cshuo opened a new issue, #19821:
URL: https://github.com/apache/hudi/issues/19821

   ## Feature Description
   
   Add support for reading Hudi VECTOR columns from Lance base files with Flink.
   
   Flink readers should preserve the VECTOR semantics declared by the requested 
Hudi schema while returning values through the existing Flink ARRAY 
representation. Files written by either Flink or Spark with the Lance format 
should be readable consistently across engines.
   
   ## Expected Behavior
   
   - FLOAT and DOUBLE VECTOR columns can be read from Lance base files.
   - Nullable VECTOR values are returned correctly.
   - Full-row reads and column projections preserve VECTOR values and field 
ordering.
   - VECTOR columns can be read in COW and MOR tables, including upsert and 
compaction results.
   - Regular ARRAY columns continue to behave as regular Lance arrays.
   - File schema discovery restores Hudi VECTOR identity when the Lance file 
contains the corresponding VECTOR metadata.
   - Unsupported or incompatible VECTOR encodings fail with a clear error.
   
   ## Scope
   
   This work reuses the existing Flink VECTOR configuration and Hudi VECTOR 
data model. It does not add support for INT8 VECTOR columns on Lance, nested 
VECTOR columns, or Lance schema evolution.
   
   Parent tracking issue: #19053


-- 
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