yihua opened a new issue, #932: URL: https://github.com/apache/hudi-rs/issues/932
### Feature Description Every base file the native writer produces carries a record-key bloom filter and key range in its parquet footer, because `hoodie.parquet.bloom.filter.enabled` defaults to true as in Java. At Java's default sizing (false positive rate 1e-9) each key goes through 30 chained MurmurHash passes, which makes the filter one of the largest costs of a bulk load. hudi-rs locates keys with the record index and reads none of these footers, so a table pays for the filter without using it. Proposed: write the footer only when `hoodie.parquet.bloom.filter.enabled` is set to true, or when `hoodie.index.type` is a bloom index (a Java writer on that table needs it), and keep Java's sizing when it is written. -- 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]
