Point3D encapsulation of knowledge

[Jones-Adam]

Propose to remove the IntersectionOf, MirrorAbout, ProjectOn methods from Point3D.

These methods require that the fundamental type point should have knowledge of higher order structures like Ray and Plane. Structurally these methods are better placed on the higher order classes rather than point3D

[RiSearcher]

How do you plan to distinguish between Point.MirrorAbout.Plane and Plane.MirrorAbout.Point?

[Jones-Adam]

I understand finding the mirror of a point on the opposite side of a plane. For which I would expect its signature to be myplane.MirrorPoint(Point3D point);

What would mirroring a plane about a point mean? a point doesn't have an opposite side, so I'm not quite following what this should be understood as?

[RiSearcher]

Reflection operation could be performed about point, line or plane. And it is three different operations. Reflection about point also called central inversion.

[Jones-Adam]

Thanks didn't know that From https://en.wikipedia.org/wiki/Point_reflection

In three dimensions, a point reflection can be described as a 180-degree rotation composed with reflection across a plane perpendicular to the axis of rotation

So as I understand it, you would want to do generically

shape.MirrorAbout(Point3D point)
shape.MirrorAbout(Line3D line)
shape.MirrorAbout(Plane plane)

where shape could be a point, a line, a plane, a shape?

I wonder if this is not better supported by providing matrix transforms rather specific operations per type:

shape.ApplyTransform(Matrix m)

[RiSearcher]

Well, yes. I would expect that good library will support all three methods for every shape. The question is how rich and complete API you plan to support in "Math.NET Spatial".

Not all reflections can be described through matrix transforms.

[Jones-Adam]

Well the answer to that has as much to do with people contributing to the library as it does API design. If you want feel free to join us on Gitter https://gitter.im/mathnet/mathnet-spatial. Also if you have the time, it would be great if you could put together a list of operations you would expect to be standard on every shape?

[RiSearcher]

Usually I provide next methods for every geometrical primitive: translate, rotate about point, reflect about point, line and plane. Scaling (for finite objects) probably would be good too, but I haven't implemented it yet.