peterxcli opened a new issue, #5547:
URL: https://github.com/apache/datafusion-comet/issues/5547

   ### Problem
   
   The native Arrow C Data exporter constructs `FFI_ArrowSchema` from 
`ArrayData.data_type()`. That preserves the physical datatype but drops the 
top-level Field name, nullability, and metadata. Logical types represented by 
Field metadata cannot survive that boundary reliably.
   
   Parent: #5546
   
   ### Solution
   
   - Pass the corresponding RecordBatch Field to every native array-export call.
   - Construct `FFI_ArrowSchema` from that Field.
   - Keep `FFI_ArrowArray` export unchanged.
   - Ensure offset-normalized arrays still use the original output Field.
   - Substitute embedded NUL only in the exported C-string name while 
preserving datatype, nullability, and metadata.
   
   This task must not add Variant scan admission or Variant-specific schema 
inference.
   
   ### Tests
   
   - Top-level Field name, nullability, and metadata survive schema export.
   - Offset-normalized arrays retain the original Field.
   - A top-level name containing NUL does not panic or lose metadata.
   - Existing non-Variant array exports remain unchanged.
   


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to