W3C home > Mailing lists > Public > www-style@w3.org > October 2008

Re: Selectors

From: Simon Fraser <smfr@me.com>
Date: Thu, 16 Oct 2008 08:45:21 -0700
Cc: Anne van Kesteren <annevk@opera.com>, W3C Style List <www-style@w3.org>
Message-id: <C41FA6E3-1608-4B3A-AFC0-3BB4738D4381@me.com>
To: Daniel Glazman <daniel.glazman@disruptive-innovations.com>

On Oct 16, 2008, at 8:27 am, Daniel Glazman wrote:

> Anne van Kesteren wrote:
>> On Thu, 16 Oct 2008 15:55:35 +0200, Daniel Glazman <daniel.glazman@disruptive-innovations.com 
>> > wrote:
>>>     * Does :hover apply to parent of hovered element if that element
>>>       is positioned "outside" of its parent?
>> Seems that :hover should rely on hit testing just like mouse  
>> events. Not sure why it matters where an element is positioned.
>
> That's exactly why it is an issue !
> Imagine an element having a sub-element positioned outside of its
> parent. IF :hover works like a mouse event, then it bubbles from the
> original target to the parent. But the mouse event occured OUTSIDE of
> the box of the parent, which is not easy to understand from a CSS
> author's perspective since he/she is not manipulating a mouse event in
> a DOM but a pointer hovering some elements in a window...

I think :hover has to always have the same targeting behavior as
hit testing for mouse events, so that page authors can rely on
:hover style as a way of communicating to the user that clicking
on an element will do something (even if the :hover style and
event handler are on a parent of the hit element). Changing
this would both break existing pages, and complicate implementations.

Now that we have positioning/transforms etc. that can move elements
outside the box of their parents, the DOM behavior of propagating
events up the parent chain seems anachronistic. You could imagine
an alternative, like sending the event to each element at the hit
point in z-order. However, such a behavioral change would be a large
undertaking.

Earlier (<http://lists.w3.org/Archives/Public/www-style/2008Jul/0428.html 
 >)
we proposed that the SVG pointer-events property be applied to HTML.
Both Mozilla and WebKit intend to implement this, and it allows
authors to have some control over hit testing via CSS.

Simon
Received on Thursday, 16 October 2008 15:46:29 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:55:15 GMT