Create a gist now

Instantly share code, notes, and snippets.

@paulirish /rAF.js
Last active May 24, 2018

What would you like to do?
Download ZIP
requestAnimationFrame polyfill
Raw
rAF.js
// http://paulirish.com/2011/requestanimationframe-for-smart-animating/
// http://my.opera.com/emoller/blog/2011/12/20/requestanimationframe-for-smart-er-animating
// requestAnimationFrame polyfill by Erik Möller. fixes from Paul Irish and Tino Zijdel
// MIT license
(function() {
var lastTime = 0;
var vendors = ['ms', 'moz', 'webkit', 'o'];
for(var x = 0; x < vendors.length && !window.requestAnimationFrame; ++x) {
window.requestAnimationFrame = window[vendors[x]+'RequestAnimationFrame'];
window.cancelAnimationFrame = window[vendors[x]+'CancelAnimationFrame']
|| window[vendors[x]+'CancelRequestAnimationFrame'];
}
if (!window.requestAnimationFrame)
window.requestAnimationFrame = function(callback, element) {
var currTime = new Date().getTime();
var timeToCall = Math.max(0, 16 - (currTime - lastTime));
var id = window.setTimeout(function() { callback(currTime + timeToCall); },
timeToCall);
lastTime = currTime + timeToCall;
return id;
};
if (!window.cancelAnimationFrame)
window.cancelAnimationFrame = function(id) {
clearTimeout(id);
};
}());
@charliewalnut

This comment has been minimized.

Show comment Hide comment
@charliewalnut

charliewalnut Jan 12, 2012

Not a bad shim. A few improvements could be made, though:

1.) use Date.now() instead of new Date().getTime(), avoids one object allocation

2.) Date.now() and setTimeout() operate in terms of integer milliseconds, but 60fps doesn't map to an integer number of milliseconds. This tries to map to 16ms/frame, which is 62.5 and will produce aliasing (meaning bad looking frames) every few seconds. See https://lists.webkit.org/pipermail/webkit-dev/2011-September/018095.html for a very wordy description of why. It is possible to do better, however, by tracking the error between the actual callback rate and the target. http://trac.webkit.org/browser/trunk/Source/WebCore/platform/graphics/chromium/cc/CCDelayBasedTimeSource.cpp?rev=104644#L101 describes in detail how to do this with some C++ code

3.) It's slightly better to set the timer before ticking rather than after - you don't have to worry about setTimeout() clamping if the callback takes >12ms to run and the precision errors are less of a concern. The easiest way to do this is to setTimeout() before invoking the callbacks assuming that somebody will call requestAnimationFrame() within the ticks, and then if they don't end up doing that call clearTimeout() after running the callbacks.

charliewalnut commented Jan 12, 2012

Not a bad shim. A few improvements could be made, though:

1.) use Date.now() instead of new Date().getTime(), avoids one object allocation

2.) Date.now() and setTimeout() operate in terms of integer milliseconds, but 60fps doesn't map to an integer number of milliseconds. This tries to map to 16ms/frame, which is 62.5 and will produce aliasing (meaning bad looking frames) every few seconds. See https://lists.webkit.org/pipermail/webkit-dev/2011-September/018095.html for a very wordy description of why. It is possible to do better, however, by tracking the error between the actual callback rate and the target. http://trac.webkit.org/browser/trunk/Source/WebCore/platform/graphics/chromium/cc/CCDelayBasedTimeSource.cpp?rev=104644#L101 describes in detail how to do this with some C++ code

3.) It's slightly better to set the timer before ticking rather than after - you don't have to worry about setTimeout() clamping if the callback takes >12ms to run and the precision errors are less of a concern. The easiest way to do this is to setTimeout() before invoking the callbacks assuming that somebody will call requestAnimationFrame() within the ticks, and then if they don't end up doing that call clearTimeout() after running the callbacks.
@paulirish

This comment has been minimized.

Show comment Hide comment
@paulirish

paulirish Jan 17, 2012

  1. Date.now() doesnt exist in oldIE.. https://developer.mozilla.org/en/JavaScript/Reference/Global_Objects/Date/now so how about +new Date

  2. true. dunno how else we can tackle this in JS land though..

  3. can i ask you to fork this and make those changes? I can then review and pull 'em in.
Owner

paulirish commented Jan 17, 2012

  1. Date.now() doesnt exist in oldIE.. https://developer.mozilla.org/en/JavaScript/Reference/Global_Objects/Date/now so how about +new Date

  2. true. dunno how else we can tackle this in JS land though..

  3. can i ask you to fork this and make those changes? I can then review and pull 'em in.
@geddski

This comment has been minimized.

Show comment Hide comment
@geddski

geddski Jan 20, 2012

Current Chrome (16) supports webkitRequestAnimationFrame but not webkitCancelAnimationFrame (or webkitCancelRequestAnimationFrame). This breaks the polyfill because it starts the animation loop with requestAnimationFrame, but there's no way to stop it. I've patched the polyfill to return the setTimeout in this case

Not sure how to send a pull request for a gist...

geddski commented Jan 20, 2012

Current Chrome (16) supports webkitRequestAnimationFrame but not webkitCancelAnimationFrame (or webkitCancelRequestAnimationFrame). This breaks the polyfill because it starts the animation loop with requestAnimationFrame, but there's no way to stop it. I've patched the polyfill to return the setTimeout in this case

Not sure how to send a pull request for a gist...
@also

This comment has been minimized.

Show comment Hide comment
@also

also Jan 22, 2012

It should be CancelRequestAnimationFrame not RequestCancelAnimationFrame. The only references to RequestCancelAnimationFrame are this gist.

webkitCancelRequestAnimationFrame exists in current Chrome (16.0.912.75)

also commented Jan 22, 2012

It should be CancelRequestAnimationFrame not RequestCancelAnimationFrame. The only references to RequestCancelAnimationFrame are this gist.

webkitCancelRequestAnimationFrame exists in current Chrome (16.0.912.75)
@paulirish

This comment has been minimized.

Show comment Hide comment
@paulirish

paulirish Jan 23, 2012

lol dyslexia. thx also. Fixed.

fwiw, the new name is present in canary.
Owner

paulirish commented Jan 23, 2012

lol dyslexia. thx also. Fixed.

fwiw, the new name is present in canary.
@desandro

This comment has been minimized.

Show comment Hide comment
@desandro

desandro Feb 20, 2012

I ran into an issue with Firefox 10, where mozRequestAnimationFrame is supported, but not an equivalent for cancelAnimationFrame. So I was able to start an animation, but not properly stop it. if cancelAnimationFrame is a deal-breaker for you then https://gist.github.com/1866474 might be more your flavor.

desandro commented Feb 20, 2012

I ran into an issue with Firefox 10, where mozRequestAnimationFrame is supported, but not an equivalent for cancelAnimationFrame. So I was able to start an animation, but not properly stop it. if cancelAnimationFrame is a deal-breaker for you then https://gist.github.com/1866474 might be more your flavor.
@paulirish

This comment has been minimized.

Show comment Hide comment
@paulirish

paulirish Feb 23, 2012

Why cAF is not at all in FF10 is a complete mystery to me. Seems no tickets at all filed about it either...
Owner

paulirish commented Feb 23, 2012

Why cAF is not at all in FF10 is a complete mystery to me. Seems no tickets at all filed about it either...
@Breton

This comment has been minimized.

Show comment Hide comment
@Breton

Breton Mar 5, 2012

couldn't you simulate cAF by wrapping requestAnimationFrame in a function grabs the ID and passes in a wrapped function that only calls its inner function if cancelAnimationFrame has not been called with a particular ID? (which you can check by having your fake cancelAnimation set the passed in ID on a private/closure variable object to true, and then checking that key in the wrap function.)
or maybe that's all just too complicated/too much overhead etc.

Breton commented Mar 5, 2012

couldn't you simulate cAF by wrapping requestAnimationFrame in a function grabs the ID and passes in a wrapped function that only calls its inner function if cancelAnimationFrame has not been called with a particular ID? (which you can check by having your fake cancelAnimation set the passed in ID on a private/closure variable object to true, and then checking that key in the wrap function.)
or maybe that's all just too complicated/too much overhead etc.
@cacheflowe

This comment has been minimized.

Show comment Hide comment
@cacheflowe

cacheflowe Mar 6, 2012

I tried this out on an iOS Phonegap project, and it runs at a really terrible frame rate on my iPhone 4. I switched back to using a simple setTimeout and am getting close to 60fps. The timeToCall calculation is not good on iOS.

cacheflowe commented Mar 6, 2012

I tried this out on an iOS Phonegap project, and it runs at a really terrible frame rate on my iPhone 4. I switched back to using a simple setTimeout and am getting close to 60fps. The timeToCall calculation is not good on iOS.
@taybenlor

This comment has been minimized.

