You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kylin.apache.org by "ASF subversion and git services (JIRA)" <ji...@apache.org> on 2018/03/18 08:28:00 UTC
[jira] [Commented] (KYLIN-3281) OLAPProjectRule can't normal
working with projectRel[input=sortRel]
[ https://issues.apache.org/jira/browse/KYLIN-3281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16403916#comment-16403916 ]
ASF subversion and git services commented on KYLIN-3281:
--------------------------------------------------------
Commit 8a9ed59213bb2f2d06ef99e836d7b21e37413112 in kylin's branch refs/heads/sync from [~hn5092]
[ https://gitbox.apache.org/repos/asf?p=kylin.git;h=8a9ed59 ]
KYLIN-3281 OLAPProjectRule can't normal working with projectRel[input=sortRel]
> OLAPProjectRule can't normal working with projectRel[input=sortRel]
> --------------------------------------------------------------------
>
> Key: KYLIN-3281
> URL: https://issues.apache.org/jira/browse/KYLIN-3281
> Project: Kylin
> Issue Type: Bug
> Components: Query Engine
> Affects Versions: v2.3.0
> Reporter: yiming.xu
> Assignee: yiming.xu
> Priority: Critical
>
> for queries like:
> select sum("count") from (SELECT LSTG_FORMAT_NAME , COUNT(*) AS "count" FROM TEST_KYLIN_FACT GROUP BY "LSTG_FORMAT_NAME" ORDER BY "count" DESC) as a
> the expected logical plan after volcano is:
> {code:java}
> OLAPToEnumerableConverter
> OLAPLimitRel(ctx=[], fetch=[50000])
> OLAPAggregateRel(group=[{}], EXPR$0=[SUM($0)], ctx=[])
> OLAPProjectRel(count=[$1], ctx=[])
> OLAPSortRel(sort0=[$1], dir0=[DESC], ctx=[])
> OLAPAggregateRel(group=[{0}], count=[COUNT()], ctx=[])
> OLAPProjectRel(LSTG_FORMAT_NAME=[$3], ctx=[])
> OLAPTableScan(table=[[DEFAULT, TEST_KYLIN_FACT]], ctx=[], fields=[[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26]])
> {code}
> however the actual is:
> {code:java}
> EnumerableLimit(fetch=[50000])
> EnumerableAggregate(group=[{}], EXPR$0=[SUM($0)])
> EnumerableCalc(expr#0..1=[{inputs}], count=[$t1])
> EnumerableSort(sort0=[$1], dir0=[DESC])
> EnumerableAggregate(group=[{0}], count=[COUNT()])
> OLAPToEnumerableConverter
> OLAPProjectRel(LSTG_FORMAT_NAME=[$3], ctx=[])
> OLAPTableScan(table=[[DEFAULT, TEST_KYLIN_FACT]], ctx=[], fields=[[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26]])
> {code}
> VolcanoPlanner apply OLAPProjectRule to logicalProject,convert child node should use child traitset not current proejct traitset.
> If we use current project trait set, the child node trait set is [olap][0 DESC], it will create a new subset but the subset doesn't have any rel. the cost is inf,
> so volcanoPlanner choose a the EnumerableAggregate.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)