µ-ffsm: Micro fluent API helper

Very tiny helper function to construct fluent interfaces.

> npm i mu-ffsm

Import

var FFSM = require('mu-ffsm');

Create language/machine.

// internal state is an Array
var Talker = FFSM({
	0:    function() { return []; }, // TODO allow const
	talk: function(say, what) { say.push(what); return say; },
	1:    function(say, sep) { return say.join(sep || ' '); }
});

Construct sentences/instances:

var cowboyGreeting = Talker()
	.talk('howdy')
	.talk('cowboy');

// make dramatic
console.log(cowboyGreeting(', ...'));

Idea

We consider a sort of 'state machine' that maintains an internal state of type S. We then consider three types of functions on it:

• entry : * ⟶ S
• transition : S ⟶ * ⟶ S
• exit : S ⟶ * ⟶ *

We write the entry function as 0, the exit function as 1 and then name all the transition functions however we like.

Then

	var M = FFSM({
		0: function(i)    { return /* initial state */; } // entry function
		a: function(s, t) { return /* new state     */; } // transition 'a'
		b: function(s, t) { return /* new state     */; } // transition 'b'
		1: function(s, x) { return /* final value   */; } // exit function
	});

Now

	var i = M(entry)	// x : S <- 0(entry)
		.a(trigger_0)	// y : S <- a(x, trigger_0)
		.b(trigger_1)	// z : S <- b(y, trigger_1)
		.a(trigger_2);  // i : S <- a(z, trigger_2)

Finally

	var y = i(x);		// y <- 1(i, x)

So we have

• First M(entry) creates a new machine instance of type M. It's initial state derived from entry (0(entry)).

• Then .a(trigger_0) transitions the machine with transition a to a new state, using the previous state and the data from trigger_0 to compute the new state.

• Similarly .b(t_1), .a(t_2).

• Finally, the i(x) call constructs an element out of the internal state and the argument using the exit function 1(i,x).