Mismatched behavior of sorting functions for DataFrame vs. Series

[dlenski]

The behavior of these methods differs in confusing ways:

DF:

• DataFrame.sort_index takes a by parameter, in which case the index is resorted by a specific column's values
• DataFrame.sort defaults to inplace=False
• DataFrame.order does not exist

Series:

• Series.sort defaults to inplace=True
• Series.order defaults to inplace=False, but is otherwise identical to Series.sort
• Series.sort_index does not take a by parameter

This makes it very hard to remember which one to use, especially after an aggregating function. For example:

df.max(axis=1).order(ascending=False)      # max returns a Series in this case, use Series methods
df.max(level='Foo').sort_index(by='Bar')      # max returns a DF in this case, use DF methods

Any reason not to make the inplace behavior identical for DataFrame.sort and Series.sort? Also, Series.sort_index could take by=0 to replicate the behavior of DataFrame.

[jreback]

see #9816 and others for discussion