JingsongLi commented on PR #9397:
URL: https://github.com/apache/paimon/pull/9397#issuecomment-5449968636

   **Suggestion: keep the side-effect draining change focused on 
FormatTableCommit**
   
   The shared `ThreadPoolUtils` rewrite is substantially larger than the Format 
Table feature itself (about 218 production lines plus 588 test lines), and it 
also routes the existing public `sequentialBatchedExecuteCloseable` path used 
by manifest rewrite through the new state machine, TCCL/ACC propagation, 
failure-stop, and interrupt-normalization behavior. The only new production 
consumer that needs “cancel unstarted work but never interrupt started side 
effects, then drain before rollback” is `FormatTableCommit`.
   
   Please prefer keeping the bounded submission and failure/drain logic local 
to `FormatTableCommit`, leaving the existing manifest iterator implementation 
untouched. If reuse is important, another reasonable option is a minimal 
explicit wait/drain capability on `CloseableBatchIterator` (or a separate 
implementation for the new close policy) without changing the existing 
scheduling and cancellation path. Normal iteration can still wait on the 
success path; the extra API only needs to guarantee quiescence from the 
failure/finally path. This would keep the PR scope and regression surface 
aligned with the feature.


-- 
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