Why is sampling something with a single keys, the output must be designated as a field (even though the logging value is single)?

Edit: The problem is that if we are sampling values from the surface of the cube, then whether it is a field or not should not matter. The number of results (single or other domain) does not depend on this.

Edit 2:

auto fn = std::make_unique<RaycastFunction>(std::move(target), std::move(field), mapping);
auto op = std::make_shared<FieldOperation>(FieldOperation(
    std::move(fn),
    {std::move(position_field), std::move(direction_field), std::move(length_field)}));

This line of code should mean exactly that the fact that the output field depends only on the fact that it is created based on position, vector and length.

I see your point. You're probably right but I'm not totally sure. Single values can still be fields that we depend on.

Single values can still be fields that we depend on.

Not sure what you meant

If imagine a little:
You can sample some value. In this case, from the surface of the cube. This value can be either a single value for all points or a field.

Sampling, like querying a value from a cube, is an operation, the amount of which depends on how many keys to sample. That is, there are as many results as there are keys (position, length, vetcore).

If the sample is from a surface with a single value, this simply means that all requests for a value receive the same thing.