danny0405 opened a new issue, #20408:
URL: https://github.com/apache/hudi/issues/20408

   ### Task Description
   
   Express a keyed streaming flow with primaryKey, preCombineField, partition 
fields, and existing Hudi record-merger settings. Use the existing upsert 
writer with a boolean _hoodie_is_deleted input marker; do not introduce a new 
merge engine or claim Spark Auto CDC/SCD1/2 support.
   
   Add a deterministic functional fixture containing duplicate keys, increasing 
and out-of-order sequence values, updates across batches, delete markers, and a 
later reinsertion. Select and document the merger/ordering mode so expected 
results are unambiguous. Verify both a newly created target and an existing 
table whose persisted settings are loaded.
   
   This tests delete input carried by the flow; it does not assume a 
latest-state upstream Hudi incremental read propagates hard deletes.
   
   ### Acceptance criteria
   
   - [ ] Intra-batch duplicate keys collapse according to the configured 
precombine/merger contract.
   - [ ] Later updates replace prior state and lower-order updates do not 
overwrite newer state under the selected ordering mode.
   - [ ] Valid delete markers remove the intended keys; a later valid 
reinsertion yields the expected record.
   - [ ] Assert exact row contents after each batch and after restart/retry, 
not only row counts.
   - [ ] Negative tests cover missing key/partition/precombine fields, 
non-boolean delete markers, and attempts to change persisted identity.
   - [ ] CLI/Connect functional coverage and a runnable SQL/Python example 
document the behavior.
   
   ### Task Type
   
   Test enhancement
   
   ### Related Issues
   
   Parent feature issue: #20401
   RFC: #19997 (under review; implementation must follow the agreed contract).
   Dependencies for end-to-end acceptance: #20406, #20403.
   


-- 
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