CGAL users discussion list

[Ben Supnik <>]

Hi Hitchem,

> I am not an expert in computational geometry, nor in exact computations. 
> However, I think that the lazy-exact-NT adapter is intended to use 
> sufficient precision for the computation at hand. So its use allows to 
> speed up computations without affecting their accuracy.

Right...it will provide faster answers when it can using intervals.  But 
its judgment on when the interval is safe to use vs. needing to do a 
more complete slow exact calculation might be affected by FP issues.  At 
least, this is my impression from this note in the code:

>  * Interface of the rep classes:
>  * - .approx()      returns Interval_nt<> (assumes rounding=nearest).
>  *                  [ only called from the handle, and declared in the base ]

> I tried to do the same job (computing intersections) by changing from 
> the CGAL::Simple_cartesian<CGAL::Lazy_exact_nt<Gmpq> > kernel to the 
> CGAL::Lazy_exact_nt<Gmpq> kernel (without the Lazy_exact_nt adapter) and 
> i still have the same error as before. Does this mean that absence of 
> evidence is evidence of absence ;-)

Hrm...perhaps both?  It surely indicates that lazy-exact-NT is not the 
problem...I do not know how any settings affect gmpq...I just don't know 
much about gmpq at all.  So I cannot say that you have ruled out 
possible configuration issues, due to my lack of knowledge of the guts 
of the gmp-based numeric type.

cheers
Ben

--
Scenery Home Page: http://scenery.x-plane.com/
Scenery blog: http://xplanescenery.blogspot.com/
Plugin SDK: http://www.xsquawkbox.net/xpsdk/
X-Plane Wiki: http://wiki.x-plane.com/
Scenery mailing list: 

Developer mailing list: