Tsrs: A Port of TypeScript to Rust

Re:Tsrs: A Port of TypeScript to Rust by dabinat on Wednesday October 21, 2026 @03:29AM (#50177232)

This is what bothers me about all the vibe-coded Rust ports. The LLMs are mostly porting the existing architecture and code as-is, which was intended for a different language with different features and tradeoffs. A port that actually uses Rust’s features and is architected in a way that makes sense for Rust code would be a lot better, more maintainable (if that matters anymore) and probably faster too.

Another advantage human ports have is the opportunity to reconsider earlier architectural decisions and change them, instead of just enshrining them in the new language.

[ Parent | on HN ]

Re:Tsrs: A Port of TypeScript to Rust by wewxjfq on Wednesday October 21, 2026 @04:41AM (#50177724)

I never see this caveat mentioned in all the "... vibe-ported to Rust" threads, which makes me wonder who comments on these projects. I hand-ported a few TS hobby projects to Rust+WASM pre-AI and realized there are only two ways to go: Do a low effort translation and create very unidiomatic Rust code or redesign the application from the ground up. There's no middle ground. Option 1 is almost a waste of time and option 2 is incredibly difficult.

[ Parent | on HN ]

Re:Tsrs: A Port of TypeScript to Rust by rk06 on Wednesday October 21, 2026 @06:55AM (#50178829)

option 2 is also impractical for most part. doing a rewrite is big thing already, changing implementation paradigm would make "catching up" impossible.

Only practical option is to do low level rewrite (otpion 1) and then rewrite module by module into better approaches.

[ Parent | on HN ]