I’ve spent the past 4 months developing an alternative Rust LSP prioritizing minimal memory consumption. Rust Glancer targets <100mb for typical projects and enables instant indexing after restarts by saving results to disk. Benchmarks on a MacBook Pro M1 with 8GB RAM confirm it runs smoothly without consuming excessive system resources.

This architecture deliberately avoids heavy in-memory databases and complex syntax trees. By storing frozen analysis results on the filesystem and loading them only when necessary, it trades minor real-time latency for significantly lower RAM overhead. Workspace metadata stays permanently offloaded until actively queried.

The tool currently supports a full indexing pipeline, type inference, and a trait solver. Core LSP actions like goto definition, hover, hints, and completions function reliably. Though still evolving with known bugs, developers can experiment with this memory-efficient alternative using the VS Code extension or by compiling it from source.