[] (__getitem__) boolean indexing assignment bug with nans

[skellys]

See repro below:

import pandas as pd
import numpy as np

temp = pd.Series(np.random.randn(10))
temp[3:6] = np.nan
temp[8] = np.nan
nan_index = np.isnan(temp)

# this works
temp1 = temp.copy()
temp1[nan_index] = [99, 99, 99, 99]
temp1[nan_index]

3    99
4    99
5    99
8    99
dtype: float64

# this doesn't - values look like they're being assigned in a different order?
temp2 = temp.copy()
temp2[nan_index] = [99, 99, 99, np.nan]

3   NaN
4    99
5    99
8    99
dtype: float64

# ... but it works properly when using .loc
temp2 = temp.copy()
temp2.loc[nan_index] = [99, 99, 99, np.nan]

3    99
4    99
5    99
8   NaN
dtype: float64

output of show_versions():

INSTALLED VERSIONS
------------------
commit: None
python: 2.7.9.final.0
python-bits: 64
OS: Windows
OS-release: 7
machine: AMD64
processor: Intel64 Family 6 Model 60 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None

pandas: 0.16.0
nose: 1.3.4
Cython: 0.21.2
numpy: 1.9.2
scipy: 0.14.0
statsmodels: 0.5.0
IPython: 3.0.0
sphinx: 1.2.3
patsy: 0.2.1
dateutil: 2.4.1
pytz: 2015.2
bottleneck: 0.8.0
tables: 3.1.1
numexpr: 2.3.1
matplotlib: 1.4.0
openpyxl: 2.0.2
xlrd: 0.9.3
xlwt: 0.7.5
xlsxwriter: 0.6.6
lxml: 3.4.2
bs4: 4.3.2
html5lib: 0.999
httplib2: 0.8
apiclient: None
sqlalchemy: 0.9.8
pymysql: None
psycopg2: None