sarutak commented on issue #25260: [SPARK-28520][SQL] WholeStageCodegen does 
not work property for LocalTableScanExec
URL: https://github.com/apache/spark/pull/25260#issuecomment-516636379
 
 
   Thanks @reactormonk for the comment!
   
   >  There are a few things I'm wondering about, though. Is LocalTableScanExec 
the only case where we're doing this kind of "if root then don't codegen" 
cutoff? If there are other cases, it might make more sense to extract a trait 
and make the insertWholeStageCodegen logic cleaner and more flexible.
   
   Yes for now. First I considered adding new trait but I found the case we 
need to cut off was only `LocaTableScanExec` so I decided to change 
`insertWholeStageCodegen` like a `SMJ` case in `insertInputAdapter`.  But it's  
worth re-considering to add new trait. I'll try it.
   
   > Also, what about the case when the immediate parent of a 
LocalTableScanExec is a Spark stage boundary, e.g. an Exchange? That case seems 
to be covered by this PR already (which doesn't insert a WholeStageCodegenExec 
right above LocalTableScanExec. Is that correct?
   
   Yes, this PR cover the case you mentioned. Actually, I added a test case for 
test that case but I removed in response to the review comment. Do you think 
it's still good to add a test case for the case?
   
   

----------------------------------------------------------------
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.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to