CGAL users discussion list

Hello!

First time writing on the mailing list, so I hope it is the right place
to discuss, since I'm not super convinced, whether my finding is a bug
or not. I transform a 3d point into 2D using a plane and transform it
back. I would have expected approximately the same point.

I'm using CGAL with Python SWIG bindings in the newest version (commit
2479e35) on Arch Linux. I took a look at the bindings code and could
not find much code, that could do something wrong. Therefore I suspect
the error being somewhere in the C++ part. 

Here is my example:
```python
from CGAL.CGAL_Kernel import Plane_3, Point_3

# plane with distance 1 in x-direction
plane_3d_x = Plane_3(1, 0, 0, 1)

# this point obviously lies on the plane
point_3d_unit_x = Point_3(1, 0, 0)

# results into a (0,0)
point_2d = plane_3d_x.to_2d(point_3d_unit_x)

print(str(point_2d))

assert point_2d.x() == 0
assert point_2d.y() == 0

# i would have expected the same point (except for minor numerical
errors)
point_3d_retransformed = plane_3d_x.to_3d(point_2d)

print(str(point_3d_retransformed))

# this assertion fails, since x == -1
assert point_3d_retransformed.x() == 1
assert point_3d_retransformed.y() == 0
assert point_3d_retransformed.z() == 0```

Can anyone tell, if this is a bug or explain to me, why this is
correct?

Thanks in advance!