carloea2 opened a new pull request, #8172:
URL: https://github.com/apache/texera/pull/8172

   ### What changes were proposed in this PR?
   
   Preserve numeric precision while the Scala range shuffle calculates a 
bucket. `DOUBLE` keys no longer lose their fractional part, while integer and 
long keys retain exact arithmetic.
   
   Before: with two receivers over -10 to 9, -0.5 went to bucket 1 in Scala and 
bucket 0 in Python.
   
   After: both implementations send -0.5 to bucket 0, while the 0.0 boundary 
remains in bucket 1.
   
   ### Any related issues, documentation, discussions?
   
   Closes #8171
   
   ### How was this PR tested?
   
   Added a regression test that covers a negative fractional key and the 
adjacent zero boundary.
   
   Red before the source change, then 7 tests passed:
   
   ```text
   sbt -no-colors "WorkflowExecutionService/testOnly 
org.apache.texera.amber.engine.architecture.messaginglayer.RangeBasedShuffleSpec"
   ```
   
   Formatting and static checks passed:
   
   ```text
   sbt -no-colors scalafmtCheckAll "scalafixAll --check"
   ```
   
   The production Scala partitioner was also run directly in the Scala console. 
It returned bucket 0 for -0.5 and bucket 1 for 0.0.
   
   ### Was this PR authored or co-authored using generative AI tooling?
   
   Generated-by: OpenAI Codex, GPT-5


-- 
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