Gabriel39 opened a new pull request, #67573:
URL: https://github.com/apache/doris/pull/67573
### What problem does this PR solve?
Issue Number: None
Related PR: None
Legacy parquet-mr and parquet-cpp writers can under-report Parquet null
counts. Format v2 treated a reported zero as proof that a row group or page had
no NULL values, so `IS NULL` pruning could drop matching rows.
### What is changed?
- Derive null-count trust from Parquet `created_by` in format v2.
- Conservatively retain NULL candidates for affected legacy writer versions
at row-group and page-index pruning layers.
- Preserve min/max pruning and existing behavior for fixed or unknown
writers.
### Release note
Avoid incorrect `IS NULL` pruning for Parquet files produced by affected
legacy writers.
### Check List (For Author)
- Test: Unit Test
- `NativeParquetStatisticsTest.*` (19 tests passed with ASAN)
- Behavior changed: Yes. Format v2 no longer trusts null counts from
affected legacy Parquet writers for metadata pruning.
- 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]