Show comment Hide comment
@taybenlor

taybenlor Mar 13, 2012

I think the timeToCall calculation is mostly useless anyway. It falls apart as soon as you call RAF twice in a row. And as mentioned by the commenter above adds a bunch of overhead.

taybenlor commented Mar 13, 2012

I think the timeToCall calculation is mostly useless anyway. It falls apart as soon as you call RAF twice in a row. And as mentioned by the commenter above adds a bunch of overhead.
@jrus

This comment has been minimized.

Show comment Hide comment
@jrus

jrus Mar 14, 2012

@paulirish [edit: whoops, was looking at a previous version of the gist. Removing some of the criticism here.]

To handle @taybenlor’s concern about calling RAF multiple times in a row, we can do something like:

var targetTime = 0;
if (!window.requestAnimationFrame)
  window.requestAnimationFrame = function(callback) {
    var currentTime = +new Date;
    targetTime = Math.max(targetTime + 16, currentTime);
    var timeoutCb = function() { callback(+new Date); }
    return window.setTimeout(timeoutCb, targetTime - currentTime);
  };

This also clarifies the code a bit (the previous variable names were a bit confusing), and makes sure that the timestamp passed to the callback is correct, because it is determined (by getting +new Date again) right as the callback is being called.

jrus commented Mar 14, 2012

@paulirish [edit: whoops, was looking at a previous version of the gist. Removing some of the criticism here.]

To handle @taybenlor’s concern about calling RAF multiple times in a row, we can do something like:

var targetTime = 0;
if (!window.requestAnimationFrame)
  window.requestAnimationFrame = function(callback) {
    var currentTime = +new Date;
    targetTime = Math.max(targetTime + 16, currentTime);
    var timeoutCb = function() { callback(+new Date); }
    return window.setTimeout(timeoutCb, targetTime - currentTime);
  };

This also clarifies the code a bit (the previous variable names were a bit confusing), and makes sure that the timestamp passed to the callback is correct, because it is determined (by getting +new Date again) right as the callback is being called.
@taybenlor

This comment has been minimized.

Show comment Hide comment
@taybenlor

taybenlor Mar 14, 2012

@jrus I think you misunderstand what is going on

The time passed into the callback is currTime + timeToCall which gives you the time at which we expect the callback to be called (not necessarily when it will actually be called).

taybenlor commented Mar 14, 2012

@jrus I think you misunderstand what is going on

The time passed into the callback is currTime + timeToCall which gives you the time at which we expect the callback to be called (not necessarily when it will actually be called).
@jrus

This comment has been minimized.

Show comment Hide comment
@jrus

jrus Mar 14, 2012

@taybenlor yeah, but the semantics of requestAnimationFrame (as far as I can understand from reading the spec; it’s not entirely clear/explicit what time should be passed to the callback) should send the time when it’s actually called into the callback. [Also, I edited my previous comment a bit: was looking at the previous version of the gist when I first wrote it.]

jrus commented Mar 14, 2012

@taybenlor yeah, but the semantics of requestAnimationFrame (as far as I can understand from reading the spec; it’s not entirely clear/explicit what time should be passed to the callback) should send the time when it’s actually called into the callback. [Also, I edited my previous comment a bit: was looking at the previous version of the gist when I first wrote it.]
@jrus

This comment has been minimized.

Show comment Hide comment
@jrus

jrus Mar 14, 2012

Here’s a cute CoffeeScript port of the whole thing

do ->
    w = window
    for vendor in ['ms', 'moz', 'webkit', 'o']
        break if w.requestAnimationFrame
        w.requestAnimationFrame = w["#{vendor}RequestAnimationFrame"]
        w.cancelAnimationFrame = (w["#{vendor}CancelAnimationFrame"] or
                                  w["#{vendor}CancelRequestAnimationFrame"])

    targetTime = 0
    w.requestAnimationFrame or= (callback) ->
        targetTime = Math.max targetTime + 16, currentTime = +new Date
        w.setTimeout (-> callback +new Date), targetTime - currentTime

    w.cancelAnimationFrame or= (id) -> clearTimeout id

jrus commented Mar 14, 2012

Here’s a cute CoffeeScript port of the whole thing

do ->
    w = window
    for vendor in ['ms', 'moz', 'webkit', 'o']
        break if w.requestAnimationFrame
        w.requestAnimationFrame = w["#{vendor}RequestAnimationFrame"]
        w.cancelAnimationFrame = (w["#{vendor}CancelAnimationFrame"] or
                                  w["#{vendor}CancelRequestAnimationFrame"])

    targetTime = 0
    w.requestAnimationFrame or= (callback) ->
        targetTime = Math.max targetTime + 16, currentTime = +new Date
        w.setTimeout (-> callback +new Date), targetTime - currentTime

    w.cancelAnimationFrame or= (id) -> clearTimeout id
@jrus

This comment has been minimized.

Show comment Hide comment
@jrus

jrus Mar 14, 2012

@Breton, @GEDDesign this version should I think do cAF properly in browsers that only implement rAF, assuming they actually return a unique ID for each call to rAF:

do ->
    w = window
    for vendor in ['ms', 'moz', 'webkit', 'o']
        break if w.requestAnimationFrame
        w.requestAnimationFrame = w["#{vendor}RequestAnimationFrame"]
        w.cancelAnimationFrame = (w["#{vendor}CancelAnimationFrame"] or
                                  w["#{vendor}CancelRequestAnimationFrame"])

    # deal with the case where rAF is built in but cAF is not.
    if w.requestAnimationFrame
        return if w.cancelAnimationFrame
        browserRaf = w.requestAnimationFrame
        canceled = {}
        w.requestAnimationFrame = (callback) ->
            id = browserRaf (time) ->
                if id of canceled then delete canceled[id]
                else callback time
        w.cancelAnimationFrame = (id) -> canceled[id] = true

    # handle legacy browsers which don’t implement rAF
    else
        targetTime = 0
        w.requestAnimationFrame = (callback) ->
            targetTime = Math.max targetTime + 16, currentTime = +new Date
            w.setTimeout (-> callback +new Date), targetTime - currentTime

        w.cancelAnimationFrame = (id) -> clearTimeout id

jrus commented Mar 14, 2012

@Breton, @GEDDesign this version should I think do cAF properly in browsers that only implement rAF, assuming they actually return a unique ID for each call to rAF:

do ->
    w = window
    for vendor in ['ms', 'moz', 'webkit', 'o']
        break if w.requestAnimationFrame
        w.requestAnimationFrame = w["#{vendor}RequestAnimationFrame"]
        w.cancelAnimationFrame = (w["#{vendor}CancelAnimationFrame"] or
                                  w["#{vendor}CancelRequestAnimationFrame"])

    # deal with the case where rAF is built in but cAF is not.
    if w.requestAnimationFrame
        return if w.cancelAnimationFrame
        browserRaf = w.requestAnimationFrame
        canceled = {}
        w.requestAnimationFrame = (callback) ->
            id = browserRaf (time) ->
                if id of canceled then delete canceled[id]
                else callback time
        w.cancelAnimationFrame = (id) -> canceled[id] = true

    # handle legacy browsers which don’t implement rAF
    else
        targetTime = 0
        w.requestAnimationFrame = (callback) ->
            targetTime = Math.max targetTime + 16, currentTime = +new Date
            w.setTimeout (-> callback +new Date), targetTime - currentTime

        w.cancelAnimationFrame = (id) -> clearTimeout id
@superhero

This comment has been minimized.

Show comment Hide comment
@superhero

superhero May 21, 2012

I created a script somewhat based on this gist.
You find the code at: https://github.com/erik-landvall/animator

superhero commented May 21, 2012

I created a script somewhat based on this gist.
You find the code at: https://github.com/erik-landvall/animator
@leodutra

This comment has been minimized.

Show comment Hide comment
@leodutra

leodutra Jul 17, 2012

Suggestion:

var id = window.setTimeout(function() { callback(currTime + timeToCall); }, timeToCall);
lastTime = currTime + timeToCall;
return id;

to

lastTime = currTime + timeToCall;
return window.setTimeout(function() { callback(currTime + timeToCall); }, timeToCall);

leodutra commented Jul 17, 2012

Suggestion:

var id = window.setTimeout(function() { callback(currTime + timeToCall); }, timeToCall);
lastTime = currTime + timeToCall;
return id;

to

lastTime = currTime + timeToCall;
return window.setTimeout(function() { callback(currTime + timeToCall); }, timeToCall);
@rwaldron

This comment has been minimized.

