My Thoughts on Bun's Rust Rewrite
The article discusses the implications of Bun's recent rewrite in Rust, emphasizing the foundational role of Zig in its development. It raises concerns about the lack of human oversight in the code review process, given that the rewrite was completed by AI. The author highlights the shift in risk from the original developer to the users of Bun after its acquisition by a larger company.
- ▪Bun's architecture was initially built using Zig, which allowed for rapid prototyping of a high-performance JS runtime.
- ▪The recent rewrite in Rust involved 6,755 commits, all generated by AI without human review.
- ▪The author warns that passing tests do not guarantee the correctness of the code under all conditions, especially with complex interactions.
Opening excerpt (first ~120 words) tap to expand
My Thoughts on Bun's Rust Rewrite2026-05-16 Updated: 2026-05-16 tags: zig rust bunTable of ContentsThe Real Issue No One Dares to Say"All Tests Pass" Doesn't Mean What You ThinkAfter the Acquisition, the Risk Bearer Has ChangedZig's "Problems" Were MisdiagnosedSo, Will This Rewrite Work?ReferencesBefore we discuss Rewrite Bun in Rust, there's something that needs to be said, because no one is saying it.Bun stands where it does today because of Zig.Jarred chose Zig back then not because it was "cool," but because Zig enabled a small team to rapidly prototype a high-performance JS runtime without a GC, without a heavy runtime.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Jiacai Liu's personal website.