css transitions on new elements

In Firefox, it does appear to be a race between layout completing and the CSS transition. Chrome is much more predictable. If I set the class name on a setTimeout(), Chrome always works, Firefox only works if the setTimeout() time is long.

With this code in Firefox (even using the setTimeout()), the text shows immediately:

function f() {
    var a = document.createElement('a');
    a.id = 'id';
    a.innerHTML = ' fading in?';
    document.getElementsByTagName('p')[0].appendChild(a);
    // at this point I expect the span element to be with opacity=0

    setTimeout(function() {
        a.className="fadeIn";
    }, 10);
    return false;
}

But, if I force a reflow by requesting a property that can only be returned after layout, it then starts to work in Firefox:

function f() {
    var a = document.createElement('a');
    a.id = 'id';
    a.innerHTML = ' fading in?';
    document.getElementsByTagName('p')[0].appendChild(a);
    // at this point I expect the span element to be with opacity=0

    // request property that requires layout to force a layout
    var x = a.clientHeight;
    setTimeout(function() {
        a.className="fadeIn";
    }, 10);
    return false;
}

Furthermore, once I’ve request that property to force a layout, I can even remove the setTimeout() and the animation works in Firefox.

function f() {
    var a = document.createElement('a');
    a.id = 'id';
    a.innerHTML = ' fading in?';
    document.getElementsByTagName('p')[0].appendChild(a);
    // at this point I expect the span element to be with opacity=0

    // request property that requires layout to force a layout
    var x = a.clientHeight;
    a.className="fadeIn";
    return false;
}

You can see this last one work here in both Chrome and Firefox: http://jsfiddle.net/jfriend00/phTdt/

And, here’s an article that discusses the phenomenon: http://gent.ilcore.com/2011/03/how-not-to-trigger-layout-in-webkit.html