Show comment Hide comment
@rwaldron

rwaldron Aug 9, 2012

Just for fun... https://gist.github.com/1058681

rwaldron commented Aug 9, 2012

Just for fun... https://gist.github.com/1058681
@rma4ok

This comment has been minimized.

Show comment Hide comment
@rma4ok

rma4ok Aug 16, 2012

Hello!

There is my version of the polyfill

It has a lot of size and performance optimizations

rma4ok commented Aug 16, 2012

Hello!

There is my version of the polyfill

It has a lot of size and performance optimizations
@bradley

This comment has been minimized.

Show comment Hide comment
@bradley

bradley Sep 21, 2012

Just some minor tweaks in case anyone is interested: implemented the single var pattern and LeoDutra's suggestion.

(function() {
    var lastTime = 0,
        vendors = ['ms', 'moz', 'webkit', 'o'],
        x,
        length,
        currTime,
        timeToCall;

    for(x = 0, length = vendors.length; x < length && !window.requestAnimationFrame; ++x) {
        window.requestAnimationFrame = window[vendors[x]+'RequestAnimationFrame'];
        window.cancelAnimationFrame = 
          window[vendors[x]+'CancelAnimationFrame'] || window[vendors[x]+'CancelRequestAnimationFrame'];
    }

    if (!window.requestAnimationFrame)
        window.requestAnimationFrame = function(callback, element) {
            currTime = new Date().getTime();
            timeToCall = Math.max(0, 16 - (currTime - lastTime));
            lastTime = currTime + timeToCall;
            return window.setTimeout(function() { callback(currTime + timeToCall); }, 
              timeToCall);
        };

    if (!window.cancelAnimationFrame)
        window.cancelAnimationFrame = function(id) {
            clearTimeout(id);
        };
}());

bradley commented Sep 21, 2012

Just some minor tweaks in case anyone is interested: implemented the single var pattern and LeoDutra's suggestion.

(function() {
    var lastTime = 0,
        vendors = ['ms', 'moz', 'webkit', 'o'],
        x,
        length,
        currTime,
        timeToCall;

    for(x = 0, length = vendors.length; x < length && !window.requestAnimationFrame; ++x) {
        window.requestAnimationFrame = window[vendors[x]+'RequestAnimationFrame'];
        window.cancelAnimationFrame = 
          window[vendors[x]+'CancelAnimationFrame'] || window[vendors[x]+'CancelRequestAnimationFrame'];
    }

    if (!window.requestAnimationFrame)
        window.requestAnimationFrame = function(callback, element) {
            currTime = new Date().getTime();
            timeToCall = Math.max(0, 16 - (currTime - lastTime));
            lastTime = currTime + timeToCall;
            return window.setTimeout(function() { callback(currTime + timeToCall); }, 
              timeToCall);
        };

    if (!window.cancelAnimationFrame)
        window.cancelAnimationFrame = function(id) {
            clearTimeout(id);
        };
}());
@EmilMarian

This comment has been minimized.

Show comment Hide comment
@EmilMarian

EmilMarian Nov 6, 2012

Just downloaded the code, really easy to use and effective.
Thank you!

EmilMarian commented Nov 6, 2012

Just downloaded the code, really easy to use and effective.
Thank you!
@unphased

This comment has been minimized.

Show comment Hide comment
@unphased

unphased Dec 31, 2012

Would it make sense to augment this shim so that the setTimeout fallback uses performance.webkitNow or performance.now to better determine a good value to setTimeout?

unphased commented Dec 31, 2012

Would it make sense to augment this shim so that the setTimeout fallback uses performance.webkitNow or performance.now to better determine a good value to setTimeout?
@jonasfj

This comment has been minimized.

Show comment Hide comment
@jonasfj

jonasfj Jan 2, 2013

Maybe it's horribly overkill, but if you can multiple animations at the same time. It could pay of to use a list and avoid multiple setTimeout's, also making sure the requests runs at the same time.
I posted my approach here: https://gist.github.com/4438815

Although, it's probably overkill for normal usage...

jonasfj commented Jan 2, 2013

Maybe it's horribly overkill, but if you can multiple animations at the same time. It could pay of to use a list and avoid multiple setTimeout's, also making sure the requests runs at the same time.
I posted my approach here: https://gist.github.com/4438815

Although, it's probably overkill for normal usage...
@jalleyne

This comment has been minimized.

Show comment Hide comment
@jalleyne

jalleyne Feb 11, 2013

I've been testing this version out and for some reason i can't cancel the animation request in webkit browsers. It does cancel in Firefox though. Has anyone else run into this?

jalleyne commented Feb 11, 2013

I've been testing this version out and for some reason i can't cancel the animation request in webkit browsers. It does cancel in Firefox though. Has anyone else run into this?
@kikito

This comment has been minimized.

Show comment Hide comment
@kikito

kikito Mar 19, 2013

Note that the latest browsers (confirmed in Chrome, Firefox and IE10) have moved away from Date-based timestamps on requestAnimationFrame. They now all use the new precision API.

If you are having issues during the first 1 or 2 frames, and you are substracting Date.now or something similar, that's the reason.

kikito commented Mar 19, 2013

Note that the latest browsers (confirmed in Chrome, Firefox and IE10) have moved away from Date-based timestamps on requestAnimationFrame. They now all use the new precision API.

If you are having issues during the first 1 or 2 frames, and you are substracting Date.now or something similar, that's the reason.
@deCastongrene

This comment has been minimized.

Show comment Hide comment
@deCastongrene

deCastongrene Mar 26, 2013

What is the licensing for this code? Is it reusable for commercial software?

deCastongrene commented Mar 26, 2013

What is the licensing for this code? Is it reusable for commercial software?
@mizzao

This comment has been minimized.

Show comment Hide comment
@mizzao

mizzao Mar 29, 2013

The rAF / cAF shim above for browsers that don't support it (read: IE9) was completely broken for me. In my app I call cAF if rAF is called before a previous callback completes, otherwise some serious lag happens. With the shim it was basically adding 16 ms for each callback so if a whole bunch of events were fired off, the actual final non-cancelled callback might not happen until many seconds later. Hence, I subtracted the 16 ms for each cAF. This works much better in IE9; the part I changed is as follows

        # handle legacy browsers which don't implement rAF
        targetTime = 0
        w.requestAnimationFrame = (callback) ->
            targetTime = Math.max targetTime + 16, currentTime = +new Date
            w.setTimeout (-> callback +new Date), targetTime - currentTime            

        w.cancelAnimationFrame = (id) ->
            targetTime -= 16
            clearTimeout id

mizzao commented Mar 29, 2013

The rAF / cAF shim above for browsers that don't support it (read: IE9) was completely broken for me. In my app I call cAF if rAF is called before a previous callback completes, otherwise some serious lag happens. With the shim it was basically adding 16 ms for each callback so if a whole bunch of events were fired off, the actual final non-cancelled callback might not happen until many seconds later. Hence, I subtracted the 16 ms for each cAF. This works much better in IE9; the part I changed is as follows

        # handle legacy browsers which don't implement rAF
        targetTime = 0
        w.requestAnimationFrame = (callback) ->
            targetTime = Math.max targetTime + 16, currentTime = +new Date
            w.setTimeout (-> callback +new Date), targetTime - currentTime            

        w.cancelAnimationFrame = (id) ->
            targetTime -= 16
            clearTimeout id
@mgol

This comment has been minimized.

Show comment Hide comment
@mgol

mgol Apr 25, 2013

You can remove the o prefix, it won't ever be implemented.

mgol commented Apr 25, 2013

You can remove the o prefix, it won't ever be implemented.
@yckart

This comment has been minimized.

Show comment Hide comment
@yckart

yckart Apr 27, 2013

You can remove the o prefix, it won't ever be implemented.

@mzgol How do you know that?

yckart commented Apr 27, 2013

You can remove the o prefix, it won't ever be implemented.

@mzgol How do you know that?
@bbudd

This comment has been minimized.

Show comment Hide comment
@bbudd

bbudd May 8, 2013

I know I'm super late to the game here, but I actually needed to adjust this polyfill because I am making use of the time argument -- which mean I not only need to fill the method, I also need to overwrite any window.requestAnimationFrame call that doesn't give me time.

