englefly commented on PR #69784:
URL: https://github.com/apache/doris/pull/69784#issuecomment-6475215834
Status update after local verification (the approach is being revised,
please hold the review):
* Repro confirmed: with the previous approach reverted, both reported
queries trip
`AdjustNullable convert slot ... from not-nullable to nullable`; the
second one is reduced to the shape which
still reproduces (kept as a regression test).
* Unit tests pass (`ConstantPropagationTest`,
`ConstantPropagationFreshExprIdTest`), and on a local cluster the
two queries pass with `fe_debug=true`.
* Regression: the mv suites `agg_optimize_when_uniform` (`mv6_0`),
`aggregate_without_roll_up` (`mv17_0`) and
`pre_write_strategy` (`mv0_1`) now report `... not chose`: renaming the
exprId of a constant grouping key
changes the identity of the column the materialized view matching relies
on.
I'm reworking the fix so that it does not rename the exprIds: keep the plan
consistent instead (do not create the
second definition of the exprId) and keep the mv matching intact.
--
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]
