do_intersect confusion

[vladgl]

In my project I am using do_intersect function to check intersection/inclusion of Iso_cuboid_3 vs Sphere_3. I've noticed that there is incoming change on the way this function acts with Sphere_3. Seems like now its not a solid ball anymore. So I want to clarify:

1. Is Iso_cuboid_3 always considered a solid object, meaning that do_intersect(sphere, cuboid) == true if cuboid contains sphere (or other object, like point, triangle, etc)?
2. Following this update, is it enough to replace do_intersect(sphere, cuboid) with not has_on_unbounded_side(sphere, cuboid) to achieve the same result: true if their surfaces intersect or if one contains the other?

[MaelRL]

The behavior of intersection functions for Sphere_3 and Circle_2 have indeed been clarified (some were one way, some the other way) to use the mathematical definition of spheres and circles, i.e., a sphere is not a ball, and a circle is not a disk.

1. Yes, Iso_cuboid_3 is a solid object and will return an intersection if another object is (strictly) contained within the cuboid.

2. Yes, not has_on_bounded_side(sphere, cuboid) is a good replacement for the old behavior of the sphere wrongly being considered a ball.

[vladgl]

I've checked, this example will return false, i.e. no intersection:

using K = CGAL::Exact_predicates_exact_constructions_kernel;
using FT = K::FT;
K::Iso_cuboid_3 iso{FT(0), FT(0), FT(0), FT(10), FT(10), FT(10)};
K::Sphere_3 s(K::Point_3{FT(5), FT(5), FT(11)}, FT(4));
std::cout << not s.has_on_unbounded_side(iso) << std::endl;

[vladgl]

@MaelRL If it is not correct behavior, I can post an issue or try to fix it myself

[vladgl]

Yes, this is wrong behavior

cgal/Maintenance/public_release/announcement/mailing-beta.eml

Line 78 in bc96ef7

involving Circle_2 and Sphere_3. The former behavior can be reproduced

[MaelRL]

Fixed with #9494, thank you.