[
https://issues.apache.org/jira/browse/DRILL-5051?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15676527#comment-15676527
]
Khurram Faraaz commented on DRILL-5051:
---------------------------------------
The problem is only when the OFFSET is specified in the subquery. When there is
no OFFSET specified in the sub-query, query returns correct results.
{noformat}
0: jdbc:drill:schema=dfs.tmp> select count(col_int) from (select col_int from
(select col_int from typeall_l limit 2) limit 1 offset 1);
+---------+
| EXPR$0 |
+---------+
| 0 |
+---------+
1 row selected (0.231 seconds)
0: jdbc:drill:schema=dfs.tmp> explain plan for select count(col_int) from
(select col_int from (select col_int from typeall_l limit 2) limit 1 offset 1);
+------+------+
| text | json |
+------+------+
| 00-00 Screen
00-01 Project(EXPR$0=[$0])
00-02 StreamAgg(group=[{}], EXPR$0=[COUNT($0)])
00-03 Limit(offset=[1], fetch=[1])
00-04 Limit(fetch=[2])
00-05 Scan(groupscan=[ParquetGroupScan [entries=[ReadEntryWithPath
[path=maprfs:///tmp/typeall_l]], selectionRoot=maprfs:/tmp/typeall_l,
numFiles=1, usedMetadataFile=false, columns=[`col_int`]]])
{noformat}
Without OFFSET in sub-query
{noformat}
0: jdbc:drill:schema=dfs.tmp> select count(1) from (select col_int from (select
col_int from typeall_l limit 2) limit 1);
+---------+
| EXPR$0 |
+---------+
| 1 |
+---------+
1 row selected (0.242 seconds)
{noformat}
> Returning incorrect number of rows while querying using both nested select
> and offset
> -------------------------------------------------------------------------------------
>
> Key: DRILL-5051
> URL: https://issues.apache.org/jira/browse/DRILL-5051
> Project: Apache Drill
> Issue Type: Bug
> Components: Query Planning & Optimization
> Affects Versions: 1.8.0
> Environment: Fedora 24 / OpenJDK 8
> Reporter: Hongze Zhang
> Fix For: Future
>
>
> My SQl:
> select count(1) from (select id from (select id from
> cp.`tpch/lineitem.parquet` LIMIT 2) limit 1 offset 1)
> This SQL returns nothing.
> Something goes wrong in LimitRecordBatch.java, and the reason is different
> with [DRILL-4884|https://issues.apache.org/jira/browse/DRILL-4884?filter=-2]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)