node package manager


Easily mix asynchronous and synchronous programming styles in node.js


Easily mix asynchronous and synchronous programming styles in node.js.

build status npm version mit license we're hiring


  • Easy-to-follow flow control for both serial and parallel execution
  • Complete stack traces, even for exceptions thrown within callbacks
  • No boilerplate code for error and exception handling
  • Conforms to standard node async API


Fibrous requires node version 0.6.x or greater.

npm install fibrous


Would you rather write this:

var updateUser = function(id, attributes, callback) {
  User.findOne(id, function (err, user) {
    if (err) return callback(err);
    user.set(attributes);, updated) {
      if (err) return callback(err);
      console.log("Updated", updated);
      callback(null, updated);

Or this, which behaves identically to calling code:

var updateUser = fibrous(function(id, attributes) {
  user = User.sync.findOne(id);
  updated =;
  console.log("Updated", updated);
  return updated;

Or even better, with CoffeeScript:

updateUser = fibrous (id, attributes) ->
  user = User.sync.findOne(id)
  updated =

Without Fibrous

Using standard node callback-style APIs without fibrous, we write (from the fs docs):

fs.readFile('/etc/passwd', function (err, data) {
  if (err) throw err;

Using sync

Using fibrous, we write:

data = fs.sync.readFile('/etc/passwd');

Using future

This is the same as writing:

future = fs.future.readFile('/etc/passwd');
data = future.wait();

Waiting for Multiple Futures

Or for multiple files read asynchronously:

futures = [
data = fibrous.wait(futures);
console.log(data[0], data[1]);

Note that fs.sync.readFile is not the same as fs.readFileSync. The latter blocks while the former allows the process to continue while waiting for the file read to complete.

Make It Fibrous

Fibrous uses node-fibers behind the scenes.

wait and sync (which uses wait internally) require that they are called within a fiber. Fibrous provides two easy ways to do this.

1. fibrous Function Wrapper

Pass any function to fibrous and it returns a function that conforms to standard node async APIs with a callback as the last argument. The callback expects err as the first argument and the function result as the second. Any exception thrown will be passed to the callback as an error.

var asynFunc = fibrous(function() {
  return fs.sync.readFile('/etc/passwd');

is functionally equivalent to:

var asyncFunc = function(callback) {
  fs.readFile('/etc/passwd', function(err, data) {
    if (err) return callback(err);
    callback(null, data);

With coffeescript, the fibrous version is even cleaner:

asyncFunc = fibrous ->

fibrous ensures that the passed function is running in an existing fiber (from higher up the call stack) or will create a new fiber if one does not already exist.

2. Express/Connect Middleware

Fibrous provides connect middleware that ensures that every request runs in a fiber. If you are using express, you'll want to use this middleware.

var express = require('express');
var fibrous = require('fibrous');
var app = express();
app.get('/', function(req, res){
  data = fs.sync.readFile('./index.html', 'utf8');

3. Wrap-and-run with is a utility function that creates a fibrous function then executes it.

Provide a callback to handle any errors and the return value of the passed function (if you need it). If you don't provide a callback and there is an error, run will throw the error which will produce an uncaught exception. That may be okay for quick and dirty work but is probably a bad idea in production code. {
  var data = fs.sync.readFile('/etc/passwd');
  return data;
}, function(err, returnValue) {
  console.log("Handle both async and sync errors here", err);

4. Waiting on a callback

Sometimes you need to wait for a callback to happen that does not conform to err, result format (for example streams). In this case the following pattern works well:

var stream = <your stream>
function wait(callback) {
  stream.on('close', function(code) {
    callback(null, code);
var code = wait.sync();


Error Handling / Exceptions

In the above examples, if readFile produces an error, the fibrous versions (both sync and wait) will throw an exception. Additionally, the stack trace will include the stack of the calling code unlike exceptions typically thrown from within callback.


Fibrous provides a test helper for jasmine-node that ensures that beforeEach, it, and afterEach run in a fiber. Require it in your shared spec_helper file or in the spec files where you want to use fibrous.

describe('My Spec', function() {
  it('tests something asynchronous', function() {
    data = fs.sync.readFile('/etc/password');

If an asynchronous method called through fibrous produces an error, the spec helper will fail the spec.

mocha-fibers provides a fiber wrapper for mocha.

If you write a helper for other testing frameworks, we'd love to include it in the project.


Fibrous makes it much easier to work with asynchronous methods in an interactive console, or REPL.

If you find yourself in an interactive session, you can require fibrous so that you can use future.

> fs = require('fs');
> require('fibrous');
> data = fs.future.readFile('/etc/passwd', 'utf8');
> data.get()

In this example, data.get() will return the result of the future, provided you have waited long enough for the future to complete. (The time it takes to type the next line is almost always long enough.)

You can't use sync in the above scenario because a fiber has not been created so you can't call wait on a future.

Fibrous does provide a bin script that creates a new interactive console where each command is run in a fiber so you can use sync. If you install fibrous with npm install -g fibrous or have ./node_modules/.bin on your path, you can just run:

$ fibrous
Starting fibrous node REPL...
> fs = require('fs');
> data = fs.sync.readFile('/etc/passwd', 'utf8');
> console.log(data);
# User Database

Or for a CoffeeScript REPL:

$ fibrous -c [or --coffee]
Starting fibrous coffee REPL...
coffee> fs = require 'fs'
coffee> data = fs.sync.readFile '/etc/passwd', 'utf8'
coffee> console.log data
# User Database


The first time you call sync or future on an object, it builds the sync and future proxies so if you add a method to the object later, it will not be proxied.

With Express and bodyParser or json

You might be getting an error in Express that you are not in context of a fiber even after adding fibrous.middleware to your stack. This can happen if you added it before express.json() or express.bodyParser(). Here's an example:

// might not work 
// or 
// should work 
// or 

Behind The Scenes


Fibrous uses the Future implementation from node-fibers.

future.wait waits for the future to resolve then returns the result while allowing the process to continue. fibrous.wait accepts a single future, multiple future arguments or an array of futures. It returns the result of the future if passed just one, or an array of results if passed multiple.

future.get returns the result of the resolved future or throws an exception if not yet resolved.

Object & Function mixins

Fibrous mixes future and sync into Function.prototype so you can use them directly as in:

readFile = require('fs').readFile;
data = readFile.sync('/etc/passwd');

Fibrous adds future and sync to Object.prototype correctly so they are not enumerable.

These proxy methods also ignore all getters, even those that may return functions. If you need to call a getter with fibrous that returns an asynchronous function, you can do:

func = obj.getter, args)


Some people don't like libraries that mix in to Object.prototype and Function.prototype. If that's how you feel, then fibrous is probably not for you. We've been careful to mix in 'right' so that we don't change property enumeration and find that the benefits of having sync and future available without explicitly wrapping objects or functions are worth the philosophical tradeoffs.


git clone git://
npm install
npm test

Fibrous is written in coffeescript with source in src/ compiled to lib/.

Tests are written with jasmine-node in spec/.

Run tests with npm test which will also compile the coffeescript to lib/.

Pull requests are welcome. Please provide tests for your changes and features. Thanks!