Rust and TypeScript's strong typing systems support features like autocomplete and memory optimization. However, agentic coding tools experience drawbacks when utilizing the Language Server Protocol (LSP), which slows down operations and complicates context without significant benefits. In practice, types are often written but ignored, complicating the coding process. While agents can progress by incorporating types into running code, this leads to delayed type error resolution and excessive backtracking, ultimately resulting in inefficient coding practices.
Having a well-integrated type system that makes sense and gives you optimization potential for memory layouts is generally a good idea.
Most agentic tools don't have access to an LSP and those that do haven't meaningfully benefited from it.
The LSP protocol slows things down and pollutes the context significantly, with models not trained sufficiently to work with type information.
The agent coding loop can make progress by writing code and adding types during compilation, but this often results in having to fix type errors later.
Collection
[
|
...
]