Thesis Defence: General Diversity Challenges in Open Source to Rust Community Insights and the Role of Code Language Models

Rohit Dandamudi, supervised by Dr. Gema Rodriguez-Perez, will defend their thesis titled “General Diversity Challenges in Open Source to Rust Community Insights and the Role of Code Language Models” in partial fulfillment of the requirements for the degree of Master of Science in Computer Science.

An abstract for Rohit Dandamudi’s thesis is included below.

Defences are open to all members of the campus community as well as the general public. Registration is not required for in-person defences.

---

ABSTRACT

Open-source software (OSS) communities drive technological innovation and collaboration in today’s software-driven globe. However, fostering a healthy and thriving OSS ecosystem necessitates a solid commitment to Diversity, Equity, and Inclusion (DEI). This thesis investigates the multifaceted aspects of diversity within OSS communities, with a particular focus on the Rust programming language. The study begins with a survey-based analysis to understand the diversity perceptions of racialized contributors, revealing significant disparities in their experiences compared to non-racialized counterparts, especially in attaining influential maintainer roles. Building on these insights, a network analysis of the Rust ecosystem was conducted by examining pull request data from key repositories, highlighting low geographic and gender diversity, as well as a concentration of contributions in a few countries and among male contributors. This analysis emphasizes the need for targeted DEI initiatives to enhance inclusivity and resilience within the Rust community. Additionally, the thesis explores the potential of Code Language Models (CLMs) as technological solutions to mitigate the high learning curve associated with Rust. An empirical evaluation of PolyCoder, a CLM designed for low-resource scenarios, was performed using translated code generation benchmarks. The findings indicate that while PolyCoder shows promise, its performance in multilingual and low-resource contexts like Rust remains limited, highlighting the necessity for more robust evaluation methodologies and enhanced tooling. By integrating survey research, network analysis, and CLM evaluation, this thesis offers a holistic view of the challenges and opportunities in promoting diversity and sustainability within OSS. The conclusions advocate for a dual approach that addresses both human-centric diversity challenges and leverages technological advancements to create a more inclusive and resilient open-source environment.