Dataset and Replication package for Platform-Specific APIs and Security Issues: A First Empirical Study

This dataset was utilized in the research paper titled "Platform-Specific APIs and Security Issues: A First Empirical Study" for publication in International Conference on Program Comprehension (ICPC). We provide a roadmap to replicate all the steps of our methodology. More information in the `REPLICABILITY.md` file. Herein, we present a concise abstract of the study: A platform-specific API is an interface implemented for a particular platform (e.g., Windows or Linux) that may not function correctly on others. This paper presents an empirical study of platform-specific APIs in the Python Standard Library, examining their relationship with defensive structures and their association with security issues. Our results show that platform-specific API usage is common, appearing in 21.40% of the analyzed files. We identified over 12k occurrences of platform-specific APIs, more than 74% of which occur without defensive structures, and most of them appear in test code (92.57%). Moreover, platform-specific APIs are occasionally associated with security issues. We identified 558 security issues and 532 occurrences of platform-specific APIs associated with them, over 53% of which appear in code without defensive structures. These findings reveal that the same APIs that facilitate cross-platform compatibility can also introduce potential security issues. We conclude by discussing implications for both practitioners and researchers concerning cross-platform development and the trade-off between compatibility and security in the Python Standard Library.