Buggy creation of int DataFrame from data containing NAs

[wesm]

In [5]: df = DataFrame([[np.nan, 1], [1, 0]], dtype=np.int64)

In [6]: df
Out[6]: 
                     0  1
0 -9223372036854775808  1
1                    1  0

[lbeltrame]

Changing dtype at runtime still causes the wrong cast, however:

In [7]: test = pandas.DataFrame(data={"Values":[1.0, 2.0, 3.0, np.nan]})

In [8]: test["Values"].astype("int64")
Out[8]: 
0                      1
1                      2
2                      3
3   -9223372036854775808
Name: Values

[wesm]

did some more work here, casting to integer with NAs raises a ValueError now