Gabriel39 opened a new pull request, #66422:
URL: https://github.com/apache/doris/pull/66422
### What problem does this PR solve?
Issue Number: None
Related PR: None
Problem Summary: Doris equality treats all NaN payloads as equal, while
Parquet Bloom filters hash their physical bits. A negative probe with a
different NaN payload could therefore skip a row group that contains matching
rows. This change keeps NaN Bloom probes conservative for FLOAT and DOUBLE
while preserving pruning for finite values.
### Release note
Fix Parquet Bloom filter pruning from incorrectly skipping row groups for
NaN equality and IN predicates.
### Check List (For Author)
- Test: Unit Test
- `./run-be-ut.sh --run --filter="ExprZonemapFilterTest.*" -j 48`
- Behavior changed: Yes. NaN Bloom probes now retain the row group instead
of trusting a raw-byte miss.
- Does this need documentation: No
--
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]
