How Rust IDEs Understand Code

A deep dive into how Rust IDEs understand code, based on a RustRover livestream with Lukas Wirth (rust-analyzer lead, Zed) and Vlad Beskrovny (JetBrains). Key topics include why IDEs must reimplement the compiler frontend for low-latency responses, the evolution from RLS to rust-analyzer, the challenges of Rust's cyclic module graph and name resolution, why procedural macros are isolated into separate processes to prevent IDE crashes, the complexity of replicating Rust's type inference rules, and how RustRover uses PSI stubs and MapReduce-style indexing while rust-analyzer uses Salsa-based query-driven memoization. The post also covers how RustRover integrates semantic analysis into debugging via a custom LLDB integration with MIR evaluation.