[jira] [Created] (HUDI-4039) Make sure builtin key-generators can efficiently fetch record-key, partition-path

["Alexey Kudinkin (Jira)" <ji...@apache.org>]

Alexey Kudinkin created HUDI-4039:
-------------------------------------

             Summary: Make sure builtin key-generators can efficiently fetch record-key, partition-path
                 Key: HUDI-4039
                 URL: https://issues.apache.org/jira/browse/HUDI-4039
             Project: Apache Hudi
          Issue Type: Improvement
            Reporter: Alexey Kudinkin
            Assignee: Alexey Kudinkin


With the introduction of row-writing Key Generators become aware of low-level Spark components like `Row`, `InternalRow` to be able to fetch record-key, and partition-path w/o the need to convert received object into Avro.

To further improve on performance of these, we should also try to avoid doing conversions from Spark's native data-types (for ex, `UTF8String`) into JVM native ones (like `String`) as this might entail unnecessary encoding/decoding, as well as underlying buffers being copied for no good use.



--
This message was sent by Atlassian Jira
(v8.20.7#820007)