Csaba Ringhofer created IMPALA-14882:
----------------------------------------
Summary: Add efficient conversion between RowBatch and arrow
Key: IMPALA-14882
URL: https://issues.apache.org/jira/browse/IMPALA-14882
Project: IMPALA
Issue Type: New Feature
Components: Backend
Reporter: Csaba Ringhofer
There is already arrow -> RowBatch conversion in PaimonScanner, but it is done
per row and is unlikely to be efficient.
Fast "transposing" in both direction would allow experimenting with arrow in
more places, for example in aggregates without group by.
At first I would focus on RowBatches with a single tuple per row.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
