JingsongLi opened a new pull request, #11324:
URL: https://github.com/apache/paimon/pull/11324

   ### Purpose
   
   Flink's `vector_search` and `full_text_search` procedures rank Data 
Evolution row IDs against a snapshot. The subsequent batch scan could resolve a 
newer snapshot. An overwrite between the two phases dropped a row that matched 
the search.
   
   Pass the snapshot captured by the search plan to the batch scan that fetches 
the selected row IDs. This also avoids resolving a time-travel selector a 
second time.
   
   ### Tests
   
   - Both procedure regressions reproduce an empty result on the merged 
baseline when the data snapshot advances after ranking. Each now reads the 
matching row from the ranking snapshot.
   - JDK 8 Maven `verify` passed with `IndexQuerySplitTest`, both procedure 
regressions, and three existing vector, primary-key vector time-travel, and 
primary-key full-text controls.
   


-- 
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