*The Rustification of Bun#
Rewriting 500,000 lines of Zig into another language would be a gargantuan undertaking if done by hand. “A rewrite in another language would take a small team of engineers a full year. It would mean freezing bugfixes, security fixes or feature development for that time,” Sumner wrote.
Instead, Sumner went with Claude. He spun up about 50 dynamic Claude Code workflows, reaching a peak of about 1,300 lines of code per minute and generating over a million lines of Rust code. The job took 11 days and cost about $165,000 at API pricing. Claude Fable did most of the heavy lifting.
Kelley is artisan, Sumner is a factory manager. Indeed they cannot be using same tools or work on the same floor or within the same building. Blind trust in AI (even with guardrails) is a recipe for disaster. They have jumped to Rust to add more guardrails to their AI lunacy but it won’t hold for long. At this point there’s no single human who actually knows or understands the code and AI code is proven to have durability issues at scale. So direct rewrite gives a quick boost as code is ported from somewhat reviewed base onto new platform, but as new code gets added this will dissipate. It sounds like Bun issue is systemic Zig Rust or Pixie dust won’t fix it.
My modest career in software development has shown me that the real value of a developer isn’t how much code they can write, it’s how well they understand the existing code base. LLMs can’t understand because that requires memory larger than an LLM’s context window, by orders of magnitude.