Quick Start

Moshi (模式) is a Julia package for algebraic data types, pattern matching, and trait derivation. This guide gets you from zero to a working example in a few minutes.

Install

From the Julia REPL, press ] to enter Pkg mode:

pkg> add Moshi

Your first ADT

Import @data and define a sum type with several variants:

using Moshi.Data: @data

@data Message begin
    Quit()
    struct Move
        x::Int
        y::Int
    end
    Write(String)
    ChangeColor(Int, Int, Int)
end

msg = Message.Move(3, 4)

Variants can be singletons (Quit()), named structs (Move), or tuple-like constructors (Write, ChangeColor).

Note: even a singleton variant is constructed with () — Message.Quit is the variant type, while Message.Quit() is the value. See Singleton Variant.

Pattern match on it

@match destructures values and returns the right-hand side of the first matching arm:

using Moshi.Match: @match

describe(msg) = @match msg begin
    Message.Quit() => "quit"
    Message.Move(; x, y) => "move to ($x, $y)"
    Message.Write(text) => text
    Message.ChangeColor(r, g, b) => (r, g, b)
end

describe(msg)  # "move to (3, 4)"

If you have used MLStyle, the syntax will feel familiar. Moshi also supports matching on arrays, tuples, and more — see Builtin Patterns.

Derive traits

Use @derive to implement Show, Hash, and Eq from your type definition:

using Moshi.Derive: @derive

@derive Message[Show, Hash, Eq]

Works on both @data types and ordinary Julia structs.

Next steps

Topic	Guide
ADT concepts	Algebraic Data Types
@data syntax	ADT Syntax
@match patterns	Pattern Matching Intro
@derive traits	Derive Intro
API docs	Moshi.Data, Moshi.Match, Moshi.Derive

Acknowledgements

Moshi is named after the Chinese word 模式 (móshì, “pattern”). Pattern matching design draws from MLStyle. The ADT encoding is inspired by SumTypes, Expronicon, and this Julia discussion on generated structs.