I don't have any experience with tensors but it seems, at least on the surface, that the convenience is not worth allowing the bugs that could be prevented by enforcing 1 <= n <= ndims(A).
However if that was the correct way to do things, then I would expect this to work (viewing a scalar as a rank 0 tensor): julia> 5[1] 5 julia> 5[1,1] ERROR: no method getindex(Int64,Int64,Int64)
