BUG/ERR: adding column with float label to Int64Index gives ValueError

[jorisvandenbossche]

When adding a column with a float label to a DataFame with integer column index, you get a ValueError: cannot insert 0.5, already exists error with an incorrect message:

In [15]: df = pd.DataFrame(np.random.randn(8,4))

In [16]: df
Out[16]:
          0         1         2         3
0 -0.482266 -1.239583 -1.177658  1.676881
1  0.314620 -0.373801 -1.535093  0.829534
2  0.887573 -0.054425  0.313028  0.590081
3 -0.665164 -0.887124  0.720014 -0.361315
4 -0.082880 -0.435376 -0.124390 -0.883354
5  2.122313  0.155019  0.762171 -0.401316
6  1.368269  0.469080 -0.274609 -1.799257
7 -0.631479  0.302715  0.008982 -0.356644

In [17]: df[0.5] = np.NaN
...
ValueError: cannot insert 0.5, already exists

In [18]: df.loc[:,0.5] = np.NaN
...
TypeError: the label [0.5] is not a proper indexer for this index type (Int64Index)

Of course, when the column labels are converted to floats (df.columns = pd.Float64Index(df.columns)), this does work.

[jreback]

the problem is [] tries to coerce it to an int (e.g. it sort of acts like .ix here.
guess this should work though (maybe)

[wesm]

This is still broken

[mroeschke]

Looks to work on master. Could use a test.

In [21]: In [17]: In [15]: df = pd.DataFrame(np.random.randn(8,4))

In [22]: In [17]: df[0.5] = np.NaN
    ...:

In [23]: df
Out[23]:
        0.0       1.0       2.0       3.0  0.5
0 -0.893317 -0.999960  0.227483 -0.616143  NaN
1 -1.557036 -0.661066  1.256377 -0.251763  NaN
2  0.874911  1.208229  0.496551  0.602761  NaN
3 -0.143278  1.483060 -0.441488 -0.036813  NaN
4 -0.423643  0.937388 -0.736469 -1.473181  NaN
5 -1.169715  0.048951  0.463316  0.474189  NaN
6 -0.064023  0.252234 -1.152730  0.603041  NaN
7  0.080979 -0.022521 -0.040599  1.601390  NaN

In [24]: In [15]: df = pd.DataFrame(np.random.randn(8,4))
    ...:

In [25]: In [18]: df.loc[:,0.5] = np.NaN
    ...:

In [26]: df
Out[26]:
        0.0       1.0       2.0       3.0  0.5
0  0.324355 -0.360758 -2.094797  1.482548  NaN
1 -1.335983  0.264566  0.436681  0.716613  NaN
2 -1.386648  1.968785  0.383127 -0.103017  NaN
3 -1.090121  1.289168  1.373147  0.181029  NaN
4  1.072814  1.546031 -0.376255  0.139408  NaN
5 -0.169053  1.594981  0.882534 -0.962147  NaN
6 -0.884327  0.781743  0.298201  0.177668  NaN
7  0.433245 -2.354488  1.608168 -0.835746  NaN

In [27]: pd.__version__
Out[27]: '0.26.0.dev0+684.g953757a3e'