kew: a lightweight (and super fast) promise/deferred framework for node.js
kew is a lightweight promise framework with an aim of providing a base set of functionality similar to that provided by the Q library.
A few answers (for a few questions)
Why'd we write it?
During our initial usage of Q we found that it was consuming 80% of the cpu under load (primarily in chained database callbacks). We spent some time looking at patching Q and ultimately found that creating our own lightweight library for server-usage would suit our needs better than figuring out how to make a large cross-platform library more performant on one very specific platform.
So this does everything Q does?
Nope! Q is still an awesome library and does way more than kew. We support a tiny subset of the Q functionality (the subset that we happen to use in our actual use cases).
What are Promises?
At its core, a Promise is a promise to return a value at some point in the future. A Promise represents a value that will be (or may return an error if something goes wrong). Promises heavily reduce the complexity of asynchronous coding in node.js-like environments. Example:
// assuming the getUrlContent() function exists and retrieves the content of a urlvar htmlPromise =// we can then filter that through an http parser (our imaginary parseHtml() function) asynchronously (or maybe synchronously, who knows)var tagsPromise = htmlPromise// and then filter it through another function (getLinks()) which retrieves only the link tagsvar linksPromise = tagsPromise// and then parses the actual urls from the links (using parseUrlsFromLinks())var urlsPromise = linksPromise// finally, we have a promise that should only provide us with the urls and will run once all the previous steps have ranurlsPromise
How do I use kew?
As a precursor to all the examples, the following code must be at the top of your page:
var Q =
Convert a literal into a promise
The easiest way to start a promise chain is by creating a new promise with a specified literal using Q.resolve() or Q.reject()
// create a promise which passes a value to the next then() callvar successPromise = Q// create a promise which throws an error to be caught by the next fail() callvar failPromise = Q
In addition, you can create deferreds which can be used if you need to create a promise but resolve it later:
// create the deferredsvar successDefer = Qdefervar failDefer = Qdefer// resolve or reject the defers in 1 second// extract promises from the deferredsvar successPromise = successDeferpromisevar failPromise = failDeferpromise
If you have a node-style callback (taking an Error as the first parameter and a response as the second), you can call the magic
makeNodeResolver() function on a defer to allow the defer to handle the callbacks:
// create the deferredvar defer = Qdefer// some node-style function// grab the outputdeferpromise
Handling successful results with
When a promise is resolved, you may call the
.then() method to retrieve the value of the promise:
.then() will in turn return a promise which will return the results of whatever it returns (asynchronously or not), allowing it to be chained indefinitely:
.then() calls may return promises themselves, allowing for complex nesting of asynchronous calls in a flat manner:
var htmlPromise =var tagsPromise = htmlPromise
Handling errors with
If a promise is rejected for some reason, you may handle the failure case with the
.fail() also returns a promise. If the
.fail() call does not throw an error, it will pass the return value of the
.fail() handler to any
.then() calls chained to it:
If you've reached the end of your promise chain, you may call
.end() which signifies that the promise chain is ended and any errors should be thrown in whatever scope the code is currently in:
getObjectPromise// this will throw an error to the uncaught exception handler if the getObjectPromise call is asynchronous
.fin() when things are finished
You may attach a handler to a promise which will be ran regardless of whether the promise was resolved or rejected (but will only run upon completion). This is useful in the cases where you may have set up resources to run a request and wish to tear them down afterwards.
.fin() will return the promise it is called upon:
var connection = dbvar itemPromise = db
Other utility methods
.all() for many things
If you're waiting for multiple promises to return, you may pass them (mixed in with literals if you desire) into
.all() which will create a promise that resolves successfully with an array of the results of the promises:
var promises =promisespromisespromisesQallpromises
If any of the promises fail, Q.all will fail as well (so make sure to guard your promises with a
.fail() call beforehand if you don't care whether they succeed or not):
var promises =promisespromisespromisesQallpromises
.delay() for future promises
If you need a little bit of delay (such as retrying a method call to a service that is "eventually consistent") before doing something else,
Q.delay() is your friend:
If two arguments are passed, the first will be used as the return value, and the second will be the delay in milliseconds.
.fcall() for delaying a function invocation until the next tick:
// Assume someFn() is a synchronous 2 argument function you want to delay.Q
You can also use
Q.fcall() with functions that return promises.
.nfcall() for Node.js callbacks
Q.nfcall() can be used to convert node-style callbacks into promises:
If your Node-style callback needs a
this context, you can use
.spread() for arrays of promises
() can be used to convert node-style callbacks into promises:
Questions, comments, bug reports, and pull requests are all welcome. Submit them at the project on GitHub.
Bug reports that include steps-to-reproduce (including code) are the best. Even better, make them in the form of pull requests that update the test suite. Thanks!
Copyright 2013 The Obvious Corporation.
Licensed under the Apache License, Version 2.0.
See the top-level file