Converting Synchronous Code to Asynchronous

I often find myself looking up patterns for converting synchronous code into async variants in Javascript. Whether it’s remembering how exactly Promise chains work or what I need to do to create a Promise - there always seems to be one part that trips me up.

I wanted to document a simple, but I think representative, example of how to take a synchronous function and convert it to be asynchronous in Javascript.

I’ll be using a node function that is reading a file from the file system.

The original function is:

const findAssetSync = (name) => { const assetPath = path.join(__dirname, 'assets', name) return fs.readFileSync(assetPath, {encoding: 'utf-8'}).toString() } 

The first step is to make this function return a promise instead.

const findAssetAsync = (name) => { const assetPath = path.join(__dirname, 'assets', name) return new Promise((resolve, reject) => { fs.readFile(assetPath, {encoding: 'utf-8'}, (err, data) => { if (err) reject(err); return resolve(data); }) }) } 

Now, let’s look at how this would actually be used. I’ll start with the synchronous version.¹

const server = http.createServer((req, res) => { const route = url.parse(req.url).pathname if (routes[route]) { const assets = findAssetSync(routes[route]) res.write(assets) res.end() } else { res.writeHead(404, ‘Not Found’) res.end() } }) 

To use the asynchronous version, however, we either need to convert the callback within createServer into an Async/Await function or now use a promise chain.

The point, however, is that now, instead of returning the string itself as we do in findAssetSync, findAssetAsync returns a promise.

Using Promise Chain

Promise chains create some complexity. Because we want to to make sure we resolve before moving onto the writing of the server response - we can’t do this:

const server = http.createServer(req, res) => { /* ... */ if (routes[route]) { let assets = findAssetAsync(routes[route]).then(results => { assets = results; }).catch(err => console.error(err)) res.write(assets) res.end() } else { /* ... */ } }) 

This would error, because while the promise is resolving, node would just continue moving along and reading the file — so we’d write assets (which would be undefined at the time) and then end the response.

To handle this - we place the response inside the .then block:

const server = http.createServer(req, res) => { /* ... */ if (routes[route]) { findAssetAsync(routes[route]).then(results => { res.write(results) res.end() }).catch(err => console.error(err)) } else { /* ... */ } }) 

It’s easy to see how, if this were to get much more complicated - and/or you wanted to carry variables forward (instead of just using the “response” variable from a Promise) how this can quickly get messy.²

Using Async/Await

The async await syntax is much simpler to reason through. While it’s not creating synchronous code - it reads as if it is. And underneath, it’s all just using Promises.

const server = http.createServer( async (req, res) => { /* ... */ if (routes[route]) { const assets = await findAssetAsync(routes[route]) res.write(assets) /* ... */ } else { /* ... */ } }) 

That’s it. We’re now waiting for the Async function to resolve before preceeding - all while not blocking other requests.

Conclusion

Converting from synchronous to asynchronous javascript code is not particularly difficult. It’s a matter of understanding what is actually happening with the event loop and then pattern recognition.

Footnotes

• ¹ I have simplified the code here a bit. For example, we’re missing the routes object that I’m looking for the route.
• ² I found Danny Moerkeke’s article on Async Programming in JavaScript particularly helpful in highlighting this challenge.