gortiz commented on PR #20061:
URL: https://github.com/apache/pinot/pull/20061#issuecomment-6959062111

   I opened [apache/pinot#20293](https://github.com/apache/pinot/pull/20293) to 
replace this PR. Looking back at the review rounds, most threads came from 
changes that touched queries that already work on master, so the new PR is 
built around one rule: the broker plans the query exactly as master does, and 
plans it again with the MV aggregate only when that plan groups by 
`ARRAY_TO_MV` and fails the validation. Every plan fixture that exists on 
master is unchanged there.
   
   It follows the three suggestions from @yashmayya's last review: the fallback 
is decided in `compile()`, the filter rule only changes the second plan, and 
`MvAggregateOperator` extends `AggregateOperator` and reports `AGGREGATE` 
stats. That covers the RLS and semi-join blockers and the stats P1 from 
@xiangfu0, and the new PR also has the fixes for the three P2s (shared group 
ids across filters, empty keys before the overflow check, and `numDocsScanned` 
in the compat fixtures). The leaf-only `HAVING` fix, the null fixes for the 
first plan, and the rejection of `GROUP BY mvCol` with the partitioned hint are 
out of scope there, since each of them changes queries that work today.
   


-- 
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