Module bytecode 

Stack-based bytecode VM for TensorLogic expressions.

This module provides a compiler from TLExpr to a flat BytecodeProgram and a lightweight virtual machine that executes it. Repeated evaluation of compiled expressions is faster than recursive interpretation because the expression tree is only traversed once during compilation; subsequent executions only walk the flat instruction array.

Quick Start

use tensorlogic_compiler::bytecode::{compile, execute, VmEnv, VmValue};
use tensorlogic_ir::TLExpr;

// Compile 2.0 + 3.0
let expr = TLExpr::add(TLExpr::Constant(2.0), TLExpr::Constant(3.0));
let program = compile(&expr).unwrap();

let env = VmEnv::new();
let result = execute(&program, &env).unwrap();
assert_eq!(result, VmValue::Num(5.0));

Structs

BytecodeProgram

A compiled, flat sequence of Instructions.

VmEnv

Variable environment passed to the VM at execution time.

VmStats

Execution statistics collected during a single VM run.

Enums

CompileError

Errors that can occur during bytecode compilation.

Instruction

Stack-based VM instruction.

VmError

Errors that can occur during VM execution.

VmValue

A runtime value on the VM stack.

Constants

DEFAULT_MAX_DEPTH

Default maximum expression depth allowed during compilation.

Functions

compile

Compile a TLExpr to a BytecodeProgram.

compile_with_config

Compile a TLExpr to a BytecodeProgram with an explicit depth limit.

execute

Execute a BytecodeProgram and return the top-of-stack value after Halt.

execute_with_stats

Execute a BytecodeProgram and return both the result and execution statistics.