JunRuiLee opened a new pull request, #686:
URL: https://github.com/apache/paimon-rust/pull/686

   ### Purpose
   
   Linked issue: N/A (draft performance investigation)
   
   Align async Parquet metadata loading with parquet-hadoop's seek-based path. 
Instead of fetching a fixed 512 KiB suffix, read the 8-byte footer first and 
then request the exact metadata and page-index ranges.
   
   OpenDAL's HDFS native service already keeps a file-scoped positioned-read 
handle, so these exact range reads reuse the same underlying HDFS file reader. 
Local HDFS benchmark validation showed better performance; detailed benchmark 
numbers can be added before marking the PR ready for review.
   
   ### Brief change log
   
   - Remove the fixed 512 KiB Parquet metadata prefetch hint.
   - Avoid loading OffsetIndex for an empty external row selection.
   - Add range-tracking tests that assert the exact footer and metadata reads.
   - Verify an empty row selection returns zero rows without page-index reads.
   
   ### Tests
   
   - `cargo fmt --all -- --check`
   - `cargo test --locked --offline -p paimon --lib 
arrow::format::parquet::tests` (47 passed)
   - `cargo test --locked --offline -p paimon --lib --features storage-hdfs` 
(2177 passed, 1 ignored)
   
   ### API and Format
   
   No API or storage-format changes.
   
   ### Documentation
   
   No documentation changes required.


-- 
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]

Reply via email to