Hollow sphere has redundant nodes

[HybridDog]

To add only the necessary nodes, I think that the sphere surface needs to be sampled from all three cartesian directions.
If you use only one direction, holes appear:

This happens because the angle between the surface normal and this direction is bigger than 45° (see left of where the arrow points).

If the surface is sampled from all three directions, no holes should appear. Positions at the opposite surface sides are trivial because of the symmetry.

Assume that the surface is sampled in Z direction (an orthographic projection). The height at (x,y) is then sqrt(r*r-x*x-y*y) if (x,y) is on the surface.
The (transposed) gradient is (-x/sqrt(r*r-x*x-y*y), -y/sqrt(r*r-x*x-y*y)). If both the (absolute value of the) x and y component of the gradient are less than one, holes cannot appear, so these positions need to be sampled in the Z direction.
|-x/sqrt(r*r-x*x-y*y)| <= 1, x <= sqrt(r*r-x*x-y*y), x*x <= r*r-x*x-y*y, 2*x*x <= r*r-y*y

forall x, y:
	if x*x+y*y <= r*r  -- on circle, this test is actually redundant
	and 2*x*x <= r*r-y*y and 2*y*y <= r*r-x*x then  -- small gradient
		z = round(sqrt(r*r-x*x-y*y))
		-- add nodes at (x,y,z) and (x,y,-z)
	end
end
-- do the same in X and Y directions

Why does the hollow sphere currently have more nodes than necessary?

[HybridDog]

Are the redundant nodes added so that connected glass domes look more connected?

[sfan5]

No idea but I'm willing to consider this a feature.