Skip to content

Instantly share code, notes, and snippets.

@joelambert
Created June 1, 2011 11:03
Show Gist options
  • Save joelambert/1002116 to your computer and use it in GitHub Desktop.
Save joelambert/1002116 to your computer and use it in GitHub Desktop.
Drop in replacements for setTimeout()/setInterval() that makes use of requestAnimationFrame() where possible for better performance
Drop in replace functions for setTimeout() & setInterval() that
make use of requestAnimationFrame() for performance where available
http://www.joelambert.co.uk
Copyright 2011, Joe Lambert.
Free to use under the MIT license.
http://www.opensource.org/licenses/mit-license.php
// requestAnimationFrame() shim by Paul Irish
// http://paulirish.com/2011/requestanimationframe-for-smart-animating/
window.requestAnimFrame = (function() {
return window.requestAnimationFrame ||
window.webkitRequestAnimationFrame ||
window.mozRequestAnimationFrame ||
window.oRequestAnimationFrame ||
window.msRequestAnimationFrame ||
function(/* function */ callback, /* DOMElement */ element){
window.setTimeout(callback, 1000 / 60);
};
})();
/**
* Behaves the same as setInterval except uses requestAnimationFrame() where possible for better performance
* @param {function} fn The callback function
* @param {int} delay The delay in milliseconds
*/
window.requestInterval = function(fn, delay) {
if( !window.requestAnimationFrame &&
!window.webkitRequestAnimationFrame &&
!(window.mozRequestAnimationFrame && window.mozCancelRequestAnimationFrame) && // Firefox 5 ships without cancel support
!window.oRequestAnimationFrame &&
!window.msRequestAnimationFrame)
return window.setInterval(fn, delay);
var start = new Date().getTime(),
handle = new Object();
function loop() {
var current = new Date().getTime(),
delta = current - start;
if(delta >= delay) {
fn.call();
start = new Date().getTime();
}
handle.value = requestAnimFrame(loop);
};
handle.value = requestAnimFrame(loop);
return handle;
}
/**
* Behaves the same as clearInterval except uses cancelRequestAnimationFrame() where possible for better performance
* @param {int|object} fn The callback function
*/
window.clearRequestInterval = function(handle) {
window.cancelAnimationFrame ? window.cancelAnimationFrame(handle.value) :
window.webkitCancelAnimationFrame ? window.webkitCancelAnimationFrame(handle.value) :
window.webkitCancelRequestAnimationFrame ? window.webkitCancelRequestAnimationFrame(handle.value) : /* Support for legacy API */
window.mozCancelRequestAnimationFrame ? window.mozCancelRequestAnimationFrame(handle.value) :
window.oCancelRequestAnimationFrame ? window.oCancelRequestAnimationFrame(handle.value) :
window.msCancelRequestAnimationFrame ? window.msCancelRequestAnimationFrame(handle.value) :
clearInterval(handle);
};
/**
* Behaves the same as setTimeout except uses requestAnimationFrame() where possible for better performance
* @param {function} fn The callback function
* @param {int} delay The delay in milliseconds
*/
window.requestTimeout = function(fn, delay) {
if( !window.requestAnimationFrame &&
!window.webkitRequestAnimationFrame &&
!(window.mozRequestAnimationFrame && window.mozCancelRequestAnimationFrame) && // Firefox 5 ships without cancel support
!window.oRequestAnimationFrame &&
!window.msRequestAnimationFrame)
return window.setTimeout(fn, delay);
var start = new Date().getTime(),
handle = new Object();
function loop(){
var current = new Date().getTime(),
delta = current - start;
delta >= delay ? fn.call() : handle.value = requestAnimFrame(loop);
};
handle.value = requestAnimFrame(loop);
return handle;
};
/**
* Behaves the same as clearTimeout except uses cancelRequestAnimationFrame() where possible for better performance
* @param {int|object} fn The callback function
*/
window.clearRequestTimeout = function(handle) {
window.cancelAnimationFrame ? window.cancelAnimationFrame(handle.value) :
window.webkitCancelAnimationFrame ? window.webkitCancelAnimationFrame(handle.value) :
window.webkitCancelRequestAnimationFrame ? window.webkitCancelRequestAnimationFrame(handle.value) : /* Support for legacy API */
window.mozCancelRequestAnimationFrame ? window.mozCancelRequestAnimationFrame(handle.value) :
window.oCancelRequestAnimationFrame ? window.oCancelRequestAnimationFrame(handle.value) :
window.msCancelRequestAnimationFrame ? window.msCancelRequestAnimationFrame(handle.value) :
clearTimeout(handle);
};
@L2L2L
Copy link

L2L2L commented Jan 26, 2018

And I like for you to consider performance.now() would be a better decision than new Date().now();

here are two sources to help you consider this:

performance date now vs date gettime
https://stackoverflow.com/questions/12517359/performance-date-now-vs-date-gettime

performance now vs date now
https://stackoverflow.com/questions/30795525/performance-now-vs-date-now

It to me sound like what you have is correct but we are looking a performance and it seem from what I read, you can get better result with performance and timing with animation using the performance.now() so since we don't really need to know about the exact time and only for a precision of how long it been... So maybe like I said what you have already is better.

@willryanuk
Copy link

I think this is still useful in 2019, but we do not need all the checks for animationFrame functionality, unless you are supporting some seriously old browsers:
https://caniuse.com/#feat=requestanimationframe

i.e. lots of the above code can be stripped out.

@alex4buba
Copy link

I tried animating a drive with Google api v3.
I took a screen shot, it looks very wrong....
Please see image
Screenshot_20200211-100018_Chrome

@abby-1b
Copy link

abby-1b commented Oct 17, 2021

I've noticed that when switching over from setInterval to requestInterval, my old delay (45 fps, don't judge me, 22.22 ms) was outputting around ~33 fps. Looking into it, this is because the specified delay isn't a multiple of the 60 fps, 16.66 ms delay. To fix this, I used the following:

requestInterval.js ln 23:

start = Date.now() - (delta % delay)

Date.now() is a bit faster, and accounting for the multiple of my delay fixed the issue completely. Without this, I believe even when setting a target delay of 60 fps, 16.66 ms, the actual fps reached would be something like 30-45, since some frames would be skipped at random.

Sorry for bringing back an old thread, just leaving this here for anyone who needs it!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment