On the weight adjacency matrix of convolutional codes by