pd.concat(objs) drops type of the inputs if all inputs are empty

[OlexiyO]

When concatenating several pd.Series of the same type, I would expect the output Series to have the same type. However, this does not hold if all input Series are empty:

In [0]: x = pd.Series([], dtype=int)
In [1]: pd.concat([x, x])
Series([], dtype: object)

How I would expect it to work:
pd.concat([x, x]) should return Series([], dtype=int)

Just as an example, this works fine:

In [2]: y = pd.Series([1], dtype=int)
In [3]: pd.concat([y, y])  # This works fine:
0    1
0    1
dtype: int64

[cpcloud]

Are you sure that's the correct example? The first x is not empty. In any event, what's the use case here? Why do you care about the dtype of an empty array?

[jreback]

this is fixed in master/0.14 by #5742

In [5]: x = Series([],dtype=int)

In [6]: concat([x,x])
Out[6]: Series([], dtype: int64)

In [7]: x = Series([],dtype=float)

In [8]: concat([x,x])
Out[8]: Series([], dtype: float64)

[OlexiyO]

Are you sure that's the correct example? The first x is not empty.
Ouch, sorry about this. I've updated the original example -- it should be

x = pd.Series([], dtype=int)

Why do you care about the dtype of an empty array?
Because some further code may depend on the array having a certain type.

this is fixed in master/0.14
Thanks, jreback!