Masked array set slice fails with np.ma.masked on right side

[taldcroft]

The following behavior is not what I would expect and seems like a bug:

In [1]: x = np.ma.MaskedArray([1,2,3])
In [2]: x[:2] = [100, np.ma.masked]
In [3]: x
Out[3]: 
masked_array(data = [100   0   3],
             mask = False,
       fill_value = 999999)

One can correctly set a single item (e.g. x[2] = np.ma.masked) or set a slice to a scalar (x[:2] = np.ma.masked).

I'm using numpy 1.11.3.

[taldcroft]

Interestingly, this works as expected:

In [1]: x = np.ma.MaskedArray([1,2,3])
In [2]: x[:2] = np.ma.asarray([100, np.ma.masked], dtype=int)
In [3]: x
Out[3]: 
masked_array(data = [100 -- 3],
             mask = [False  True False],
       fill_value = 999999)

[eric-wieser]

A related bug came up in #8510, which I'll repeat here since that wasn't the focus of that issue:

>>> arr = np.ma.arange(5)
>>> arr[1:3] = [1, np.ma.masked]
ValueError: setting an array element with a sequence.

#expected result:
>>> arr
masked_array(data = [0 1 -- 3 4],
             mask = [False False  True False False],
       fill_value = 999999)

[eric-wieser]

Your code fails in the same way as the above for me on both master and 1.11.1

[sethtroisi]

This no longer reproduces and this problem is "fixed"

>>> x = np.ma.MaskedArray([1,2,3])
>>> x[:2] = [100, np.ma.masked]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 3323, in __setitem__
    _data[indx] = dval
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 4312, in __int__
    raise MaskError('Cannot convert masked element to a Python int.')
numpy.ma.core.MaskError: Cannot convert masked element to a Python int.

>>> x = np.ma.MaskedArray([1,2,3])
>>> x[:2] = np.ma.asarray([100, np.ma.masked], dtype=int)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 7771, in asarray
    subok=False, order=order)
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 2788, in __new__
    order=order, subok=True, ndmin=ndmin)
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 4312, in __int__
    raise MaskError('Cannot convert masked element to a Python int.')
numpy.ma.core.MaskError: Cannot convert masked element to a Python int.

>>> arr = np.ma.arange(5)
>>> arr[1:3] = [1, np.ma.masked]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 3323, in __setitem__
    _data[indx] = dval
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 4312, in __int__
    raise MaskError('Cannot convert masked element to a Python int.')
numpy.ma.core.MaskError: Cannot convert masked element to a Python int.