mean broken on MV array

[dnadeau4]

From @doutriaux1 on August 15, 2016 15:32

In [1]: import MV2

In [2]: a=MV2.ones((13,14))

In [3]: a.mean()
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-4-dc887f7a7bfb> in <module>()
----> 1 a.mean()

/export/doutriaux1/anaconda2/envs/blah/lib/python2.7/site-packages/numpy/ma/core.pyc in mean(self, axis, dtype, out)
   4779         """
   4780         if self._mask is nomask:
-> 4781             result = super(MaskedArray, self).mean(axis=axis, dtype=dtype)
   4782         else:
   4783             dsum = self.sum(axis=axis, dtype=dtype)

/export/doutriaux1/anaconda2/envs/blah/lib/python2.7/site-packages/numpy/core/_methods.pyc in _mean(a, axis, dtype, out, keepdims)
     67     if isinstance(ret, mu.ndarray):
     68         ret = um.true_divide(
---> 69                 ret, rcount, out=ret, casting='unsafe', subok=False)
     70     elif hasattr(ret, 'dtype'):
     71         ret = ret.dtype.type(ret / rcount)

/export/doutriaux1/anaconda2/envs/blah/lib/python2.7/site-packages/numpy/ma/core.pyc in __array_wrap__(self, obj, context)
   2897                 return masked
   2898             else:
-> 2899                 result._mask = m
   2900                 result._sharedmask = False
   2901         #....

AttributeError: 'numpy.ndarray' object has no attribute '_mask'

Copied from original issue: CDAT/cdat#2088

[dnadeau4]

@doutriaux1 What is your cdms2.MV2.numpy.version.version. I have 1.11.1 and this test works fine.

[dnadeau4]

From @doutriaux1 on August 15, 2016 17:53

@dnadeau4 that was my suspicion, I'm using 1.9.2, let's try to merge numpy 1.11 in as fast as possible.

[dnadeau4]

From @doutriaux1 on August 15, 2016 17:53

oh and let's add a test for this as well.

[dnadeau4]

a.mean() was broken again for numpy version 1.12.0

Need to pass a () for to numpy.mean getitem and not [] in tvariable.expertSlice