Implementation of two Sliding Windows over a multivariate sequence of data in python

I am trying to construct two sliding windows over a multivariate sequence of data (m*n). The first window should be fixed and the second one is rolling over the data samples. Both windows have the same size. I followed this postdistance calculations, FFT, Sliding window and built upon that. I am struggling to implement it. The python snippet of the code is below:

sliding_dist(data,window_size):
  window_size = 10
  n = len(data)
  dist = np.zeros(n)
  for i in range(n - window_size):
    fixed_window = data[i:window_size, :]
    rolling_window = data[i:i + window_size,:]
    ditance = np.linalg.norm(fixed_window - rolling_window)
    dist[i] = ditance
  return dist

My problem is how to fix the first window and keep the second on rolling over data sample n. Logically, I am not quite sure whether the implementation of for loop and indexes are correct or should I change them. Also, I have issue with the indexes I got an error about the shape: ValueError: operands could not be broadcast together with shapes (9,3) (10,3)

Edit:

I just edited the code to take the fixed window out of the for loop. It solved the previous error ValueError: operands could not be broadcast together with shapes (9,3) (10,3) :

sliding_dist(data,window_size):
  window_size = 10
  n = len(data)
  dist = np.zeros(n)
  fixed_window = data[:window_size, :]
  for i in range(n - window_size):
    rolling_window = data[i:i + window_size,:]
    ditance = np.linalg.norm(fixed_window - rolling_window)
    dist[i] = ditance
  return dist

I am still not sure about whether the way I implemented the two sliding windows is correct or not? Please could anyone have some thoughts and help on this.