clintropolis commented on a change in pull request #10613:
URL: https://github.com/apache/druid/pull/10613#discussion_r553840570
##########
File path:
processing/src/main/java/org/apache/druid/query/groupby/epinephelinae/vector/GroupByVectorColumnProcessorFactory.java
##########
@@ -64,7 +65,16 @@ public GroupByVectorColumnSelector
makeMultiValueDimensionProcessor(
ValueType.STRING == capabilities.getType(),
"groupBy dimension processors must be STRING typed"
);
- throw new UnsupportedOperationException("Multi-value dimensions not yet
implemented for vectorized groupBys");
+ throw new UnsupportedOperationException("Multi-value dimensions are not
yet implemented for vectorized groupBys");
+ }
+
+ @Override
+ public GroupByVectorColumnSelector makeObjectProcessor(
+ ColumnCapabilities capabilities,
+ VectorObjectSelector selector
+ )
+ {
+ throw new UnsupportedOperationException("Object columns are not yet
implemented for vectorized groupBys");
Review comment:
yes, I have a related branch that I will open as a follow-up to this PR
that adds support for group-by queries on vectorized string expressions, using
an object selector and a newly added dictionary building
`GroupByVectorColumnSelector` similar to the non-vectorized group by code-path.
In that branch this part checks that the column capabilities are string typed,
and if so then will make the new dictionary building selector, else throw an
exception for being an unsupported type.
It seems beneficial to not have to make the string expression vector results
have to masquerade as dictionary encoded, to be translated to string values to
be re-encoded into a new dictionary, which is part of my reason for pushing for
object selectors for the use case
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
