jayshrivastava commented on issue #27543:
URL: https://github.com/apache/datafusion/issues/27543#issuecomment-7012301247

   > Calling datafusion::physical_plan::execution_plan::reset_plan_states 
before encoding works around it for the filter a node produces. As documented, 
it does not reset a dynamic filter that planning pushed into a child (for 
example a hash join's filter pushed into a Parquet scan), so it is not a full 
fix.
   
   I think this is the bug. The contents of the filer should be cleared imo. 
I'm not sure why it doesn't get cleared today - typically mutating the producer 
mutates the consumer because they point to the same memory. I wonder if the 
consumer is caching the expression - this should be reset.
   
   > should encoding default to the planned state (say, with an opt in to keep 
the live one) 
   This feels the same as the `reset_state` / `reset_plan_states` API. If a 
user doesn't want to reset the plan, they just skip calling `reset_state`.
   


-- 
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