3 min read

On JavaScript Type Tricks

Everybody knows that JavaScript is a dynamically typed and multi-paradigm language. After having recently worked with some languages that are statically typed (Golang, Elm), I fully appreciate the benefits and safety that come with having a compiler.

I recently reluctantly added Flow to a React project that I’ve been doing in my spare time, and in so doing I remembered in a flash of bright white light that I had written a function ten years ago for my jsLite JavaScript library that enforced type checking (and parameter length). Ah, those halcyon days of yore when everyone wrote their own JavaScript library/framework! Definitely the most fun I’ve had when programming JavaScript. It was a different time, when little bands of merry men roamed the wilds and frontend developers were expected to know the language itself rather than a framework.

Anyhoo, I thought the idea to force the language to do type checks at runtime was pretty nifty. I worked out an implementation and included it my library, where it augmented the Function.prototype object.* The function was called assert and ensured not only that the function was invoked with the correct types but also the correct number of arguments that the function was expecting. Here is an example:

const foobar = (a, b) {
}.assert(Boolean, Number);

If foobar is called with exactly two arguments, the first being a Boolean type and the second a Number, everything is right with the world. Otherwise, it will throw. Let’s verify that in the console:

foobar(false, 1992);

Uncaught Error: Function arguments do not equal number of expected arguments.
    at <anonymous>:7:23
    at <anonymous>:1:1

Uncaught Error: Function arguments do not equal number of expected arguments.
    at <anonymous>:7:23
    at <anonymous>:1:1

foobar(true, 'asdf');
Uncaught Error: Wrong data type for function argument 2.
    at Array.from.forEach (<anonymous>:12:27)
    at Array.forEach (<anonymous>)
    at <anonymous>:10:35
    at <anonymous>:1:1

The implementation is rather straightforward. The assert function wraps the base function, which is returned as a closure.

Here are the important bits:

  1. The outer wrapper function (foobar) captures a reference to itself and its arguments (the constructors, i.e., Boolean and Number) and returns an inner function (closure).

  2. The closure is invoked (i.e., foobar(false, 42)), and then two checks are made:

    • Its argument length is compared to the expected number of arguments of the wrapper function (fn.length) and throws an exception if it’s not the same.

    • The type of its parameters are checked against those of the arguments captured in the wrapper function (available via the [[Scope]] chain). The equality checks are ensuring that the “constructor” functions are the same.

  3. Finally, if everything is okey dokey, call the wrapper function with the arguments object passed to the closure. Note that each function has a different arguments object, which is why we’re having to capture the outer arguments object in the args free variable that the closure closes over.

Here is the implementation:

Function.prototype.assert = function () {
    const fn = this;
    const args = arguments;

    return function () {
        if (fn.length !== arguments.length) {
            throw new Error('Function arguments do not equal number of expected arguments.');

        Array.from(arguments).forEach((a, i) => {
            if (a.constructor !== args[i]) {
                throw new Error(`Wrong data type for function argument ${++i}.`);

        // Finally, invoke the parent function.
        return fn.apply(this, arguments);


* I know, augmenting native types is “bad”, just like mixing JavaScript with HTML.