yihua opened a new pull request, #20252: URL: https://github.com/apache/hudi/pull/20252
### Describe the issue this Pull Request addresses closes #20251 Row-writer bulk insert always writes the parquet footer bloom filter, ignoring `hoodie.parquet.bloom.filter.enabled=false`. The config and the shared `HoodieFileWriterFactory.enableBloomFilter` gate came in with #10457 and were wired into the Avro, Spark and Flink writer factories, but `HoodieInternalRowFileWriterFactory` (from #5523) kept its own check and was missed, and again when #19205 moved the gate to `MetaFieldsMode`. The bloom adds about 430 KB to every base file footer, which every reader loads. ### Summary and Changelog `HoodieInternalRowFileWriterFactory` now uses the shared `enableBloomFilter` and `createBloomFilter`, so the row writer writes the bloom filter and min/max record key footers under the same condition as the other writers: meta fields fully populated and either the config is on or the index is BLOOM / GLOBAL_BLOOM. `TestBulkInsertBloomFilterFooter` covers bulk insert with the row writer on and off; the row-writer case with the config off fails without the fix. ### Impact Only writers that set `hoodie.parquet.bloom.filter.enabled=false` see a change: their bulk-inserted (and row-writer clustered) files now omit the bloom, as upserts and inserts already do. Defaults are unchanged and BLOOM / GLOBAL_BLOOM still force it. ### Risk Level low ### Documentation Update none ### Contributor's checklist - [ ] Read through [contributor's guide](https://hudi.apache.org/contribute/how-to-contribute) - [ ] Enough context is provided in the sections above - [ ] Adequate tests were added if applicable -- 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]
