This repository contains an implementation of an event emitter. Working with emitters can be frustrating. These frustrations led me to make an ES2015 based implementation contained in this repository.
Key features:
- Based on
WeakMap
so you don't have to unregister callbacks when you drop references to an emitter. - Unregister a callback using a reference (like
setTimeout
). - Give a registered callback a call count to unregister itself.
See below for why these are important.
This library comes with both UMD and ES2015 versions. If you're using rollup, it'll use the ES2015 version. Node will use the UMD version automatically. Since this library provides UMD and ES2015 versions, it's supported by every major module system (or no module system at all). It has no production dependencies, making it easy to include too!
The EventEmitter
is used to construct new emitter objects. It takes
no arguments.
const emitter = new EventEmitter();
Add a listener function callback
to the emitter for the given name
. name
must be a string. count
is the number of times the event can be called before
the listener is automatically unregistered. count
defaults to Infinity
when
not given. When given it must be a positive integer greater than 1
, or
Infinity
.
A reference
object is returned, which may later be used to unregister the
listener.
// Callback called every time 'event-name' triggered.
emitter.on('event-name', callback);
// Callback each time 'event-name' triggered, unregistering after 10 calls.
emitter.on('event-name', callback, 10);
The count parameter can be used to simulate a Node.js style once
method.
Unregister an event listener using the reference
object returned by
emitter.on
.
// The on method returns a reference object.
const ref = emitter.on('event-name', callback);
// Use the off method to unregister the callback.
emitter.off(ref);
When called with a name string, all events for that name are removed from the emitter. When called without a name string, all events for all names are removed.
Trigger all handlers for the given name with the remaining arguments args
. The
callbacks are called with the emitter as this
.
function testCallback(a, b, c) {
console.log(a, b, c);
}
emitter.on('some-event', testCallback);
emitter.trigger('some-event', 1, 2, 3) // logs: 1, 2, 3
Alias for emitter.trigger
.
For example, if you add an event listener to a Backbone event emitter using
on
, it will stay there until something removes it. Backbone tries to get
around this with listenTo
, which allows the emitter itself to unregister
events in batch. This is no fault of existing implementations. JavaScript itself
made it an impossible problem to solve until recently.
Luckily, one of the earlier features of ES2015 to make it into browsers was
WeakMap
, which allows the garbage collector to clean up members when no other
references to them remain. The Vertebrate event emitter uses these to avoid
memory leaks.
In Node, you might have code like:
import EventEmitter from 'events';
const emitter = new EventEmitter();
function testCallback() {
console.log('Hello, world!');
}
emitter.on('test', testCallback); // Add a listener to the 'test' event.
emitter.removeListener('test', testCallback); // Remove the listener.
That looks fine, but what happens when you add the same callback for an event twice? Does the callback get called twice per emission, or just once? If twice, what happens when you remove the listener? Does it remove both or just one?
This ambiguity bothers me.
When a Vertebrate emitter has a listener registered for an event, it returns a
reference object, a lot like setTimeout
does. Unregistering the event is done
using this reference object:
import EventEmitter from 'vertebrate-event-emitter';
const emitter = new EventEmitter();
function testCallback() {
console.log('Hello, world!');
}
const ref = emitter.on('test', testCallback); // Add a listener.
emitter.off(ref); // Remove the listener. No need to use the event name.
You get a fresh reference object each time a listener is registered, so the ambiguity never arises.