Torja (Symbol Resolver)

Torja is the core Symbol Resolver of LightVM. It acts as a bridge between high-level bytecode—which uses human-readable names for variables and functions—and the high-performance execution engine that relies on memory-efficient numerical indices.

How Torja Works

Before your bytecode reaches the execution phase, Torja performs a crucial pass to resolve all symbolic references into fixed-position indices.

• Symbol Mapping: Scans through all imports and dynamic instructions to map every unique variable or function name to a stable integer index.
• Dynamic Resolution: If a variable or function is encountered for the first time during the resolution pass, Torja dynamically assigns a new index, ensuring a unique ID for every symbol throughout the lifecycle of the program.
• Instruction Specialization: It transforms generic, name-based instructions (e.g., get, set, inc) into their specialized index-based counterparts (get_idx, set_idx, inc_idx). This minimizes runtime lookups and significantly reduces CPU overhead during execution.
• Value Promotion: During the symbol pass, Torja also optimizes push instructions by promoting generic Value types into specialized, type-specific opcodes (e.g., push_int16, push_float64, push_string), ensuring the VM knows the exact data size and type immediately.
• Functional Scope Tracking: Torja identifies function parameters and scoped identifiers, ensuring that all local symbols are correctly tracked within the symbol table and prepared for the VM's stack-based architecture.