Ordering of python set from NumPy array

Question

I am having trouble figuring out why I make a set from a NumPy array, Python swaps the order of elements:

import numpy as np
A = np.array([2])
B = np.array([2, 8])
setA = set(A)
setB = set(B)

In [6]: A
Out[6]: [2]

In [7]: B
Out[7]: [2, 8]

In [8]: setA
Out[8]: set([2])

In [9]: setB
Out[9]: set([8, 2])

In [10]: list(setA.union(setB))
Out[10]: [8, 2]

In [11]: np.union1d(A,B).tolist()
Out[11]: [2, 8]

Why isn't the order wouldn't be maintained when I created set(B) ?

Answer 1

set s by definition have no order - they are instead created so as to optimize certain operations such as those testing for containment. Therefore, you should never rely on order preservation when you create / add elements to a set.

Answer 2

Sets are unordered collections of unique elements , so set([2,8]) and set([8, 2]) are exactly the same. Why do you care? Maybe a set is not what you need...