yihua opened a new issue, #19673:
URL: https://github.com/apache/hudi/issues/19673

   ### Problem
   
   Hudi has no bundle that ships a native execution engine. Anyone wanting
   [Apache DataFusion Comet](https://datafusion.apache.org/comet/) alongside 
Hudi has to assemble and
   version-match the jars themselves, against a Comet release matrix that does 
not line up with
   Hudi's Spark profiles: Comet publishes one artifact per Spark minor version 
and, at 1.0.0, covers
   only Spark 3.4, 3.5, 4.0 and 4.1.
   
   ### Proposal
   
   Publish `hudi-native-spark-bundle`, the contents of `hudi-spark-bundle` plus 
Comet, for the Spark
   versions Comet actually releases for.
   
   Scope note: Comet does not read Hudi tables natively (it rejects Hudi's file 
format with
   `Unsupported file format HoodieFileGroupReaderBasedFileFormat`), so the 
Parquet decode stays on
   Spark. What the bundle enables is native execution of the rest of the plan: 
joins, shuffles, sorts
   and aggregates over Hudi tables. Making the Hudi scan itself native is 
separate work.
   


-- 
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