rdblue commented on PR #4512:
URL: https://github.com/apache/iceberg/pull/4512#issuecomment-1094375187

   I don't think that this is the right approach.
   
   In most cases, the residual evaluator is small because the initial 
expression is small. Plus, the residual evaluator is reused for all of the 
tasks, so there is only one in memory. Eagerly evaluating the residual could 
create a huge number of expressions in memory, one for each partition at a 
minimum and probably on per file split.
   
   I also don't think this would help very much because if the large expression 
with thousands of literal values is not eliminated by the residual evaluator, 
it would be retained in the resulting expressions.
   
   Overall, I think that the right balance is to keep the residual evaluator 
here. There will be some cases where that is expensive, but much better in the 
average case. The cases where it is expensive are cases where the large 
expression is already in memory anyway.


-- 
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