LsomeYeah opened a new pull request, #11043:
URL: https://github.com/apache/paimon/pull/11043

   ### Purpose
   
   Allow users of `merge_map_with_keytime` to opt into replacing a map entry 
when the incoming ROW has the same timestamp as the accumulated ROW.
   
   - Add the field-level boolean option 
`fields.<field-name>.overwrite-on-equal-ts`, defaulting to `false`. Existing 
tables keep the strict-greater-than behavior; enabling it also replaces the 
entire ROW on equal timestamps.
   - Apply the option through the Java aggregator factory and the PyPaimon 
Python implementation. Timestamp ordering, null handling, and key deletion 
behavior remain unchanged.
   - Document that equal-timestamp overwrites follow merge order, not 
wall-clock arrival order. Writers, readers, and compaction jobs must use 
consistent settings; changing the option cannot recover values discarded by 
previous merges.
   - Keep table-level coverage in `PrimaryKeySimpleTableTest`. Allow this class 
up to 5,000 lines while retaining the 4,000-line limit for other files.
   
   ### Tests
   
   - Java: 187 targeted tests passed across `FieldAggregatorTest`, 
`PartialUpdateMergeFunctionTest`, `AggregateMergeFunctionTest`, and the new 
parameterized method in `PrimaryKeySimpleTableTest`.
   - Unit coverage includes omitted/false/true settings, equal/newer/older 
timestamps, reversed aggregation, null timestamps, null maps, and key deletion.
   - Table-level coverage checks aggregation and partial-update engines with 
both settings: in-buffer aggregation, merging across commits/files, and results 
before and after compaction.
   - Python: all 128 tests in `test_field_aggregators.py` passed.
   - Spotless, Checkstyle, Python flake8, and `git diff --check` passed.
   


-- 
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