Canonical has announced its co-funding of a three-year PhD project focused on the translation of large C codebases into the Rust programming language using artificial intelligence. This initiative, revealed by Engineering Vice President Jon Seager on Ubuntu’s Discourse forum, will be conducted at the University of Bristol’s Programming Languages Research Group.
This project is not a sweeping overhaul of Ubuntu’s codebase but rather a methodical investigation into the feasibility of using AI for this task. The objective is to determine if a large language model (LLM) can effectively decompose extensive C programs, which may consist of hundreds of thousands of lines, and rewrite them into “safe, behaviourally correct and maintainable Rust.” This approach aims to address the safety concerns often associated with C programming.
Seager emphasized that while traditional source-to-source translators exist, they often fail to produce code that aligns with Rust’s safety standards. He noted, “Traditional source-to-source translators can process substantial amounts of code, but often preserve the structure of the C too literally.” This can lead to Rust code that still relies on unsafe operations and retains C idioms that are not ideal for Rust maintainers.
The project will specifically examine two existing tools: snap-confine and AppArmor. The latter has seen a decline in usage in favor of SELinux in some distributions, making a Rust implementation potentially beneficial for its remaining users.
Canonical is not undertaking this project alone; it is co-sponsoring it with UK Research and Innovation, a public body associated with the UK government. Seager will oversee the project alongside Professor Meng Wang and Dr. Cristina David from the University of Bristol.
While the initiative is promising, skepticism remains regarding the challenges of decomposing large codebases into manageable components for AI processing. The complexities involved in this task echo long-standing difficulties in algorithmic decomposition for parallel processing, suggesting that the project may face significant hurdles.
Ultimately, this PhD research aims to provide empirical evidence on the viability of AI-driven code translation, a critical step in assessing the potential for improving software reliability through innovative methods.
This article was produced by NeonPulse.today using human and AI-assisted editorial processes, based on publicly available information. Content may be edited for clarity and style.
