A developer has released an early version of Rust Glancer, an alternative Rust language server built around low memory use. The four-month-old project is available through a Visual Studio Code extension, but its creator cautions that it remains incomplete, contains known bugs and does not yet cover the full Language Server Protocol feature set.
Rust Glancer already includes workspace indexing, type inference and a trait solver based on Chalk. Common editor features such as go-to-definition, hover information, inlay hints and completions work for much ordinary Rust syntax. In a published demonstration, memory remained below 100MB, and the developer reported satisfactory use on a 2020 M1 MacBook Pro with 8GB of RAM. Those figures come from the project and are not an independent benchmark.
The architecture deliberately differs from rust-analyzer. Instead of keeping a lazy, incremental analysis graph in memory, Rust Glancer creates a frozen analysis result, stores indexed data on the filesystem and invalidates it when files are saved. Queries load only the required information for their duration. This reduces persistent memory demand but makes disk loading and deserialization slower than accessing already-resident data.
To preserve interactive completions, the server performs a shallow analysis of the current function body while the user types and reuses the most recent full index. New imports, structures and traits are not added to that index until the document is saved. The design therefore asks users to accept delayed awareness of some edits in exchange for lower resource use.
The project also accounts for code changes made outside the editor, including those produced by coding agents. A custom file watcher addressed inlay hints becoming misaligned, and external changes receive lower indexing priority so a burst of agent edits does not continuously trigger expensive work.
Rust Glancer began as what its author imagined would be a smarter form of ctags. Adding body lowering, type propagation, trait resolution and standard-library support gradually turned it into a more complete language server. The developer has professional Rust experience and has contributed to rustc, Clippy and rust-analyzer, but still describes the scale of the task as substantial.
The stated goal is not to replace rust-analyzer for everyone. Rust-analyzer is expected to remain preferable when completeness and keystroke-level accuracy are priorities. Rust Glancer is aimed at older or constrained computers and users willing to accept slower or deferred analysis to save RAM. Its current release is best understood as a working architectural experiment with usable features, not a drop-in equivalent.
The on-disk index is consequently both the project's defining optimization and its main compromise. Users gain a smaller memory footprint by accepting more filesystem work and a save boundary before some newly introduced symbols become visible to the server.
