carloea2 opened a new issue, #8257:
URL: https://github.com/apache/texera/issues/8257

   ### Current behavior
   
   The JVM range partitioner converts `Double.NaN` to long zero before routing, 
while the Python partitioner performs arithmetic on NaN and raises 
`ValueError`. A workflow containing a NaN range key therefore crashes only in 
the Python engine.
   
   Before: NaN range key raises ValueError in Python
   
   After: NaN is treated as zero and routed like the JVM
   
   ### Steps to reproduce
   
   On `50321e403c82df299a13deb50a7f9849dd93bdba`, range-shuffle a NaN key 
across receivers A and B with minimum -10 and maximum 9.
   
   Observed result:
   
   ```text
   ValueError: cannot convert float NaN to integer
   scala_nan_to_long_index=1
   ```
   
   ### Expected behavior
   
   Python should treat a NaN range key as zero, matching JVM double-to-long 
conversion and routing it to receiver B for this range.


-- 
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