡献。本研究旨在探究：项目团队外部人员以拉取请求（Pull Requests，PR）形式提交的贡献（即外部拉取请求（External PR））对第三方库的支持力度。具体而言，我们对1,076,123个PR进行分析，以考察外部PR的流行程度、机器人（bots）参与情况以及PR的特征。研究结果显示，外部贡献极为普遍：多数软件包接收的外部PR占比中位数达73.45%。此外，贡献者提交的PR中，外部PR的占比中位数同样高达87.62%。结果表明，机器人提交的PR在接收情况上与已废弃或处于待处理状态的PR存在统计学差异。进一步对比外部PR与内部PR，我们发现内部PR的被接收概率更高。经统计学分析，内部PR所提交的补丁（即提交指标与代码指标）均优于外部PR提交的补丁。我们发现，外部PR与内部PR的核心内容并无差异，均为新增功能与缺陷修复。与之不同的是，外部PR中涉及文档内容调整的占比更高，而内部PR则更多包含代码重构相关的变更。