[
https://issues.apache.org/jira/browse/SPARK-12727?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15138930#comment-15138930
]
Xiao Li commented on SPARK-12727:
---------------------------------
{code}
Aggregate [a#16L], [a#16L,(count(if ((gid#23 = 2)) b#25L else
null),mode=Complete,isDistinct=false) AS _c1#20L,(count(if ((gid#23 = 1)) c#24L
else null),mode=Complete,isDistinct=false) AS _c2#21L,(first(if ((gid#23 = 0))
sum(d)#27L else null) ignore nulls,mode=Complete,isDistinct=false) AS _c3#22L]
+- Aggregate [a#16L,c#24L,b#25L,gid#23],
[a#16L,c#24L,b#25L,gid#23,(sum(d#26L),mode=Complete,isDistinct=false) AS
sum(d)#27L]
+- Expand [List(a#16L, null, null, 0, d#19L),List(a#16L, c#18L, null, 1,
null),List(a#16L, null, b#17L, 2, null)], [a#16L,c#24L,b#25L,gid#23,d#26L]
+- Subquery t2
+- Relation[a#16L,b#17L,c#18L,d#19L] ParquetRelation
{code}
Maybe we can split the whole query to two parts. {{Aggregate}} + {{Aggregate +
Expand}} = {{Aggregate}} + {{grouping sets}}.
> SQL generation support for distinct aggregation patterns that fit
> DistinctAggregationRewriter analysis rule
> -----------------------------------------------------------------------------------------------------------
>
> Key: SPARK-12727
> URL: https://issues.apache.org/jira/browse/SPARK-12727
> Project: Spark
> Issue Type: Sub-task
> Components: SQL
> Affects Versions: 2.0.0
> Reporter: Cheng Lian
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
