Zouxxyy commented on PR #11317:
URL: https://github.com/apache/paimon/pull/11317#issuecomment-7065402886

   Thanks for the detailed reproducer and for comparing the two physical plans. 
I was able to reproduce the TopN behavior you described.
   
   #11303 already fixed the writer-side issue for newly written files. When a 
floating-point column contains NaN, PyPaimon records its file bounds as 
unknown, so TopN keeps that file as a candidate. For historical snapshots that 
still reference files with incorrect bounds, I’d leave this as a known 
limitation. A reader-side workaround would add complexity to TopN planning for 
a benefit limited to those older files.
   
   This PR fixes a separate writer issue in the manifest list. When a manifest 
contains partition values `5.0` and `NaN`, PyPaimon could record `5.0` as the 
partition upper bound. The fix records `NaN` regardless of entry order, so a 
predicate such as `p > 5` can still find the NaN partition.
   


-- 
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