CGAL users discussion list

[Jens Rasmussen <>]

Hi

I've been tampering with the intersection points on segments demo, and
would like to be verified of its validity as only 1-3 of 2000-3000
constructed intersection-points are found to lie on the segments.

I'm familiar with the Robustness issue, so I know it's due to the
underlying data types and the arithmetic used on them -- I've even
traced the primitives used in construction and in the predicate -- but
in: Invited Lecture: Real Numbers and Robustness in Computational
Geometry, the stated numbers from the robustness demo in CGAL 2.x is 20%
and not 0.05%-0.1%. 

Further, I've altered the code so that an 'X' will mark the spot of the
intersection-point(s) that was(were) found to lie on the intersection,
and it is often nowhere near any intersection -- though this might be
due to an error from my part as when drawing the point I do:
 *widget << (*fip)
where *fip is of type CartesianDouble::Point_2 (resp.
CartesianFloat::Point_2) and not C_double::Point_2 (resp.
C_real::Point_2) as is otherwise used in the file robustness.cpp.

Hope you can help!

Best regards
Jens