Remove curve2d.full_circle2 to avoid confusion

[orbingol]

Ref: orbingol/NURBS-Python#67

[orbingol]

Seems like it is working fine but there is an important fine print here: it is not possible to generate a correct full circle using 7 control points. It will be always 3 arcs connected together. The NURBS Book also says the parameterization is not as good as the 9-point full circle.

That was a fun experiment for me at the time of the development, but considering today, I think it is best to remove curve2d.full_circle2 to avoid confusion.

The reference for 7-point full circle is on pages 301 and 302 of The NURBS Book, 2nd Edition.

[cshorler]

actually 7 point is possible - see the later comment.
orbingol/NURBS-Python#67 (comment)

[orbingol]

@cshorler I still don't believe that is a valid full circle. Using 9-points makes sure that you are creating the correct full circle at all times.

I think it would be the best to contribute that script to the examples repo.

[cshorler]

I also made some changes to the remaining function for compatibility with the 6.x branch.

Also - when developing the package name conflicts with a parallel (egg link) install of NURBS-Python / geomdl. I can submit a PR for this as well.

[orbingol]

geomdl-shapes will be merged with 6.x along with a couple of example curves, surfaces and volumes from the examples repository. It will still be maintained for 5.x

[orbingol]

Also - when developing the package name conflicts with a parallel (egg link) install of NURBS-Python / geomdl. I can submit a PR for this as well.

I'd be glad if you could open a new ticket for this.

[cshorler]

I now read the section in the NURBS book on these two circle constructions
The way I interpret it is that the 7 point circle construction is worse than the 9 point circle - due to two factors:

• the parameterisation leads to a less regular interpolation (this can be seen looking at the figures)
• the convex hull is looser

I guess the detail is in the discussion of higher order circles (later in the same chapter) and discontinuities in the basis functions.

[cshorler]

Also - when developing the package name conflicts with a parallel (egg link) install of NURBS-Python / geomdl. I can submit a PR for this as well.

I'd be glad if you could open a new ticket for this.

okay will do