read_csv does not set index name on empty dataframe

[corr723]

If foo.csv contains

x,y
1,2

then you get the normal behavior:

>>> df = pd.read_csv('foo.csv', index_col=0)
>>> list(df.columns)
['y']
>>> df.index.name
'x'

but if you remove the data row "1,2" in the CSV the index name gets dropped:

>>> df = pd.read_csv('foo.csv', index_col=0)
>>> list(df.columns)
['y']
>>> df.index.name
None

This means round-tripping with to_csv/read_csv fails for empty dataframes.

[jreback]

ok, will mark as a bug. want to do a pull-request to fix?

[corr723]

No, I don't have a patch.

[jreback]

@corr723 well if you would like to work on one would be gr8

[corr723]

No plan to work on one, sorry

[jorisvandenbossche]

Closed by #10302