CGAL users discussion list

Hello,
I would like to build a kdtree for points that belong to the 
cartesian product R^3 x S^2 
to store and retrieve points with normals efficiently.
If a point is represented as u=(p,n) with p its 3d location and n the unit 
normal on S^2 my distance is the following:
d(u1,u2) = max( ||p1-p2||/r, acos( n1.n2 )/(pi/2) )
where r is a constant.

In the example about kdtree 
35.3.5   Example Illustrating Use of User Defined Point and Distance Class

it seems required that a distance to an iso-rectangle be defined.
So, in my case it seems that I should have my own implementation of an 
iso-rectangle, the analogue of a Kd_tree_rectangle,
right ?

Is-it easy to make my own implementation of such a thing for the cgal kdtree 
implementation, by any chance ?

- Best regards,

Manuel Yguel