CGAL users discussion list

[Gärtner Bernd <>]

I agree in principle, but it could still be the case that a too naive 
floating-point implementation is the reason for unnecessarily bad behavior 
under double arithmetic. For example, the planes that Amir sent define a 3x3 
linear system (for computing the intersection point) that seems to be 
well-conditioned; if the solution is still way off, this could mean that the 
solution method cannot properly deal with non-critical degeneracies like the 
x- and y-coordinate of both the first and third plane being almost equal (up 
to scaling).
 
Best,
Bernd.

________________________________

Von: Fernando Cacciola 
[mailto:]
Gesendet: Mo 10.09.2007 16:41
An: 

Betreff: Re: Re: Re: [cgal-discuss] Bug in 3 Planes intersection?



Hello Amir,

One way to see if this is a robustness issue or a bug in CGAL is to try
with:

Exact_predicate_exact_constructions_kernel

If you get incorrect results using that then almost certainly there is a bug
in CGAL.

But conversely, if results are correct using that kernel, then you just have
a robustness issue which you need to solve by following the advice given.

HTH

Fernando Cacciola
GeometyFactory
--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://lists-sop.inria.fr/wws/info/cgal-discuss