On Nov 3, 2010, at 8:28 pm, Robert O'Callahan wrote:
> On Thu, Nov 4, 2010 at 4:54 AM, Sylvain Galineau <sylvaing@microsoft.com> wrote:
> But why would it be any different with overflow:auto ? To oversimplify,
> my expectation is that transforms are applied to the laid-out element
> before it's painted. So if the element does not cause overflow before
> transform, I expect it to look the same after transform whether or not
> the transform is causing the parent to overflow.
>
> In Firefox, the post-transform area of an element contributes to the "scrollable overflow area" of its parent. This allows the user to scroll transformed objects into view. This seems desirable to me.
WebKit now does this too, and the spec requires it.
I think what confuses Sylvain is that the content is laid out with the scrollbars already present, and then transformed, but how did it know it was going to overflow before applying the transform?
Simon