`.diff(axis=1)` gives `NaNs` with different types.

[eoincondron]

df = pd.DataFrame({'a': np.arange(3, dtype='float32'), 'b': np.arange(3, dtype='float64')})
df.diff(axis=1)

	a	b
0	NaN	NaN
1	NaN	NaN
2	NaN	NaN

df = pd.DataFrame({'a': np.arange(3, dtype='int32'), 'b': np.arange(3, dtype='int64')})
df.diff(axis=1)
	a	b
0	NaN	NaN
1	NaN	NaN
2	NaN	NaN

Problem description

When diffing across the column axis with different numeric types in the columns we get
Not sure if this is intentional behaviour but given that df.a - df.b works as expected, I would think that .diff does the same. I think it should at least emit a warning when the types differ.

Expected Output

	a	b
0	NaN	0
1	NaN	0
2	NaN	0

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.2.final.0
python-bits: 64
OS: Linux
OS-release: 3.10.0-327.el7.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.UTF-8
LOCALE: en_GB.UTF-8

pandas: 0.20.3
pytest: 3.2.1
pip: 9.0.1
setuptools: 36.4.0
Cython: None
numpy: 1.13.1
scipy: 0.19.1
xarray: None
IPython: 6.1.0
sphinx: None
patsy: 0.4.1
dateutil: 2.6.1
pytz: 2017.2
blosc: None
bottleneck: None
tables: 3.4.2
numexpr: 2.6.2
feather: None
matplotlib: 2.0.2
openpyxl: 2.4.8
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: 0.9999999
sqlalchemy: 1.1.13
pymysql: 0.7.9.None
psycopg2: None
jinja2: 2.9.6
s3fs: None
pandas_gbq: None
pandas_datareader: None

[WillAyd]

Fair callout. I suppose the below works so should probably work for axis=1 as well:

In [9]: df.T.diff()
Out[9]: 
     0    1    2
a  NaN  NaN  NaN
b  0.0  0.0  0.0

Investigation and PRs welcome!

[aschade92]

Taking a look!

[jbrockmendel]

closed by #32995