Here's my go (note: I didn't rewrite the cancelAnimationFrame polyfill, so I didn't include it here)

var lastTime = 0,
    vendors = ['ms', 'moz', 'webkit', 'o'],
    polyfill = function(callback, element) {
        var currTime = new Date().getTime(),
            timeToCall = Math.max(0, 16 - (currTime - lastTime)),
            id = window.setTimeout(function() { 
                callback(currTime + timeToCall); 
            }, timeToCall);

        lastTime = currTime + timeToCall;

        return id;
    };
for(var x = 0, xx = vendors.length; x < xx && !window.requestAnimationFrame; ++x) {
    window.requestAnimationFrame = window[vendors[x]+'RequestAnimationFrame'];
    window.cancelAnimationFrame = window[vendors[x]+'CancelAnimationFrame'] 
                               || window[vendors[x]+'CancelRequestAnimationFrame'];
}

if (!window.requestAnimationFrame) {
    window.requestAnimationFrame = polyfill;
} else {
    window.requestAnimationFrame(function() {
        var isTime = false,
            args = Array.prototype.slice.apply(arguments);
        for (var i = 0, ii = args.length; i < ii; i++) {
            if (typeof args === 'number') {
                isTime = true;
                break;
            }
        }
        if (!isTime) {
            window.requestAnimationFrame = polyfill;
        }
    });
}

EDIT: Oh, and @yckart - because opera's moved to webkit. Hard to believe they wouldn't just use -webkit- instead of -o-

bbudd commented May 8, 2013

I know I'm super late to the game here, but I actually needed to adjust this polyfill because I am making use of the time argument -- which mean I not only need to fill the method, I also need to overwrite any window.requestAnimationFrame call that doesn't give me time.

Here's my go (note: I didn't rewrite the cancelAnimationFrame polyfill, so I didn't include it here)

var lastTime = 0,
    vendors = ['ms', 'moz', 'webkit', 'o'],
    polyfill = function(callback, element) {
        var currTime = new Date().getTime(),
            timeToCall = Math.max(0, 16 - (currTime - lastTime)),
            id = window.setTimeout(function() { 
                callback(currTime + timeToCall); 
            }, timeToCall);

        lastTime = currTime + timeToCall;

        return id;
    };
for(var x = 0, xx = vendors.length; x < xx && !window.requestAnimationFrame; ++x) {
    window.requestAnimationFrame = window[vendors[x]+'RequestAnimationFrame'];
    window.cancelAnimationFrame = window[vendors[x]+'CancelAnimationFrame'] 
                               || window[vendors[x]+'CancelRequestAnimationFrame'];
}

if (!window.requestAnimationFrame) {
    window.requestAnimationFrame = polyfill;
} else {
    window.requestAnimationFrame(function() {
        var isTime = false,
            args = Array.prototype.slice.apply(arguments);
        for (var i = 0, ii = args.length; i < ii; i++) {
            if (typeof args === 'number') {
                isTime = true;
                break;
            }
        }
        if (!isTime) {
            window.requestAnimationFrame = polyfill;
        }
    });
}

EDIT: Oh, and @yckart - because opera's moved to webkit. Hard to believe they wouldn't just use -webkit- instead of -o-
@danb-cws

This comment has been minimized.

Show comment Hide comment
@danb-cws

danb-cws May 14, 2013

webkitRequestAnimationFrame is broken in iOS if you try and use it on additional pages, you can reenable it by focusing on the url bar then back to the page. (tested on ios6.1 iPad3)
The code by KrofDrakula fixed the issue for me (although the bug he describes was in an iframe, my issue was with basic page-to-page navigation).
https://gist.github.com/KrofDrakula/5318048

danb-cws commented May 14, 2013

webkitRequestAnimationFrame is broken in iOS if you try and use it on additional pages, you can reenable it by focusing on the url bar then back to the page. (tested on ios6.1 iPad3)
The code by KrofDrakula fixed the issue for me (although the bug he describes was in an iframe, my issue was with basic page-to-page navigation).
https://gist.github.com/KrofDrakula/5318048
@darius

This comment has been minimized.

Show comment Hide comment
@darius

darius May 30, 2013

A simpler fix for the problem mizzao pointed out: update lastTime after the timeout instead of immediately, so you never have to undo it. Also fixed to make sure request and cancelAnimationFrame are defined consistently, plus gratuitous style changes.

(function() {
    var vendors = ['webkit', 'moz'];
    for (var i = 0; i < vendors.length && !window.requestAnimationFrame; ++i) {
        var vp = vendors[i];
        window.requestAnimationFrame = window[vp+'RequestAnimationFrame'];
        window.cancelAnimationFrame = (window[vp+'CancelAnimationFrame']
                                   || window[vp+'CancelRequestAnimationFrame']);
    }
    if (!window.requestAnimationFrame || !window.cancelAnimationFrame) {
        var lastTime = 0;
        window.requestAnimationFrame = function(callback) {
            var now = new Date().getTime();
            var nextTime = Math.max(lastTime + 16, now);
            return setTimeout(function() { callback(lastTime = nextTime); },
                              nextTime - now);
        };
        window.cancelAnimationFrame = clearTimeout;
    }
}());

darius commented May 30, 2013

A simpler fix for the problem mizzao pointed out: update lastTime after the timeout instead of immediately, so you never have to undo it. Also fixed to make sure request and cancelAnimationFrame are defined consistently, plus gratuitous style changes.

(function() {
    var vendors = ['webkit', 'moz'];
    for (var i = 0; i < vendors.length && !window.requestAnimationFrame; ++i) {
        var vp = vendors[i];
        window.requestAnimationFrame = window[vp+'RequestAnimationFrame'];
        window.cancelAnimationFrame = (window[vp+'CancelAnimationFrame']
                                   || window[vp+'CancelRequestAnimationFrame']);
    }
    if (!window.requestAnimationFrame || !window.cancelAnimationFrame) {
        var lastTime = 0;
        window.requestAnimationFrame = function(callback) {
            var now = new Date().getTime();
            var nextTime = Math.max(lastTime + 16, now);
            return setTimeout(function() { callback(lastTime = nextTime); },
                              nextTime - now);
        };
        window.cancelAnimationFrame = clearTimeout;
    }
}());
@darius

This comment has been minimized.

Show comment Hide comment
@darius

darius May 30, 2013

A couple people have offered versions that collect multiple callbacks into a list all called for the same frame -- my code should handle the same usage without any extra fuss, assuming multiple setTimeouts with the same target time will get called without too much time in between. (I wonder if they will.)

darius commented May 30, 2013

A couple people have offered versions that collect multiple callbacks into a list all called for the same frame -- my code should handle the same usage without any extra fuss, assuming multiple setTimeouts with the same target time will get called without too much time in between. (I wonder if they will.)
@stakach

This comment has been minimized.

Show comment Hide comment
@stakach

stakach Jun 17, 2013

Added cancellation support to @darius's mod

(function(window) {
    'use strict';

    var lastTime = 0,
        vendors = ['moz', 'webkit', 'o', 'ms'],
        x;

    // Remove vendor prefixing if prefixed and break early if not
    for (x = 0; x < vendors.length && !window.requestAnimationFrame; x += 1) {
        window.requestAnimationFrame = window[vendors[x] + 'RequestAnimationFrame'];
        window.cancelAnimationFrame = window[vendors[x] + 'CancelAnimationFrame']
                                   || window[vendors[x] + 'CancelRequestAnimationFrame'];
    }

    // Check if full standard supported
    if (!window.cancelAnimationFrame) {
        // Check if standard partially supported
        if (!window.requestAnimationFrame) {
            // No support, emulate standard
            window.requestAnimationFrame = function(callback) {
                var now = new Date().getTime(),
                    nextTime = Math.max(lastTime + 16, now);

                return window.setTimeout(function() { callback(lastTime = nextTime); }, nextTime - now);
            };

            window.cancelAnimationFrame = window.clearTimeout;
        } else {
            // Emulate cancel for browsers that don't support it
            vendors = window.requestAnimationFrame;
            lastTime = {};

            window.requestAnimationFrame = function(callback) {
                var id = x; // Generate the id (x is initialized in the for loop above)
                x += 1;
                lastTime[id] = callback;

                // Call the vendors requestAnimationFrame implementation
                vendors(function(timestamp) {
                    if (lastTime.hasOwnProperty(id)) {
                        var error;
                        try {
                            lastTime[id](timestamp);
                        } catch (e) {
                            error = e;
                        } finally {
                            delete lastTime[id];
                            if (error) { throw error; }         // re-throw the error if an error occurred
                        }
                    }
                });

                // return the id for cancellation capabilities
                return id;
            };

            window.cancelAnimationFrame = function(id) {
                delete lastTime[id];
            };
        }
    }

}(this));

stakach commented Jun 17, 2013

Added cancellation support to @darius's mod

(function(window) {
    'use strict';

    var lastTime = 0,
        vendors = ['moz', 'webkit', 'o', 'ms'],
        x;

    // Remove vendor prefixing if prefixed and break early if not
    for (x = 0; x < vendors.length && !window.requestAnimationFrame; x += 1) {
        window.requestAnimationFrame = window[vendors[x] + 'RequestAnimationFrame'];
        window.cancelAnimationFrame = window[vendors[x] + 'CancelAnimationFrame']
                                   || window[vendors[x] + 'CancelRequestAnimationFrame'];
    }

