Async/Await with Generators and Promises

Javascript Generator Series

This article is part of a series of articles about ES6 generators. The plan is to explore iterators and generators, starting with what they are and show casing different situations where they can be useful.

  1. Introduction to ES6 Iterators and Generators
  2. Working with Generators
  3. Async/Await with Generators


Generators are not only useful for generating a stream of values but can also be a powerful control structure when paired with an external runner. In this article we will take a stab at implementing async/await using only ES6 tools (they have later been added and are part of ES7).

For anyone unfamiliar with async/await, this is how it looks in practice.

async function myAsync(input) {
  try {
    const res1 = await anAsynchronousOperation(input)
    const res2 = await anotherAsync(res1)
    return res2
  catch (error) {
    // handle the error

myAsync(123).then(result => //...)

The benefit is that your programming style is very sequencial even though the code is not blocking the event loop.

Since we can’t use the special syntax that async/await introduces our code will looks slightly different but with the same semantics.

const myAsync = makeAsync(function* (input) {
  try {
    const res1 = yield anAsynchronousOperation(input)
    const res2 = yield anotherAsync(res1)
    return res2
  } catch (error) {
    // handle the error

myAsync(123).then(result => //...)

The difference is that makeAsync is now a function instead of a keyword and that await has been replaced with yield.

Passing values into the generator

As can be seen in the example above, the generator code gets a value back from calling yield. In the previous articles yield was used to pass values out of the generator. In order to send values into the generator (as return values from yield) we pass a value with the next() method.

We can use this to pass the values of a promise resolution back into the generator once the promise is resolved. In the same way we can use throw() on the iterator to throw an exception in the context of the generator.

// return a value from yield'a value')

// throw an exception in the context of the yield call
iter.throw(new Error('Exception'))

With this knowledge we can set out to implement our makeAsync function!

Implementing makeAsync

The makeAsync function itself is fairly simple. It wraps the passed generator in a function that takes the arguments (input in our example above) and pass them into the generator when called.

function makeAsync(generator) {
	return function (...args) {
		const gen = generator(...args)
		return iterateAsync(gen,

The interesting work will be done in the iterateAsync function that will be called recursively to step through the generator until the generator is done. iterateAsync returns a promise that will be resolved once the generator is done.

Here is the resulting code:

function iterateAsync(gen, { value, done }) {
  if (done) {
    return Promise.resolve(value)

  // If the value is not a promise (for example 'yield 123'), wrap it in one.
  const promise = value instanceof Promise ? value : Promise.resolve(value)
  return promise
    .then(result => iterateAsync(gen,
    .catch(error => iterateAsync(gen, gen.throw(error)))


As far as my testing went, this behaves exactly as async/await with one little caveat.

const resultGen = yield Promise.resolve(100) + 100
const resultAsync = await Promise.resolve(100) + 100
// resultGen = "[object Promise]100"
// resultAsync = 200

In the case of yield, Javascript first evaluates Promise.resolve(100) + 100 which it does by converting both to strings and then sends that to yield. The solution to this is to wrap it in parenthesis.

const resultGen = (yield Promise.resolve(100)) + 100
// resultGen = 200


This article covered using generators and promises to implement async/await which is available in ES7. The main reason for doing this was to learn more about how to work with generators.

The code can be found at:

While finishing up this article I was looking around for a proper implementation of this as it can be quite useful with Node 6.11 (which is still the version used by Google Cloud Functions) I stumbled over an article series by Kyle Simpson.

His excellent talk Coordinated Concurrency: Reactive (Observables) vs. CSP at JS.LA in 2015 was the reason I started learning more about generators in the first place. And since the fourth article in his series ES6 Generators: Complete Series is about CSP with generators, which is what I had planned on writing about next, I suggest reading that instead.