Why does `.rename(columns={'b': 'b'}, copy=False)` followed by inplace method not update the original dataframe?

Question

Here's my example:

In [1]: import pandas as pd

In [2]: df = pd.DataFrame({'a': [1,2,3], 'b': [4,5,6]})

In [3]: df1 = df.rename(columns={'b': 'b'}, copy=False)

In [4]: df1.isetitem(1, [7,8,9])

In [5]: df
Out[5]:
   a  b
0  1  4
1  2  5
2  3  6

In [6]: df1
Out[6]:
   a  b
0  1  7
1  2  8
2  3  9

If df1 was derived from df with copy=False, then I'd have expected an in-place modification of df1 to also affect df. But it doesn't. Why?

I'm using pandas version 2.2.1, with no options (e.g. copy-on-write) enabled

Answer 1

copy=False means that the underlying data (numpy) is shared.

If we modify one item of the underlying numpy array, it is reflected since the data is shared:

df = pd.DataFrame({'a': [1,2,3], 'b': [4,5,6]})
df1 = df.rename(columns={'b': 'c'}, copy=False)
df1.values[0, 1] = 999

print(df)
   a    b
0  1  999
1  2    5
2  3    6

print(df1)
   a    c
0  1  999
1  2    5
2  3    6

print(id(df.values), id(df1.values))
# 140335530654768 140335530654768

The DataFrame objects (containers, indices, column names…) are however not identical.

print(id(df), id(df1))
# 140335696175520 140335696171536

By setting a new column with df1.isetitem(1, [7,8,9]) you just add a new Series and force a copy.