[
https://issues.apache.org/jira/browse/ARROW-15591?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17535917#comment-17535917
]
Weston Pace commented on ARROW-15591:
-------------------------------------
First of all, how did you get that cool red text?
The measure:
{noformat}
"measure": {
"functionReference": 0,
"args": [],
"sorts": [],
"phase": "AGGREGATION_PHASE_INITIAL_TO_RESULT",
"outputType": {
"i64": {}
}
}
{noformat}
will have to map to:
{noformat}
Aggregate{"hash_count", CountOptions{}}
{noformat}
{quote}
If we take a look at the measures(Optional, required if no grouping sets), it
contains the `functionReference` which can be
extracted from the extensions where the `function_anchor` is what is referred
by `functionReference`.
{quote}
Yes. We have to do the same thing for scalar functions. We should reuse code
where we can. Mapping aggregate functions is probably a separate JIRA all on
its own. You might want to take a look at ARROW-15582 where I propose a
general way of doing this. I believe [~sanjibansg] has a prototype for scalar
functions but he is out until next week so this ticket might be best to wait
until he is back and we can collaborate.
> [C++] Add support for aggregation to the Substrait consumer
> -----------------------------------------------------------
>
> Key: ARROW-15591
> URL: https://issues.apache.org/jira/browse/ARROW-15591
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: Weston Pace
> Assignee: Vibhatha Lakmal Abeykoon
> Priority: Major
> Labels: substrait
>
> The streaming execution engine supports aggregation (i.e. group by). The
> Substrait consumer does not currently consume aggregation relations. We
> should add support for this.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
