concat Series[sparse] converts sp_values to NaN

[TomAugspurger]

In [1]: import pandas as pd

In [2]: s = pd.Series(pd.SparseArray([0, 1], fill_value=0))

In [3]: pd.concat([s, s], axis=1, keys=['a', 'b'])
Out[3]:
     a    b
0  NaN  NaN
1  1.0  1.0

[TomAugspurger]

Note, the only happens when we choose to return a SparseDataFrame instead of a DataFrame of sparse values. I thought we changed that to return a DataFrame..

[TomAugspurger]

This is the behavior on 0.23.4

In [1]: import pandas as pd

In [2]: s = pd.Series(pd.SparseArray([0, 1], fill_value=0))

In [3]: pd.concat([s, s], axis=1, keys=['a', 'b'])
Out[3]:
   a  b
0  1  1
1  1  1

[TomAugspurger]

Huh, those are dense values too:

In [13]: pd.concat([s, s], axis=1, keys=['a', 'b']).a.values
Out[13]: array([1, 1])