complete rework

scalar

• dios.loc[scalar,scalar] -> scalar
• dios.iloc[int,int] -> scalar

Series

• dios['a'] -> pd.Series
• dios[slice] -> pd.Series
• dios.loc[x,scalar] -> pd.Series
• dios.iloc[x,scalar] -> pd.Series with x a slice, or iterabel, bool-list, bool-series

column-label Series

• dios.loc[scalar] -> pd.Series
• dios.iloc[int] -> pd.Series
• dios.loc[scalar, x] -> pd.Series
• dios.iloc[scalar, x] -> pd.Series with x a slice, or iterabel, bool-list, bool-series

dios

• dios[iterable] -> dios
• dios[bool-series] -> dios (slices columns)
• dios[boolx] -> dios (with x dios/df/nested-list)
• dios.iloc[x,y] -> pd.Series
• dios.iloc[x,y] -> pd.Series with x/y a slice, or iterabel, bool-list, bool-series

Haven't looked into the code yet, but the following points are a bit unclear to me:

• Series:
  • I think dios[slice] -> pd.Series should be dios[slice] -> dios instead. If you use a slice on a pd.DataFrame pandas returns a pd.DataFrame.
  • dios.iloc[x,scalar] -> pd.Series should be dios.iloc[x, int] -> pd.Series as I think, iloc should only support int-scalars

Apart from that, I think the choices are reasonable.

yes you're right in all points, and it is implemented that way.

1. dios[slice] indeed return a dios
2. dios.iloc[x,int] is's just a typo

closed

please do not merge, i forgot something serious