在 python 多处理管理器命名空间中，为什么不能直接分配

Question

Can anyone do a favor to figure out why we cant change dataframe directly? add_new_derived_column_NOT_work dont work as I expected

#-*- coding: UTF-8 -*-'
import pandas as pd
import numpy as np
from multiprocessing import *
import multiprocessing.sharedctypes as sharedctypes
import ctypes

def add_new_derived_column_work(ns):
    dataframe2 = ns.df
    dataframe2['new_column']=dataframe2['A']+dataframe2['B'] / 2
    print (dataframe2.head())
    ns.df = dataframe2

def add_new_derived_column_NOT_work(ns):
    ns.df['new_column']=ns.df['A']+ns.df['B'] / 2
    print (ns.df.head())

if __name__ == "__main__":

    mgr = Manager()
    ns = mgr.Namespace()

    dataframe = pd.DataFrame(np.random.randn(100000, 2), columns=['A', 'B'])
    ns.df = dataframe
    print (dataframe.head())

    # then I pass the "shared_df_obj" to Mulitiprocessing.Process object
    process=Process(target=add_new_derived_column_work, args=(ns,))
    process.start()
    process.join()

    print (ns.df.head())

Answer 1

As per Python documentation:

• a namespace object has "writable attributes"

• but about proxy objects , you can read that "If standard (non-proxy) list or dict objects are contained in a referent, modifications to those mutable values will not be propagated through the manager because the proxy has no way of knowing when the values contained within are modified. However, storing a value in a container proxy does propagate through the manager and so to effectively modify such an item"

This why, I think, add_new_derived_column_work works, as it assign a new dataframe to ns.df, whereas add_new_derived_column_NOT_work fails, as it tries to add a column to the dataframe by mutating it, but this mutation does not actualy affect ns.df.