XiaoHongbo-Hope opened a new pull request, #11269:
URL: https://github.com/apache/paimon/pull/11269

   ## Purpose
   
   PyPaimon's Arrow read APIs resolve an omitted `blob_parallelism` to 1 and 
explicitly forward it to the native reader, serializing BLOB range reads within 
each split even when callers did not request serial reads.
   
   Default this parameter to 16. Explicit overrides, including 1 for serial 
reads, remain supported. Keep existing backend scheduling: Core owns native 
concurrency, and the Python fallback retains its existing total BLOB worker 
cap. Update the API documentation and test default propagation through both 
Arrow APIs, explicit overrides, and fallback capping.
   
   ## Validation
   
   - `native_read_test.py` and `reader_parallel_test.py`: 119 passed.
   - BLOB concurrency and cap tests: 5 passed.
   - `git diff --check`.
   
   This is a candidate default, not a claim of a universal optimum. Small 
batches, local storage and multiple training workers may need lower explicit 
concurrency. No read-amplification or planning changes are included.
   


-- 
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