zhengruifeng commented on code in PR #53059:
URL: https://github.com/apache/spark/pull/53059#discussion_r2726808232
##########
python/pyspark/sql/connect/dataframe.py:
##########
@@ -169,6 +169,12 @@ def __reduce__(self) -> Tuple:
},
)
+ def __getstate__(self) -> Any:
+ return self.__dict__.copy()
+
+ def __setstate__(self, state: Any) -> None:
Review Comment:
the two methods are needed, otherwise the SERDES of connect dataframe will
fail after the change in `__getattr__`.
We already have `__reduce__` in this class, I am not very sure how it works
with `__getstate__` and `__setstate__`
The official doc said
> Although powerful, implementing
[__reduce__()](https://docs.python.org/3/library/pickle.html#object.__reduce__)
directly in your classes is error prone. For this reason, class designers
should use the high-level interface (i.e.,
[__getnewargs_ex__()](https://docs.python.org/3/library/pickle.html#object.__getnewargs_ex__),
[__getstate__()](https://docs.python.org/3/library/pickle.html#object.__getstate__)
and
[__setstate__()](https://docs.python.org/3/library/pickle.html#object.__setstate__))
whenever possible.
I did have a try to remove the `__reduce__`, but cannot make it works.
cc @gaogaotiantian could you please help review this PR? thanks
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
