raman118 opened a new pull request, #41489:
URL: https://github.com/apache/beam/pull/41489

   Flink's ordered-list state used the timestamp as a map key when reading 
stored values. This collapsed entries with the same timestamp. A range clear 
then wrote the collapsed list back to state, dropping entries outside the range 
as well.
   
   Sort and filter the stored list directly in the shared Flink sources and the 
Flink 2.0 override. The state descriptor and serializer stay the same. The 
regression covers distinct entries and repeated identical entries at the same 
timestamp, chronological reads, and the inclusive-start, exclusive-end range 
boundaries.
   
   Fixes #39782.
   
   **Validation**
   
   - The new regression failed with the original checkout sources on Flink 
1.19.0 and 2.0.1.
   - All 26 tests in each FlinkStateInternalsTest suite passed with the patched 
checkout sources compiled against Beam 2.78.0 dependencies.
   - Google Java Format 1.17.0 and git diff --check passed.
   
   **AI disclosure**
   
   Codex (GPT-6) investigated the issue and generated the patch, regression 
tests, and PR description. The commit includes a Generated-by trailer.
   


-- 
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]

  • [PR] [Flink] Preserve entries ... raman118 (via GitHub) via github

Reply via email to