On Wed, 2008-02-27 at 13:00 +0000, Matthew Allum wrote:
> > Most of the pick() implementations in core clutter just use the default
> > implementation in ClutterActor, which draws a rectangle. I can't find
> > any non-container pick() implementation in clutter or tidy that doesn't
> > use that. 
> 
> Right, because its usually enough. A shaped alpha texture could probably
> paint something a little more (but hasn't really been a problem so far)

> Every Clutter actor has a globally unique ID. When picking this ID is
> converted to a color and passed to the actor. The actor then paints its
> silhouette in *just* this color and offscreen. When a paint is complete
> the pixel is read back from this offscreen framebuffer at the location
> required (i.e where the pointer event was) its color is then converted
> back to an ID and then viola a lookup and we can get the actor the
> pointer event was for.  

Ah. Now I feel less confused.

This ID-to-color stuff doesn't need to be done by the pick()
implementation right? Whatever color pick() uses to paint(), everything
will actually be painted in this special color?

-- 
[EMAIL PROTECTED]
www.murrayc.com
www.openismus.com
> 

-- 
To unsubscribe send a mail to [EMAIL PROTECTED]

Reply via email to