    // Check if full standard supported
    if (!window.cancelAnimationFrame) {
        // Check if standard partially supported
        if (!window.requestAnimationFrame) {
            // No support, emulate standard
            window.requestAnimationFrame = function(callback) {
                var now = new Date().getTime(),
                    nextTime = Math.max(lastTime + 16, now);

                return window.setTimeout(function() { callback(lastTime = nextTime); }, nextTime - now);
            };

            window.cancelAnimationFrame = window.clearTimeout;
        } else {
            // Emulate cancel for browsers that don't support it
            vendors = window.requestAnimationFrame;
            lastTime = {};

            window.requestAnimationFrame = function(callback) {
                var id = x; // Generate the id (x is initialized in the for loop above)
                x += 1;
                lastTime[id] = callback;

                // Call the vendors requestAnimationFrame implementation
                vendors(function(timestamp) {
                    if (lastTime.hasOwnProperty(id)) {
                        var error;
                        try {
                            lastTime[id](timestamp);
                        } catch (e) {
                            error = e;
                        } finally {
                            delete lastTime[id];
                            if (error) { throw error; }         // re-throw the error if an error occurred
                        }
                    }
                });

                // return the id for cancellation capabilities
                return id;
            };

            window.cancelAnimationFrame = function(id) {
                delete lastTime[id];
            };
        }
    }

}(this));
@ngryman

This comment has been minimized.

Show comment Hide comment
@ngryman

ngryman Jul 7, 2013

Just published an slightly modified version of this polyfill on bower and jamjs if you use those package managers: https://github.com/ngryman/raf.js.

ngryman commented Jul 7, 2013

Just published an slightly modified version of this polyfill on bower and jamjs if you use those package managers: https://github.com/ngryman/raf.js.
@andrewdeandrade

This comment has been minimized.

Show comment Hide comment
@andrewdeandrade

andrewdeandrade Jul 13, 2013

I went through all the comments and revisions. Of all the iterations posted (or not), is there a best version of the polyfill to use as of July 12th 2013?

andrewdeandrade commented Jul 13, 2013

I went through all the comments and revisions. Of all the iterations posted (or not), is there a best version of the polyfill to use as of July 12th 2013?
@matthew-dean

This comment has been minimized.

Show comment Hide comment
@matthew-dean

matthew-dean Jul 19, 2013

As @danb-66 noted, requestAnimationFrame is completely borked in a variety of scenarios: iframes, page-to-page navigation, and, in our observation, in conjunction with infinitely running CSS animations.

CanIuse.com mentions that requestAnimationFrame has bugs. Other notable mentions:
http://shitwebkitdoes.tumblr.com/post/47186945856/native-requestanimationframe-broken-on-ios-6
http://forums.greensock.com/topic/6639-tweens-not-working-on-ios6/?p=24455#entry24455

The problem has not been fixed (thus far) in iOS7. IMO, this shim should detect iOS and simply not support it, for the time being.

matthew-dean commented Jul 19, 2013

As @danb-66 noted, requestAnimationFrame is completely borked in a variety of scenarios: iframes, page-to-page navigation, and, in our observation, in conjunction with infinitely running CSS animations.

CanIuse.com mentions that requestAnimationFrame has bugs. Other notable mentions:
http://shitwebkitdoes.tumblr.com/post/47186945856/native-requestanimationframe-broken-on-ios-6
http://forums.greensock.com/topic/6639-tweens-not-working-on-ios6/?p=24455#entry24455

The problem has not been fixed (thus far) in iOS7. IMO, this shim should detect iOS and simply not support it, for the time being.
@paulirish

This comment has been minimized.

Show comment Hide comment
@paulirish

paulirish Aug 30, 2013

I think it's time we move this to a repo. :D I won't be able to maintain it so I'm looking for some responsible people.

Also we need to address the iOS issues. I haven't researched them in-depth myself but we need to resolve them ASAP.

@KrofDrakula (his iOS-fix fork) or @darius .. either of you interested?
Owner

paulirish commented Aug 30, 2013

I think it's time we move this to a repo. :D I won't be able to maintain it so I'm looking for some responsible people.

Also we need to address the iOS issues. I haven't researched them in-depth myself but we need to resolve them ASAP.

@KrofDrakula (his iOS-fix fork) or @darius .. either of you interested?
@geedew

This comment has been minimized.

Show comment Hide comment
@geedew

geedew Sep 4, 2013

@paulirish I would be happy to maintain it; I'm about to release code at work (SproutSocial) that relies on a very similar polyfil)

geedew commented Sep 4, 2013

@paulirish I would be happy to maintain it; I'm about to release code at work (SproutSocial) that relies on a very similar polyfil)
@davevv

This comment has been minimized.

Show comment Hide comment
@davevv

davevv Sep 14, 2013

very nice, thanks for sharing

i made two little modifications, instead of checking for the method to be falsy like this:

!window.requestAnimationFrame

i check for it to be explicitly undefined

window.requestAnimationFrame === undefined

and also when declaring vars i do it in a single var statement:

var lastTime = 0,vendors = ['ms', 'moz', 'webkit', 'o'],x;

davevv commented Sep 14, 2013

very nice, thanks for sharing

i made two little modifications, instead of checking for the method to be falsy like this:

!window.requestAnimationFrame

i check for it to be explicitly undefined

window.requestAnimationFrame === undefined

and also when declaring vars i do it in a single var statement:

var lastTime = 0,vendors = ['ms', 'moz', 'webkit', 'o'],x;
@darius

This comment has been minimized.

Show comment Hide comment
@darius

darius Sep 16, 2013

@paulirish, I just made a repo at https://github.com/darius/requestAnimationFrame. It has KrofDrakula's test merged into my version above, though I haven't tested that on an iOS device yet.

darius commented Sep 16, 2013

@paulirish, I just made a repo at https://github.com/darius/requestAnimationFrame. It has KrofDrakula's test merged into my version above, though I haven't tested that on an iOS device yet.
@kof

This comment has been minimized.

Show comment Hide comment
@kof

kof Sep 22, 2013

just wrote an even better shim https://github.com/kof/animation-frame

kof commented Sep 22, 2013

just wrote an even better shim https://github.com/kof/animation-frame
@Gaubee

This comment has been minimized.

Show comment Hide comment
@Gaubee

Gaubee Oct 15, 2013

  • How about:
var now = Date.now || function() {
        return +new Date
    }
  • And the long list of strings and attribute name is not friendly to compression,recommended to use the variable string concatenation if readability is in the extent permitted.

For example:

var _KEY_AnimationFrame = 'AnimationFrame',
    _KEY_requestAnimationFrame = 'request' + _KEY_AnimationFrame;
window[_KEY_requestAnimationFrame] = function(){/*............*/}
  • But as an shim, readability is consider can not do, need to be more concerned about is the compression ratio.

It is my fork: https://gist.github.com/Gaubee/6991570

Gaubee commented Oct 15, 2013

  • How about:
var now = Date.now || function() {
        return +new Date
    }
  • And the long list of strings and attribute name is not friendly to compression,recommended to use the variable string concatenation if readability is in the extent permitted.

For example:

var _KEY_AnimationFrame = 'AnimationFrame',
    _KEY_requestAnimationFrame = 'request' + _KEY_AnimationFrame;
window[_KEY_requestAnimationFrame] = function(){/*............*/}
  • But as an shim, readability is consider can not do, need to be more concerned about is the compression ratio.

It is my fork: https://gist.github.com/Gaubee/6991570
@Ugoku

This comment has been minimized.

Show comment Hide comment
@Ugoku

Ugoku Jan 16, 2014

FWIW, -o-requestAnimationFrame and -ms-requestAnimationFrame have never been used. Both browsers (IE 10+ and Opera 15+) use the unprefixed version. Firefox supports the unprefixed version since FF23, which was released last August (we are now 3 versions further).
So lines 10-15 can be rewritten to:

if (!window.requestAnimationFrame)
{
    window.requestAnimationFrame = window['webkitRequestAnimationFrame'];
    window.cancelAnimationFrame = window['webkitCancelAnimationFrame'] || window['webkitCancelRequestAnimationFrame'];
}

Ugoku commented Jan 16, 2014

FWIW, -o-requestAnimationFrame and -ms-requestAnimationFrame have never been used. Both browsers (IE 10+ and Opera 15+) use the unprefixed version. Firefox supports the unprefixed version since FF23, which was released last August (we are now 3 versions further).
So lines 10-15 can be rewritten to:

if (!window.requestAnimationFrame)
{
    window.requestAnimationFrame = window['webkitRequestAnimationFrame'];
    window.cancelAnimationFrame = window['webkitCancelAnimationFrame'] || window['webkitCancelRequestAnimationFrame'];
}
@pseidemann

This comment has been minimized.

