CGAL users discussion list

[David Keller <>]

Dear Ron,

thanks for your fast answer. I was actually not aware of the error in my 
input file, sorry for this. Maybe it would be a good decision to check 
the validity of the point sequence during the curve construction if it 
is not too expensive at all.

I don't explicitly use the constructor _Bezier_point_2 (const Algebraic& 
x, const Algebraic& y), this is true.
So from my point of view your proposal is perfectly ok and would help me 
to use the Bezier-arrangement package in a more efficient way.

Thanks and kind regards,
David


 wrote:
> Dear David,
>
> I checked the assertion violation you have reported.
> What caused the problem is curve #23 in your input file:
> It has 4 control points, the first two are identical (both are (0, 0)).
> This is in fact a forbidden input (two conescutive control points must 
> not
> be the same - perhaps we should add it as a precondition).
> When I fix curve #23 to have 3 control point by removing one (0, 0),
> the arrangement is constructed without any problem.
>
> As for the second issue you have raised:
> I think it is a good idea to introduce the constructor:
>   _Bezier_point_2 (const Rational& x, const Rational& y);
> In fact, in order to perform point-location or to insert isolated points,
> it would be difficult to carry such an operation out if the point has
> algebraic coefficients.
> I think that I'll add the constructor, and hide the constructor for a 
> pair
> of algebraic numbers (unless you explicitly use it, which I doubt).
>
> Best regards,
> Ron