Figure out what sort of special-casing IE and Chrome do for data: (and what other URIs)?

It looks like the "Check the usability of the image argument" algorithm from the spec returns "good" for data: images in IE and Chrome even when per current spec it should not. Testcase goes like so:

var img = new Image();
img.src = 'data:stuff'; // Need actual valid image data: URI here
var canvas = document.createElement('canvas');
var pattern = canvas.getContext("2d").createPattern(img, 'no-repeat');
canvas.getContext("2d").drawImage(img, 0, 0)

Per spec, pattern should be null here (modulo #655) and there should be nothing painted to the canvas. See https://bugzilla.mozilla.org/show_bug.cgi?id=1164458#c9 for the spec analysis. It sounds like IE and Chrome are doing something else instead. It would be good to understand what, exactly, this "something else" is...

@domenic @travisleithead @jacobrossi any idea who would have that info?

Start with the “Check the usability of the image argument” algorithm and the supplied Image, data: URI, createPattern, and drawImage testcase. Reproduce it with a valid data: image in IE and Chrome, compare the result with the current specification and the linked Bugzilla analysis, and document the browser behavior or required specification change.