CVM++ Stack-Based Virtual Machine

A fully working custom scripting language called CVM that compiles down to proprietary bytecode and runs on your own stack-based Virtual Machine — all built in C++ with zero external libraries. You will write a Lexer that tokenizes raw .cvm source files into typed tokens with line numbers, a Recursive-Descent Parser that builds a typed Abstract Syntax Tree, a Bytecode Compiler that walks the AST and emits flat binary bytecode with 16 custom opcodes and jump back-patching for loops and branches, and a VM that fetches, decodes, and executes each instruction in a switch loop. The CLI supports three modes: run a .cvm script file, open an interactive REPL, or use --debug mode to print the full token list, indented AST, hex dump, and human-readable disassembly before executing.

• How programming languages work end-to-end — from raw text to machine execution
• Lexical analysis: scanning source characters and producing typed tokens with line numbers
• Recursive descent parsing: translating tokens into a typed Abstract Syntax Tree (AST)
• Operator precedence climbing: correctly handling * / before + - before == <
• AST node design using polymorphism and std::unique_ptr for memory safety
• Designing a custom Instruction Set Architecture (ISA) with 16 opcodes
• Bytecode compilation: walking the AST and emitting flat uint8_t bytecode
• Jump back-patching: emitting placeholder offsets and fixing them after compiling blocks
• Building a stack-based VM with a fetch-decode-execute loop using a switch statement
• Tagged union value types to support both int and bool on the VM stack
• Variable slot allocation using a symbol table (map<string, int>)
• Runtime error handling: division by zero, stack underflow, unknown opcode
• Building a hex dump and human-readable bytecode disassembler
• Building a CLI in C++ with file mode, REPL mode, and debug mode
• Clean separation of concerns: each module in its own .h/.cpp pair
• Compiling a multi-file C++ project with CMake and g++