Show comment Hide comment
@pseidemann

pseidemann Feb 4, 2014

I did cleaned up the original a little bit:

// http://paulirish.com/2011/requestanimationframe-for-smart-animating/
// http://my.opera.com/emoller/blog/2011/12/20/requestanimationframe-for-smart-er-animating

// requestAnimationFrame polyfill by Erik Möller. fixes from Paul Irish and Tino Zijdel

// MIT license

(function (window, rAF, cAF) {
    var lastTime = 0, vendors = ['ms', 'moz', 'webkit', 'o'], x;

    for (x = 0; x < vendors.length && !window[rAF]; ++x) {
        window[rAF] = window[vendors[x] + 'RequestAnimationFrame'];
        window[cAF] = window[vendors[x] + 'CancelAnimationFrame']
            || window[vendors[x] + 'CancelRequestAnimationFrame'];
    }

    if (!window[rAF]) {
        window[rAF] = function (callback) {
            var currTime = new Date().getTime(),
                timeToCall = Math.max(0, 16 - (currTime - lastTime)),
                id = window.setTimeout(function () { callback(currTime + timeToCall); }, timeToCall);

            lastTime = currTime + timeToCall;

            return id;
        };
    }

    if (!window[cAF]) {
        window[cAF] = function (id) {
            window.clearTimeout(id);
        };
    }
}(this, 'requestAnimationFrame', 'cancelAnimationFrame'));

which compresses to 423 bytes:

!function(a,b,c){var f,d=0,e=["ms","moz","webkit","o"];for(f=0;f<e.length&&!a[b];++f)a[b]=a[e[f]+"RequestAnimationFrame"],a[c]=a[e[f]+"CancelAnimationFrame"]||a[e[f]+"CancelRequestAnimationFrame"];a[b]||(a[b]=function(b){var c=(new Date).getTime(),e=Math.max(0,16-(c-d)),f=a.setTimeout(function(){b(c+e)},e);return d=c+e,f}),a[c]||(a[c]=function(b){a.clearTimeout(b)})}(this,"requestAnimationFrame","cancelAnimationFrame");

pseidemann commented Feb 4, 2014

I did cleaned up the original a little bit:

// http://paulirish.com/2011/requestanimationframe-for-smart-animating/
// http://my.opera.com/emoller/blog/2011/12/20/requestanimationframe-for-smart-er-animating

// requestAnimationFrame polyfill by Erik Möller. fixes from Paul Irish and Tino Zijdel

// MIT license

(function (window, rAF, cAF) {
    var lastTime = 0, vendors = ['ms', 'moz', 'webkit', 'o'], x;

    for (x = 0; x < vendors.length && !window[rAF]; ++x) {
        window[rAF] = window[vendors[x] + 'RequestAnimationFrame'];
        window[cAF] = window[vendors[x] + 'CancelAnimationFrame']
            || window[vendors[x] + 'CancelRequestAnimationFrame'];
    }

    if (!window[rAF]) {
        window[rAF] = function (callback) {
            var currTime = new Date().getTime(),
                timeToCall = Math.max(0, 16 - (currTime - lastTime)),
                id = window.setTimeout(function () { callback(currTime + timeToCall); }, timeToCall);

            lastTime = currTime + timeToCall;

            return id;
        };
    }

    if (!window[cAF]) {
        window[cAF] = function (id) {
            window.clearTimeout(id);
        };
    }
}(this, 'requestAnimationFrame', 'cancelAnimationFrame'));

which compresses to 423 bytes:

!function(a,b,c){var f,d=0,e=["ms","moz","webkit","o"];for(f=0;f<e.length&&!a[b];++f)a[b]=a[e[f]+"RequestAnimationFrame"],a[c]=a[e[f]+"CancelAnimationFrame"]||a[e[f]+"CancelRequestAnimationFrame"];a[b]||(a[b]=function(b){var c=(new Date).getTime(),e=Math.max(0,16-(c-d)),f=a.setTimeout(function(){b(c+e)},e);return d=c+e,f}),a[c]||(a[c]=function(b){a.clearTimeout(b)})}(this,"requestAnimationFrame","cancelAnimationFrame");
@knod

This comment has been minimized.

Show comment Hide comment
@knod

knod Feb 5, 2014

@paulirish - Is this being hosted somewhere new now? There are links to various repos, but I couldn't tell which one of them is "officially sanctioned".

knod commented Feb 5, 2014

@paulirish - Is this being hosted somewhere new now? There are links to various repos, but I couldn't tell which one of them is "officially sanctioned".
@stoikerty

This comment has been minimized.

Show comment Hide comment
@stoikerty

stoikerty Mar 6, 2014

@knod - As far as I can tell, the requestAnimationFrame polyfill-repo from @darius is the most up-to-date one.

If you want sub-millisecond precision, you can try adding the performance.now polyfill from @timhall after @darius's one. @timhall also uses the more compatible shorthand notation instead of Date.now().

var hasPerformance = !!(window.performance && window.performance.now);

// Add new wrapper for browsers that don't have performance
if (!hasPerformance) {
    // Store reference to existing rAF and initial startTime
    var rAF = window.requestAnimationFrame,
        startTime = +new Date;

    // Override window rAF to include wrapped callback
    window.requestAnimationFrame = function (callback, element) {
        // Wrap the given callback to pass in performance timestamp
        var wrapped = function (timestamp) {
            // Get performance-style timestamp
            var performanceTimestamp = (timestamp < 1e12) 
                ? timestamp 
                : timestamp - startTime;

            return callback(performanceTimestamp);
        };

        // Call original rAF with wrapped callback
        rAF(wrapped, element);
    }        
}

stoikerty commented Mar 6, 2014

@knod - As far as I can tell, the requestAnimationFrame polyfill-repo from @darius is the most up-to-date one.

If you want sub-millisecond precision, you can try adding the performance.now polyfill from @timhall after @darius's one. @timhall also uses the more compatible shorthand notation instead of Date.now().

var hasPerformance = !!(window.performance && window.performance.now);

// Add new wrapper for browsers that don't have performance
if (!hasPerformance) {
    // Store reference to existing rAF and initial startTime
    var rAF = window.requestAnimationFrame,
        startTime = +new Date;

    // Override window rAF to include wrapped callback
    window.requestAnimationFrame = function (callback, element) {
        // Wrap the given callback to pass in performance timestamp
        var wrapped = function (timestamp) {
            // Get performance-style timestamp
            var performanceTimestamp = (timestamp < 1e12) 
                ? timestamp 
                : timestamp - startTime;

            return callback(performanceTimestamp);
        };

        // Call original rAF with wrapped callback
        rAF(wrapped, element);
    }        
}
@pyrsmk

This comment has been minimized.

Show comment Hide comment
@pyrsmk

pyrsmk Mar 15, 2014

Currently, I found these :

I can't tell which code is included in another one or not, and which is better... Maybe @kof has the answer ^^

pyrsmk commented Mar 15, 2014

Currently, I found these :

I can't tell which code is included in another one or not, and which is better... Maybe @kof has the answer ^^
@kof

This comment has been minimized.

Show comment Hide comment
@kof

kof Mar 15, 2014

I have created my library after reading all of this shims/polyfills. It includes all improvements and some additional features, like custom frame rate, feature detected raf (required in cases where raf is implemented but just doesn't work), using raf from the iframe, increased performance for multiple parallel animations when using non native version by grouping callback calls and using just one timer ...

https://github.com/kof/animation-frame

kof commented Mar 15, 2014

I have created my library after reading all of this shims/polyfills. It includes all improvements and some additional features, like custom frame rate, feature detected raf (required in cases where raf is implemented but just doesn't work), using raf from the iframe, increased performance for multiple parallel animations when using non native version by grouping callback calls and using just one timer ...

https://github.com/kof/animation-frame
@pyrsmk

This comment has been minimized.

Show comment Hide comment
@pyrsmk

pyrsmk Mar 15, 2014

Thanks a lot for your quick answer, now I can move forward in my project with serenity ;)

pyrsmk commented Mar 15, 2014

Thanks a lot for your quick answer, now I can move forward in my project with serenity ;)
@Bnaya

This comment has been minimized.

