What is the current behavior?
Currently Arrow backed dataframes (polars etc.) must be converted to pandas before write_pandas to write local dataframes into Snowflake
What is the desired behavior?
Support an Arrow native write API similar to write_pandas
How would this improve snowflake-connector-python?
Streamlines non-pandas based workflows (ex: polars)
References and other background
Implemented in snowpark-python @ snowflakedb/snowpark-python#2959