np.random.shuffle corrupts astropy.table.Table

[lpsinger]

Calling np.random.permutation() on an instance of Table correctly returns a randomly reordered version of the table, but calling np.random.shuffle() to randomly reorder the table in place eventually corrupts the data. I am using Astropy 1.3.3 and Numpy 1.13.0 with Python 2.7.13 from MacPorts on macOS Sierra (10.12.5). See example below.

>>> from astropy.table import Table
>>> import numpy as np
>>> np.random.seed(0)
>>> t = Table(rows=[[1, 2, 3], [4, 5, 6]])
>>> t
<Table length=2>
 col0  col1  col2
int64 int64 int64
----- ----- -----
    1     2     3
    4     5     6
>>> np.random.seed(0)
>>> np.random.permutation(t)
array([(4, 5, 6), (1, 2, 3)],
      dtype=[('col0', '<i8'), ('col1', '<i8'), ('col2', '<i8')])
>>> np.random.shuffle(t); t
<Table length=2>
 col0  col1  col2
int64 int64 int64
----- ----- -----
    1     2     3
    4     5     6
>>> np.random.shuffle(t); t
<Table length=2>
 col0  col1  col2
int64 int64 int64
----- ----- -----
    1     2     3
    4     5     6
>>> np.random.shuffle(t); t
<Table length=2>
 col0  col1  col2
int64 int64 int64
----- ----- -----
    1     2     3
    1     2     3