Show comment Hide comment
@Bnaya

Bnaya Apr 23, 2014

I think that the approach of using new timer for each callback can cause unexpected behaviour.
Say you have several elements you want to hide,
if each hide() will be on its own timer callback, they might get hidden not on the same frame

You can overcome this using callback queue and single timer.
(its not the full polyfill, just the way of work)

(function (global) {
    var callbacksQueue = [];

    global.setInterval(function () {
        for (var i = 0; i < callbacksQueue.length; i++) {
            if (callbacksQueue[i] !== false) {
                callbacksQueue[i].call(null);
            }
        }

        callbacksQueue = [];
    }, 1000 / 60);

    global.requestAnimationFrame = function (callback) {
        return callbacksQueue.push(callback) - 1;
    }

    global.cancelAnimationFrame = function (id) {
        callbacksQueue[id] = false;
    }
}(window));

Bnaya commented Apr 23, 2014

I think that the approach of using new timer for each callback can cause unexpected behaviour.
Say you have several elements you want to hide,
if each hide() will be on its own timer callback, they might get hidden not on the same frame

You can overcome this using callback queue and single timer.
(its not the full polyfill, just the way of work)

(function (global) {
    var callbacksQueue = [];

    global.setInterval(function () {
        for (var i = 0; i < callbacksQueue.length; i++) {
            if (callbacksQueue[i] !== false) {
                callbacksQueue[i].call(null);
            }
        }

        callbacksQueue = [];
    }, 1000 / 60);

    global.requestAnimationFrame = function (callback) {
        return callbacksQueue.push(callback) - 1;
    }

    global.cancelAnimationFrame = function (id) {
        callbacksQueue[id] = false;
    }
}(window));
@CMTegner

This comment has been minimized.

Show comment Hide comment
@CMTegner

CMTegner May 15, 2014

raf (npm) v1.0 was released a few hours ago. It's a CommonJS module (i.e. use it in node or in the browser via browserify) that is loosely based on this gist, but with a polyfill which follows the spec as close as possible.

CMTegner commented May 15, 2014

raf (npm) v1.0 was released a few hours ago. It's a CommonJS module (i.e. use it in node or in the browser via browserify) that is loosely based on this gist, but with a polyfill which follows the spec as close as possible.
@Catnemesis

This comment has been minimized.

Show comment Hide comment
@Catnemesis

Catnemesis Mar 18, 2015

If you use JSLint:

/*global window*/
/*jslint maxlen: 100, plusplus: true, unparam: true */

(function () {
    "use strict";

    var lastTime = 0,
        vendors = ['ms', 'moz', 'webkit', 'o'],
        x;

    for (x = 0; x < vendors.length && !window.requestAnimationFrame; ++x) {
        window.requestAnimationFrame = window[vendors[x] + 'RequestAnimationFrame'];
        window.cancelAnimationFrame  = window[vendors[x] + 'CancelAnimationFrame']
                                   || window[vendors[x] + 'CancelRequestAnimationFrame'];
    }

    if (!window.requestAnimationFrame) {
        window.requestAnimationFrame = function (callback, element) {
            var currTime = new Date().getTime(),
                timeToCall = Math.max(0, 16 - (currTime - lastTime)),
                id = window.setTimeout(function () {
                    callback(currTime + timeToCall);
                }, timeToCall);
            lastTime = currTime + timeToCall;
            return id;
        };
    }

    if (!window.cancelAnimationFrame) {
        window.cancelAnimationFrame = function (id) {
            window.clearTimeout(id);
        };
    }
}());

Catnemesis commented Mar 18, 2015

If you use JSLint:

/*global window*/
/*jslint maxlen: 100, plusplus: true, unparam: true */

(function () {
    "use strict";

    var lastTime = 0,
        vendors = ['ms', 'moz', 'webkit', 'o'],
        x;

    for (x = 0; x < vendors.length && !window.requestAnimationFrame; ++x) {
        window.requestAnimationFrame = window[vendors[x] + 'RequestAnimationFrame'];
        window.cancelAnimationFrame  = window[vendors[x] + 'CancelAnimationFrame']
                                   || window[vendors[x] + 'CancelRequestAnimationFrame'];
    }

    if (!window.requestAnimationFrame) {
        window.requestAnimationFrame = function (callback, element) {
            var currTime = new Date().getTime(),
                timeToCall = Math.max(0, 16 - (currTime - lastTime)),
                id = window.setTimeout(function () {
                    callback(currTime + timeToCall);
                }, timeToCall);
            lastTime = currTime + timeToCall;
            return id;
        };
    }

    if (!window.cancelAnimationFrame) {
        window.cancelAnimationFrame = function (id) {
            window.clearTimeout(id);
        };
    }
}());
@davidchambers

This comment has been minimized.

Show comment Hide comment
@davidchambers

davidchambers Apr 14, 2015

What is the element parameter?

window.requestAnimationFrame = function(callback, element) {

It's unused, and according to MDN the function takes just one argument.

davidchambers commented Apr 14, 2015

What is the element parameter?

window.requestAnimationFrame = function(callback, element) {

It's unused, and according to MDN the function takes just one argument.
@techniq

This comment has been minimized.

Show comment Hide comment
@techniq

techniq Jun 2, 2015

To better match the timestamp passed to the requestAnimationFrame callback (ie. window.performance.now()), currTime should be adjusted for the navigationStart

var currTime = new Date().getTime() - window.performance.timing.navigationStart;

Without this change my animations on IE9 were not correctly calculating compared to the browsers with native requestAnimationFrame support.

Note: window.performance.timing is only available on IE9+ (which really only applies to IE9 since requestAnimationFrame is already available on IE10+

techniq commented Jun 2, 2015

To better match the timestamp passed to the requestAnimationFrame callback (ie. window.performance.now()), currTime should be adjusted for the navigationStart

var currTime = new Date().getTime() - window.performance.timing.navigationStart;

Without this change my animations on IE9 were not correctly calculating compared to the browsers with native requestAnimationFrame support.

Note: window.performance.timing is only available on IE9+ (which really only applies to IE9 since requestAnimationFrame is already available on IE10+
@techniq

This comment has been minimized.

Show comment Hide comment
@techniq

techniq Jun 4, 2015

@davidchambers - It was removed from the spec - https://bugs.webkit.org/show_bug.cgi?id=74232

techniq commented Jun 4, 2015

@davidchambers - It was removed from the spec - https://bugs.webkit.org/show_bug.cgi?id=74232
@glued

This comment has been minimized.

Show comment Hide comment
@glued

glued Jul 4, 2015

ES2015 flavored + performance.now()

{

  const vendors = ['ms', 'moz', 'webkit', 'o']
  const af = 'AnimationFrame'
  let lastTime = 0

  if('performance' in window == false)
      window.performance = {}

  if(!Date.now)
    Date.now = () => new Date().getTime()

  if ('now' in window.performance == false){
    let nowOffset = Date.now()

    if(performance.timing && performance.timing.navigationStart)
      nowOffset = performance.timing.navigationStart

    window.performance.now = () => Date.now() - nowOffset
  }

  for(let x = 0; x < vendors.length && !window.requestAnimationFrame; ++x) {
      const vendor = vendors[x]
      window.requestAnimationFrame  = window[`${vendor}Request${af}`]
      window.cancelAnimationFrame   = window[`${vendor}Cancel${af}`] || window[`${vendor}CancelRequest${af}`]
  }

  if(!window.requestAnimationFrame){
    window.requestAnimationFrame = callback =>{
        const currTime    = Date.now()
        const timeToCall  = Math.max(0, 16 - (currTime - lastTime))
        const id          = window.setTimeout(() => callback(currTime + timeToCall), timeToCall)

        lastTime = currTime + timeToCall
        return id
    }
  }

  if(!window.cancelAnimationFrame)
    window.cancelAnimationFrame = id => clearTimeout(id)

}

glued commented Jul 4, 2015

ES2015 flavored + performance.now()

{

  const vendors = ['ms', 'moz', 'webkit', 'o']
  const af = 'AnimationFrame'
  let lastTime = 0

  if('performance' in window == false)
      window.performance = {}

  if(!Date.now)
    Date.now = () => new Date().getTime()

  if ('now' in window.performance == false){
    let nowOffset = Date.now()

    if(performance.timing && performance.timing.navigationStart)
      nowOffset = performance.timing.navigationStart

    window.performance.now = () => Date.now() - nowOffset
  }

  for(let x = 0; x < vendors.length && !window.requestAnimationFrame; ++x) {
      const vendor = vendors[x]
      window.requestAnimationFrame  = window[`${vendor}Request${af}`]
      window.cancelAnimationFrame   = window[`${vendor}Cancel${af}`] || window[`${vendor}CancelRequest${af}`]
  }

  if(!window.requestAnimationFrame){
    window.requestAnimationFrame = callback =>{
        const currTime    = Date.now()
        const timeToCall  = Math.max(0, 16 - (currTime - lastTime))
        const id          = window.setTimeout(() => callback(currTime + timeToCall), timeToCall)

        lastTime = currTime + timeToCall
        return id
    }
  }

  if(!window.cancelAnimationFrame)
    window.cancelAnimationFrame = id => clearTimeout(id)

}
@imthenachoman

This comment has been minimized.

