Re: [selectors] [css21] :hover pseudo-class

That is the rule for authors. But what about UA? Is the rule either:
1) always reflow (and flicker)
2) always reflow when mouse moved
3) never reflow
4) reflow but don't recompute pseudo-classes
4) do whatever you want

Whichever we choose, I think we should write something in the spec, for
clarity.

Giovanni

2009/1/26 Tab Atkins Jr. <jackalmage@gmail.com>

> On Mon, Jan 26, 2009 at 7:17 AM, Giovanni Campagna
> <scampa.giovanni@gmail.com> wrote:
> > Look at this CSS example:
> >
> > #my-div {
> > display:block;
> > width:150px;
> > height:150px;
> > background-color:red;
> > }
> > #my-div:hover {
> > width:100px;
> > heigth:100px;
> > }
> >
> > As you see, when the target is hovered, it is shrinked. The interesting
> fact
> > is that after pseudo-class application, the mouse pointer maybe outside
> of
> > the box, so the element won't be hovered any more, so it will get back
> > "default" width and height, so it will be hovered, etc.
> > That is, the current :hover definition may cause an infinite loop, a
> > flashing red box in my example.
> > Also some implementations (WebKit) apply :hover only when the mouse is
> > moved, other (Gecko) instead apply them continuosly.
> > Is this the expected behaviour?
> >
> > CSS21 says:
> > "User agents are not required to reflow a currently displayed document
> due
> > to pseudo-class transitions".
> >
> > Does this mean that any box-model related property (any property that is
> > applied after reflowing) is ignored in a selector with :active, :hover or
> > :focus?
> >
> > Will something be included also in CSS3 Selectors?
>
> It means that they *may* be ignored, if the UA doesn't want the
> complexity and perf hit of reflowing a document like that.  Frex, a
> mobile browser may not have the processing power to do that well.
>
> The simple rule is, don't do anything in a :hover class that might
> change whether or not the target is hovered.  ^_^
>
> ~TJ
>