vecs.i.(j,k) = images.i.j.k This converts a 3D array to a 2D array, with the indices of the 2nd dimension being tuples. This syntax is really nice. If the arr

vecs.i.(j,k) = images.i.j.k This converts a 3D array to a 2D array, with the indices of the 2nd dimension being tuples. This syntax is really nice. If the array shapes are types though, then what happens when you have to read a matrix from file and you don't know the shape of the matrix?