Skip to main content

State Machines

State machine helps you model an application logic in a way that is expressive of domain knowledge.

Getting started​

pnpm i @constellar/core
import { objectMachine, simpleStateMachine } from '@constellar/core'

const machine = simpleStateMachine({
	init: (a: string) => a.length,
	events: {
		inc: ({ n }: { n: number }, s) => n + s,
		step: (_: object, s) => s - 1,
	},
})

Then you can consume your machine, here as an object.

const m = objectMachine(machine('hello')) // m.state -> 5
m.send({ n: 2, type: 'inc' }) // m.state -> 7
m.send('step') // m.state -> 6

Prior art​

  • xState is the most popular implementation of finite state machines in TypeScript. Main differences:
    • We represent state machines as pure function and let integration actually implementing the store and subscription mechanism.
    • Qualitative part of the state (what xState calls the "context") can have a different type depending on the qualitative part of the state (what xState calls "state").
    • We don't have devtools (definitively a strong point of xState).
    • We don't support machine composition yet (working on it).