Consider a date-indexed DataFrame:

d0 = datetime.date(2024,5,5)
d1 = datetime.date(2024,5,10)
df0 = pd.DataFrame({"a":[1,2],"b":[10,None],"c":list("xy")}, index=[d0,d1])
df0.index
Index([2024-05-05, 2024-05-10], dtype='object')

Note that df0.index.dtype is object.
Now, lookup works for date:

df0.loc[d0]
a       1
b    10.0
c       x
Name: 2024-05-05, dtype: object

but both df0.loc[str(d0)] and df0.loc[pd.Timestamp(d0)] raise KeyError.
This seems to be reasonable.

However, consider

df1 = df0.reindex(pd.date_range(d0,d1))
df1.index
DatetimeIndex(['2024-05-05', '2024-05-06', '2024-05-07', '2024-05-08',
               '2024-05-09', '2024-05-10'],
              dtype='datetime64[ns]', freq='D')

Note that df1.index.dtype is datetime64.
Now, lookup works for both df1.loc[pd.Timestamp(d0)] (as expected) and df1.loc[str(d0)] (why?!) but not for df1.loc[d0] (if it works for a string, why not date?!)

Is this the expected behavior (a bug with tenure)? Is this intentional?