Rust Glancer: Rust LSP using 100x less RAM
Points and comments are a snapshot, not live.
A new Rust LSP uses disk-based indexing to cut memory usage below 100 MB.
Rust Glancer is an alternative LSP for Rust that stores analysis results on disk instead of keeping them all in memory like rust-analyzer does. This approach targets machines with limited RAM: tested on a 2020 M1 MacBook Pro with 8 GB, memory stayed under 100 MB during use. Indexing times are comparable to rust-analyzer (5-9 seconds for base indexing), but the trade-off is that live analysis freezes until save. The project is 4 months old and incomplete, lacking some features like proc macro support and many code actions. The author built it using heavy LLM assistance but reviews all code personally.
What commenters are saying
Commenters were intrigued by the memory savings and the disk-based approach, especially those frustrated by rust-analyzer's memory usage in large workspaces. Several asked about Zed and Neovim support (the author said it's coming in the next release). The author clarified that peak RAM during indexing can exceed rust-analyzer's, but the gain comes from not needing frequent reindexing and from loading only needed data on demand. One commenter noted that JetBrains' Rust Rover may already achieve similar compactness, which the author said they haven't tested.