Show comment Hide comment
@imthenachoman

imthenachoman Jul 10, 2015

This was very helpful for me. Thanks! I'm using it in a generic animation library I wrote. I hope this is okay. https://github.com/imthenachoman/nAnimation

imthenachoman commented Jul 10, 2015

This was very helpful for me. Thanks! I'm using it in a generic animation library I wrote. I hope this is okay. https://github.com/imthenachoman/nAnimation
@190n

This comment has been minimized.

Show comment Hide comment
@190n

190n Jul 23, 2015

@imthenachoman Same issue as you, what's the license for this? I'm planning to use this in a game engine I'm writing.

190n commented Jul 23, 2015

@imthenachoman Same issue as you, what's the license for this? I'm planning to use this in a game engine I'm writing.
@190n

This comment has been minimized.

Show comment Hide comment
@190n

190n Jul 23, 2015

@imthenachoman It says it's MIT license; you can probably do whatever you want.

190n commented Jul 23, 2015

@imthenachoman It says it's MIT license; you can probably do whatever you want.
@julienetie

This comment has been minimized.

Show comment Hide comment
@julienetie

julienetie Aug 8, 2015

img
Here's request-frame for anyone looking for a comprehensive solution that should just works on everything.

  • npm / bower request-frame
  • iOS6 fix without user-agent sniffing
  • Normalizes firefox 4 - 10 for cancellation & predictability
  • Polyfill native functions or keep intact via variables (To prevent conflicts with other API usage)
  • IE 5.5+ (tested)
  • AMD support

I believe it's best to use as few requestAnimationFrame instances as possible. I avoided frame rates and delays as they're better utilized outside of the polyfill, (vai the timestamp). Thanks to @darius for the timing tweaks, as well as everyone else on here.

julienetie commented Aug 8, 2015

img
Here's request-frame for anyone looking for a comprehensive solution that should just works on everything.

  • npm / bower request-frame
  • iOS6 fix without user-agent sniffing
  • Normalizes firefox 4 - 10 for cancellation & predictability
  • Polyfill native functions or keep intact via variables (To prevent conflicts with other API usage)
  • IE 5.5+ (tested)
  • AMD support

I believe it's best to use as few requestAnimationFrame instances as possible. I avoided frame rates and delays as they're better utilized outside of the polyfill, (vai the timestamp). Thanks to @darius for the timing tweaks, as well as everyone else on here.
@mikeblakeuk

This comment has been minimized.

Show comment Hide comment
@mikeblakeuk

mikeblakeuk Aug 17, 2015

[20:22:43] bower download https://gist.githubusercontent.com/paulirish/1579671/raw/3d42a3a76ed09890434a07be2212f376959e616f/rAF.js

Anyone else getting this:

Error 503 backend read error

backend read error

Guru Mediation:

Details: cache-lcy1125-LCY 1439843293 3968273849

Varnish cache server

mikeblakeuk commented Aug 17, 2015

[20:22:43] bower download https://gist.githubusercontent.com/paulirish/1579671/raw/3d42a3a76ed09890434a07be2212f376959e616f/rAF.js

Anyone else getting this:

Error 503 backend read error

backend read error

Guru Mediation:

Details: cache-lcy1125-LCY 1439843293 3968273849

Varnish cache server
@lenville

This comment has been minimized.

Show comment Hide comment
@lenville

lenville Sep 7, 2015

Due to an unconfirmed bug in OldIE(maybe the poor performance), many tasks were delaying running.

There are two intended bad Fibonacci demos.

  1. Check the original one in IE8(or lower version), then you may find that timeToCall is gradually growing bigger and bigger.

    raf_without_limitation

    I've made a little change to this code.
    https://gist.github.com/lenville/9e13e63af075c145d662/revisions#diff-0

    // line 20
var timeToCall = Math.max(0, 16 - Math.abs(currTime - lastTime));

  2. Check the revision.

    Now everything works very well, like this:
    raf_with_limitation

lenville commented Sep 7, 2015

Due to an unconfirmed bug in OldIE(maybe the poor performance), many tasks were delaying running.

There are two intended bad Fibonacci demos.

  1. Check the original one in IE8(or lower version), then you may find that timeToCall is gradually growing bigger and bigger.

    raf_without_limitation

    I've made a little change to this code.
    https://gist.github.com/lenville/9e13e63af075c145d662/revisions#diff-0

    // line 20
var timeToCall = Math.max(0, 16 - Math.abs(currTime - lastTime));

  2. Check the revision.

    Now everything works very well, like this:
    raf_with_limitation
@borgeslt

This comment has been minimized.

Show comment Hide comment
@borgeslt

borgeslt Dec 24, 2015

In this article, David Geary implements a interesting version of polyfill.
http://www.ibm.com/developerworks/library/j-html5-game2/#N101B4

What do you think about it?

borgeslt commented Dec 24, 2015

In this article, David Geary implements a interesting version of polyfill.
http://www.ibm.com/developerworks/library/j-html5-game2/#N101B4

What do you think about it?
@milosdjakonovic

This comment has been minimized.

Show comment Hide comment
@milosdjakonovic

milosdjakonovic Oct 30, 2016

I've written a solution that doesn't queue each callback separately but execute all scheduled for next frame simultaneously:

https://github.com/milosdjakonovic/requestAnimationFrame-polyfill

Second (third, fourth...) opinion and eyeball check would do good.

milosdjakonovic commented Oct 30, 2016

I've written a solution that doesn't queue each callback separately but execute all scheduled for next frame simultaneously:

https://github.com/milosdjakonovic/requestAnimationFrame-polyfill

Second (third, fourth...) opinion and eyeball check would do good.
@jlmakes

This comment has been minimized.

Show comment Hide comment
@jlmakes

jlmakes Dec 12, 2016

In most situations I use miniraf ✨

jlmakes commented Dec 12, 2016 •
edited
Edited 4 times
  • @jlmakes jlmakes edited Mar 26, 2018
  • @jlmakes jlmakes edited Mar 26, 2018
  • @jlmakes jlmakes edited Mar 26, 2018
  • @jlmakes jlmakes edited Dec 12, 2016

In most situations I use miniraf ✨
@elizad

This comment has been minimized.

Show comment Hide comment
@elizad

elizad Jan 20, 2017

Maybe I'm still missing something but for me the polyfill is not working right now and I am triong to figure out what is missing

Uncaught SyntaxError: Missing initializer in const declaration

elizad commented Jan 20, 2017 •
edited
Edited 1 time
  • @elizad elizad edited Jan 20, 2017

Maybe I'm still missing something but for me the polyfill is not working right now and I am triong to figure out what is missing

Uncaught SyntaxError: Missing initializer in const declaration
@vennapusaravali

This comment has been minimized.

Show comment Hide comment
@vennapusaravali

vennapusaravali Jul 28, 2017

Hai , i am new to angular2. i have a conflict in loading the angular2 webapp in safari browser. Actually where to use this code in the code so that i can get rid of this browser problem. is this also the solution for improper loading of c3 charts in certain browsers like safari and microsoft edge. please help me out and thanks in advance.

vennapusaravali commented Jul 28, 2017

Hai , i am new to angular2. i have a conflict in loading the angular2 webapp in safari browser. Actually where to use this code in the code so that i can get rid of this browser problem. is this also the solution for improper loading of c3 charts in certain browsers like safari and microsoft edge. please help me out and thanks in advance.
@marcusbelcher

This comment has been minimized.

Show comment Hide comment
@marcusbelcher

marcusbelcher Sep 6, 2017

Hi Vennapusaravali,

More suitable for the Angular 4 repo, however if you checkout this JS file and add the script to angular-cli.jsons "scripts": [ ".../path/to/this/file/rAF.js" ] it should load correctly. More info can be found here.

marcusbelcher commented Sep 6, 2017

Hi Vennapusaravali,

More suitable for the Angular 4 repo, however if you checkout this JS file and add the script to angular-cli.jsons "scripts": [ ".../path/to/this/file/rAF.js" ] it should load correctly. More info can be found here.
@williansmartins

This comment has been minimized.

Show comment Hide comment
@williansmartins

williansmartins Dec 4, 2017

Hello, can someone help me with a question in my website? I don't know where is the issue list, or who I can ask. My website is not working the parallax.js (http://memphisnet.com.br/3.0/)

Thanks advice

williansmartins commented Dec 4, 2017

Hello, can someone help me with a question in my website? I don't know where is the issue list, or who I can ask. My website is not working the parallax.js (http://memphisnet.com.br/3.0/)

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