Whilst working with the tensor.Iterator functionality I found some strange behavior, when using iterator in a for loop like follows:

it = t.Iterator()  // t is a tensor.Dense
for i, err := it.Start(); err == nil; i, err = it.Next() {
	fmt.Printf("i = %v, coord = %v\n", i, it.Coord())
}

1. When using the iterator as stated above, the 'first' element of the tensor is always last in the iteration. This is not really a big issue if you want to iterate over all elements and order does not matter, but it is weird nonetheless. As an example consider the tensor [1, 2, 3, 4] with shape (2, 2). It will visit the elements in order:

   i = 0, coord = [0 1]
   i = 1, coord = [1 0]
   i = 2, coord = [1 1]
   i = 3, coord = [0 0]  <------------ should be first
   

2. When iterating over a vector or a scalar, the indices are off by one. Again same tensor, but as a vector (shape (4,)):

   i = 0, coord = [1]
   i = 1, coord = [2]
   i = 2, coord = [3]
   i = 3, coord = [4]  <----------- gives index error when used
   

   The same thing happens with 'scalar-like' tensors, like: [1] with shape (1, 1, 1):

   i = 0, coord = [1 0 0]  <--------- also index error, should be [0 0 0] always for scalars
   

   Interestingly, when the tensor is a vector/scalar View of a tensor that is not a vector/scalar (i.e. obtained by slicing), the issue does not happen.

What I found to work properly is to use the following for loop instead, but I don't think this is the intended way of iterating.

it = t.Iterator()
for it.Reset(); !it.Done(); it.Next() {
	...
}