Lightweight Grammar Based Parsing Without Language Server

via “lightweight grammar-based parsing without language server”

Syntax highlighting support for Machine Learning Intermediate Representation

Unique: Uses VS Code's native TextMate grammar engine for tokenization instead of implementing a custom parser or language server, eliminating the need for a separate process and reducing memory/CPU overhead by ~50-80% compared to LSP-based alternatives

Significantly faster startup and lower resource usage than language server-based highlighters (e.g., MLIR LSP), at the cost of no semantic features; ideal for syntax-only highlighting on resource-constrained systems