On Thu, Jan 9, 2014 at 11:22 PM, Dominic Mazzoni <dmazzoni@google.com>wrote:
> On Thu, Jan 9, 2014 at 5:14 PM, Robert O'Callahan <robert@ocallahan.org>wrote:
>
>> On Fri, Jan 10, 2014 at 2:08 PM, Rik Cabanier <cabanier@gmail.com> wrote:
>>
>>> Why not? If we cache the path, we should be able to draw the focus on
>>> top of the canvas.
>>>
>>
>> Because there might be other content on top of the rendering of the
>> focused element, that should not be obscured by the focus ring. Or the
>> focused element might be supposed to be clipped.
>>
>
> Are we overoptimizing for obscure theoretical cases? What are the chances
> someone wants to draw a *system* focus ring but clip it and overlay other
> elements on top of it? We can provide a good native focus ring.
>
today's focus ring can be clipped and overlayed with other pseudo canvas
elements. I'm sure that is a realistic use case...
It would be unnatural if the ring is drawn on top of other controls. (The
a11y focus would still be on top)
> I wonder if there aren't just as many plausible use cases where the focus
> ring would naturally extend outside of the canvas bounds, and it'd be
> unnatural and annoying for the author to enlarge the canvas just for focus
> rings.
>
Why would you need to enlarge the canvas? The a11y region for the focus
ring can fall outside the canvas but nothing needs to be drawn there.
If the author has a control that's outside the canvas, he probably is
"faking" the scrollbars too so you can scroll it into view