xinrong-meng commented on PR #39585:
URL: https://github.com/apache/spark/pull/39585#issuecomment-1386654628

   > Can we make the python specific stuff a binary blob instead of an actual 
message? That way you have more flexibility in the language specific bits, for 
scala for example we are planning to use a serialised JVM object instead of a 
proto message.
   
   My understanding is the one level higher abstraction 
`ScalarInlineUserDefinedFunction` may provide enough flexibility since 
language-specific metadata can be passed in the `PythonUDF` proto. And future 
updates of `PythonUDF` proto wouldn't interfere with protos of UDFS of other 
languages.
   My concern is serializing `PythonUDF` into a binary blob may hurt the 
debuggability of proto messages and introduce additional complexity to code.
   
   Please let me know if there are other benefits of serializing `PythonUDF` 
that I miss.
   


-- 
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]

Reply via email to