How to convert a Pandas DataFrame to Julia one without copy?

[ShuhuaGao]

In Python 3.9:

import pandas as pd
import numpy as np

from juliacall import Main as jl

pd_df = pd.DataFrame({'Name': ['Tom', 'nick', 'krish', 'jack'],
                      'Age': [20, 21, 19, 18]})
jl.seval("using DataFrames")
jl_df = jl.DataFrame(pd_df)
jl_df[1, 2] = -100
pd_df.iloc[0, 1]  # still 20

Since PythonCall.PyPandasDataFrame is compatible with Table.jl and Julia DataFrame's constructor does not copy a table by default, how to explain the above behaviour? Is there any way to avoid data copying?

[cjdoris]

Indeed accessing columns of a PyPandasDataFrame is currently a copying operation. That whole functionality isn't very well developed so far - but contributions are welcome.

[cjdoris]

The latest release now uses no-copy views of columns where possible, so your above example mutates pd_df as desired.