Midori(ミドリ)

A Programming Language

Statically typed functional language featuring algebraic data types, exhaustive pattern matching, classes, and a strong module system. Compiles to bytecode for a stack-based Midori VM with per-VM garbage collection and shared async runtime handles.

Static TypingFunctionalPattern MatchingClassesBytecode VMGarbage CollectionAsync/Await

Key Features

Static Type System

Strong static typing with type inference and generics. Catch errors at compile time while maintaining clean, concise code.

Pattern Matching

Exhaustive pattern matching on algebraic data types with compiler-enforced coverage, ensuring all cases are handled.

Algebraic Data Types

Express complex data structures with structs (product types) and unions (sum types) for safer, more expressive code.

Classes

Constrained generics for powerful polymorphism. Define behavior contracts and implement them for any type.

Module System

Explicit imports and exports with privacy enforcement. Build large-scale applications with clear module boundaries.

Expression-Oriented

Write more functional, composable code with natural control flow.

Quick Start

Hello World

import { "../MidoriPrelude/IO.mdr" }

IO::PrintLine("Hello, Midori!");

Functions

// Simple function
defun square(x: Int) : Int => {
    return x * x;
};

// Generic function
defun identity<T>(value: T) : T => value;

// Higher-order function
defun apply<T, R>(
    fn: fn(T) -> R,
    value: T
) : R => {
    return fn(value);
};

Algebraic Data Types

// Product types (structs)
struct Point {
    x: Float,
    y: Float,
};

// Sum types (unions)
union Option<T> = Some(T) | None;

union Result<T, E> = Ok(T) | Err(E);

def origin = new Point(0.0, 0.0);
def maybe = new Option::Some(42);

Pattern Matching

Match Expression Example

union Result<T, E> = Ok(T) | Err(E);

defun handle_result<T>(result: Result<T, Text>) : Text => {
    return match result with
        case Result::Ok(value) => "Success: " ++ (value as Text)
        case Result::Err(msg) => "Error: " ++ msg
        default => "Unknown"
    ;
};

Classes

Classes let you define shared behavior that can be implemented for multiple types while maintaining type safety.

Class Definition and Implementation

// Define a class
class Show<T> {
    show: fn(value: T) -> Text;
};

// Implement for Int
instance Show<Int> {
    defun show(value: Int) : Text => {
        return value as Text;
    };
};

// Use with constraints
defun display<T>(value: T) : Text where Show<T> => {
    return Show::show(value);
};

def message = display(42);  // "42"

Advanced Features

Pipe Operator

defun double(x: Int) : Int => x * 2;
defun add_ten(x: Int) : Int => x + 10;

def result = 5
    |> double
    |> add_ten
    |> double;  // 40

Closures

defun make_counter() : fn() -> Int => {
    def count = 0;
    return fn() : Int => {
        count = count + 1;
        return count;
    };
};

def counter = make_counter();
def first = counter();   // 1
def second = counter();  // 2

Async/Await

Concurrency in Midori uses runtime worker VMs with shared futures and shared cell handles. The compiler selects value/cell/shared-cell local storage so closure and async captures preserve by-reference semantics.

Async/Await Example

import { <IO> }

defun compute_heavy(x: Int) : Int => {
    // Computations...
    return x * x;
};

// Spawn concurrent task (scheduled on runtime workers)
def task : Future<Int> = async compute_heavy(42);

IO::PrintLine("Task running in background...");

// Await result (current task waits; runtime can execute other work)
def result = await task;

Language Operators

Arithmetic

+ - * / %

Comparison

== != < > <= >=

Logical

&& || !

Bitwise

& | ^ <~ ~>

String

++ (concatenation)

Functional

|> (pipe)

Example: Recursive Data Structures

Binary Tree Implementation

union Tree<T> = Leaf(T) | Node(Tree<T>, Tree<T>);

defun height<T>(tree: Tree<T>) : Int => {
    return match tree with
        case Tree::Leaf(value) => 1
        case Tree::Node(left, right) => {
            def left_height = height(left);
            def right_height = height(right);
            return 1 + (if left_height > right_height
                        then left_height
                        else right_height);
        }
        default => 0
    ;
};

Generic Linked List

union List<T> = Cons(T, List<T>) | Nil;

defun map<A, B>(list: List<A>, f: fn(A) -> B) : List<B> => {
    return match list with
        case List::Cons(head, tail) =>
            new List::Cons(f(head), map(tail, f))
        case List::Nil =>
            new List::Nil()
        default =>
            new List::Nil()
    ;
};

Type System

Primitive Types: Int, Float, Bool, Text, Unit

Composite Types: Array<T>, Future<T>, structs, unions

Function Types: fn(T1, T2) -> R

Generic Parameters: Single and multiple type parameters

Type Constraints: Class constraints with where clause

Type Inference: Automatic type deduction at instantiation

Standard Library

IO.mdr

Input/output operations including PrintLine for console output and file I/O utilities.

Math.mdr

Mathematical functions including trigonometry, logarithms, and common operations.

System.mdr

System interaction utilities for command line arguments, environment variables, and more.

DateTime.mdr

Timing and date operations for working with temporal data in your applications.

Foreign Function Interface (FFI)

Midori can call C/C++ functions directly using the FFI. Types like Text and Array are automatically marshalled between the VM and native code.

FFI Declaration

// Declare external C++ function
foreign "MIDORI_FFI_ReadFile" ReadFile : fn(Text) -> Array<Byte>;

// Use it like a normal function
def bytes = ReadFile("data.bin");

Architecture

Frontend: Lexer → Parser → Type Checker

Optimizer: Constant folding, tail call optimization, strength reduction

Backend: Bytecode generator → Linker

Runtime: Stack-based VM with per-VM mark-and-sweep GC and shared async runtime handles

Get Started

Start building with Midori today

View on GitHub Documentation