[Numpy-discussion] is ndarray.base the closest base or the ultimate base?

[Citi, Luca]

Hello,
I cannot quite understand whether ndarray.base is the closest base,
the one from which the view was made or the ultimate base, the one
actually containing the data.
I think the documentation and the actual behaviour mismatch.

In [1]: import numpy as np
In [2]: x = np.arange(12)
In [3]: y = x[::2]
In [4]: z = y[2:]
In [5]: x.flags
Out[5]:
  C_CONTIGUOUS : True
  F_CONTIGUOUS : True
  OWNDATA : True
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False
In [6]: y.flags
Out[6]:
  C_CONTIGUOUS : False
  F_CONTIGUOUS : False
  OWNDATA : False
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False
In [7]: z.flags
Out[7]:
  C_CONTIGUOUS : False
  F_CONTIGUOUS : False
  OWNDATA : False
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False
In [8]: z.base
Out[8]: array([ 0,  2,  4,  6,  8, 10])

It looks like the "base" of "z" is "y", i.e. its closest base,
the array from which the view "z" was created.

But the documentation says:

    base : ndarray
        If the array is a view on another array, that array is
        its `base` (unless that array is also a view).  The `base` array
        is where the array data is ultimately stored.

and it looks like the "base" should be "x", the array
where the data is ultimately stored.

I like the second one better. First, because this way I do not have
to travel all the bases until I find an array with OWNDATA set.
Second, because the current implementation keeps "y" alive
because of "z" while in the end "z" only needs "x".

In [11]: del y
In [12]: z.base
Out[12]: array([ 0,  2,  4,  6,  8, 10])

Comments?
Best,
Luca