NimVM

A basic programming language written in Nim

Project structure

The project is split into several directories and submodules to ease human inspection:

• README.md -> This file here (lol)
• docs -> Contains markdown files with the various specifications for NimVM (bytecode, grammar, etc)
  • docs/bytecode.md -> Lays out the bytecode specification for NimVM as well as serialization guidelines
  • docs/grammar.md -> Formal grammar specification in EBNF syntax
• src -> Contains source files
  • src/main.nim -> This is the main executable for NimVM (REPL, run files, etc.), currently not in this repo
  • src/backend -> Contains the backend of the language (lexer, parser and compiler)
    • src/backend/meta -> Contains meta-structures that are used during parsing and compilation
    • src/backend/lexer.nim -> Contains the tokenizer
    • src/backend/parser.nim -> Contains the parser
    • src/backend/compiler.nim -> Contains the compiler
  • src/frontend -> Contains the language's frontend (runtime environment and type system)
    • src/frontend/types -> Contains the implementation of the type system
    • src/frontend/vm.nim -> Contains the virtual machine (stack-based)
  • src/util -> Contains generic utilities used troughout the project
  • src/util/bytecode -> Contains the bytecode serializer/deserializer
    • src/util/bytecode/serializer.nim -> Contains the bytecode serializer
    • src/util/bytecode/deserializer.nim -> Contains the bytecode deserializer
    • src/util/bytecode/objects.nim -> Contains object wrappers for bytecode opcodes
  • src/util/debug.nim -> Contains the debugger

Language design

NimVM is a generic stack-based bytecode VM implementation, meaning that source files are compiled into an imaginary instruction set for which we implemented all the required operations in a virtual machine. NimVM uses a triple-pass compiler where the input is first tokenized, then parsed into an AST and finally optimized before being translated to bytecode.

The compilation toolchain has been designed as follows:

• First, the input is tokenized. This process aims to break down the source input into a sequence of easier to process tokens for the next step. The lexer (or tokenizer) detects basic syntax errors like unterminated string literals and multi-line comments and invalid usage of unknown tokens (for example UTF-8 runes)
• Then, the tokens are fed into a parser. The parser recursively traverses the list of tokens coming from the lexer and builds a higher-level structure called an Abstract Syntax Tree-- or AST for short-- and also catches the rest of static or syntax errors such as illegal statement usage (for example return inside a function), malformed expressions and declarations and much more
• After the AST has been built, it goes trough the optimizer. As the name suggests, this step aims to perform a few optimizations, namely:
  • constant folding (meaning 1 + 2 will be replaced with 3 instead of producing 2 constant opcodes and 1 addition opcode)
  • global name resolution. This is possible because NimVM's syntax only allows for globals to be defined in a way that is statically inferrable, so "name error" exceptions can be caught before any code is even ran. The optimizer also detects attempts to modify a constant's or a let's value at compile-time.
• Once the optimizater is done, the compiler takes the AST and compiles it to bytecode for it to be later interpreted by our virtual machine implementation