PERFORMANCE Signal 399
Rust Glancer introduces low-memory Rust LSP with instant re-indexing after restart
↗
Illustration only Photo by Anton Savinov on Unsplash
Rust Glancer is a new Rust language server that aims to stay under 100 MB RAM and reload its index instantly after the editor restarts.
Engineers working on older or resource-constrained machines can keep a Rust language server running without exhausting memory, which improves overall system responsiveness. The persistent on-disk index eliminates the need for a full re-analysis after each editor launch, saving developer time during frequent restarts. However, the design trades off some real-time analysis speed and feature completeness, so teams must evaluate whether the memory savings outweigh the slower incremental updates.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Rust Glancer targets less than 100 MB RAM usage even for reasonably sized Rust projects.
It writes the full workspace analysis to disk, allowing immediate reuse of the index after an editor restart.
The frozen analysis model means new symbols are only indexed on file save and many LSP features are still missing.
THE READ
What the cluster adds up to.
The project presents an alternative implementation of the Rust Language Server Protocol (LSP) that prioritises low memory consumption, aiming for a RAM footprint below 100 MB on typical projects. It also guarantees that a previously indexed workspace can be re-used instantly after the editor is restarted, avoiding a full re-index. The author built the prototype over four months and provides a VS Code extension for testing.
Rust Glancer’s architecture differs from the dominant rust-analyzer by abandoning incremental, in-memory query processing. Instead, it performs a one-time full indexing pass, serialises the results to the filesystem, and loads only the needed fragments on demand. This design removes the memory-heavy salsa database and rowan syntax trees that rust-analyzer relies on, which are cited as primary sources of high RAM usage.
Benchmarks on a 2025 MacBook Pro M4 Max with 36 GB RAM show Rust Glancer completing a full index in five seconds versus six seconds for rust-analyzer, while on an older 2020 MacBook Pro M1 with 8 GB RAM the times are six seconds versus seven seconds respectively. Throughout the recorded demo the process memory stayed under the 100 MB target, confirming the low-memory claim on both high-end and modest hardware.
The trade-off is slower incremental feedback: because the analysis is frozen, only shallow checks run on each keystroke and new symbols (imports, structs, traits) are not visible until the file is saved. The current implementation also lacks a number of LSP capabilities and contains known bugs, meaning it is not yet a drop-in replacement for rust-analyzer in feature-rich environments.
Developers can try Rust Glancer by installing the published VS Code extension or by building the VSIX from the repository. It is especially suited for machines with limited RAM, such as the author’s 8 GB MacBook Pro, but teams should weigh the reduced feature set and slower on-the-fly analysis against the memory savings before adopting it for production work.
Written by elseif from the cluster below · checked for specifics the sources never containedTHE CLUSTER