[
https://issues.apache.org/jira/browse/DRILL-2235?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14610664#comment-14610664
]
Aman Sinha commented on DRILL-2235:
-----------------------------------
This query plans successfully since we now support (since Drill 1.0)
NestedLoopJoin with scalar subqueries. Here's a plan with query agains TPC-H:
{code}
0: jdbc:drill:zk=local> explain plan for select n1.n_name from
cp.`tpch/nation.parquet` n1 where (n1.n_nationkey, n1.n_regionkey) not in
(select n2.n_nationkey, n2.n_regionkey from cp.`tpch/nation.parquet` n2 where
n2.n_regionkey < 10);
+------+------+
| text | json |
+------+------+
| 00-00 Screen
00-01 Project(n_name=[$0])
00-02 SelectionVectorRemover
00-03 Filter(condition=[NOT(CASE(=($1, 0), false, IS NOT NULL($7),
true, IS NULL($3), null, IS NULL($4), null, <($2, $1), null, false))])
00-04 HashJoin(condition=[AND(=($3, $5), =($4, $6))],
joinType=[left])
00-06 Project(n_name=[$2], $f0=[$3], $f1=[$4], f5=[$0], f6=[$1])
00-08 NestedLoopJoin(condition=[true], joinType=[inner])
00-11 Project(n_nationkey=[$2], n_regionkey=[$0], n_name=[$1])
00-14 Scan(groupscan=[ParquetGroupScan
[entries=[ReadEntryWithPath [path=classpath:/tpch/nation.parquet]],
selectionRoot=classpath:/tpch/nation.parquet, numFiles=1,
columns=[`n_nationkey`, `n_regionkey`, `n_name`]]])
00-10 StreamAgg(group=[{}], agg#0=[COUNT()], agg#1=[COUNT($0,
$1)])
00-13 Project($f0=[$0], $f1=[$1], $f2=[true])
00-16 SelectionVectorRemover
00-18 Filter(condition=[<($1, 10)])
00-20 Project(n_nationkey=[$1], n_regionkey=[$0])
00-21 Scan(groupscan=[ParquetGroupScan
[entries=[ReadEntryWithPath [path=classpath:/tpch/nation.parquet]],
selectionRoot=classpath:/tpch/nation.parquet, numFiles=1,
columns=[`n_nationkey`, `n_regionkey`]]])
00-05 Project($f00=[$0], $f10=[$1], $f2=[$2])
00-07 HashAgg(group=[{0, 1}], agg#0=[MIN($2)])
00-09 Project($f0=[$0], $f1=[$1], $f2=[true])
00-12 SelectionVectorRemover
00-15 Filter(condition=[<($1, 10)])
00-17 Project(n_nationkey=[$1], n_regionkey=[$0])
00-19 Scan(groupscan=[ParquetGroupScan
[entries=[ReadEntryWithPath [path=classpath:/tpch/nation.parquet]],
selectionRoot=classpath:/tpch/nation.parquet, numFiles=1,
columns=[`n_nationkey`, `n_regionkey`]]])
{code}
However, note that the StreamAgg is doing a COUNT($0, $1) .. it seems Calcite
generates such an aggregate expression. I am not sure what is the semantics of
count(a, b). Running this query fails during execution because we don't
support this function.
> Assert when NOT IN clause contains multiple columns
> ---------------------------------------------------
>
> Key: DRILL-2235
> URL: https://issues.apache.org/jira/browse/DRILL-2235
> Project: Apache Drill
> Issue Type: Bug
> Components: Query Planning & Optimization
> Affects Versions: 0.8.0
> Reporter: Victoria Markman
> Assignee: Aman Sinha
> Fix For: 1.2.0
>
>
> {code}
> 0: jdbc:drill:schema=dfs> select * from t1;
> +------------+------------+------------+
> | a1 | b1 | c1 |
> +------------+------------+------------+
> | 1 | aaaaa | 2015-01-01 |
> | 2 | bbbbb | 2015-01-02 |
> | 3 | ccccc | 2015-01-03 |
> | 4 | null | 2015-01-04 |
> | 5 | eeeee | 2015-01-05 |
> | 6 | fffff | 2015-01-06 |
> | 7 | ggggg | 2015-01-07 |
> | null | hhhhh | 2015-01-08 |
> | 9 | iiiii | null |
> | 10 | jjjjj | 2015-01-10 |
> +------------+------------+------------+
> 10 rows selected (0.056 seconds)
> 0: jdbc:drill:schema=dfs> select * from t2;
> +------------+------------+------------+
> | a2 | b2 | c2 |
> +------------+------------+------------+
> | 0 | zzz | 2014-12-31 |
> | 1 | aaaaa | 2015-01-01 |
> | 2 | bbbbb | 2015-01-02 |
> | 2 | bbbbb | 2015-01-02 |
> | 2 | bbbbb | 2015-01-02 |
> | 3 | ccccc | 2015-01-03 |
> | 4 | ddddd | 2015-01-04 |
> | 5 | eeeee | 2015-01-05 |
> | 6 | fffff | 2015-01-06 |
> | 7 | ggggg | 2015-01-07 |
> | 7 | ggggg | 2015-01-07 |
> | 8 | hhhhh | 2015-01-08 |
> | 9 | iiiii | 2015-01-09 |
> +------------+------------+------------+
> 13 rows selected (0.069 seconds)
> {code}
> IN clause returns correct result:
> {code}
> 0: jdbc:drill:schema=dfs> select count(*) from t1 where (a1, b1) in (select
> a2, b2 from t2);
> +------------+
> | EXPR$0 |
> +------------+
> | 7 |
> +------------+
> 1 row selected (0.258 seconds)
> {code}
> NOT IN clause asserts:
> {code}
> 0: jdbc:drill:schema=dfs> select count(*) from t1 where (a1, b1) not in
> (select a2, b2 from t2);
> Query failed: AssertionError: AND(AND(NOT(IS TRUE($7)), IS NOT NULL($3)), IS
> NOT NULL($4))
> Error: exception while executing query: Failure while executing query.
> (state=,code=0)
> {code}
> {code}
> #Thu Feb 12 12:13:26 EST 2015
> git.commit.id.abbrev=de89f36
> {code}
> drillbit.log
> {code}
> 2015-02-12 22:47:11,730 [2b22d290-315e-4450-8b3f-9b3590eb20c3:foreman] INFO
> o.a.drill.exec.work.foreman.Foreman - State change requested. PENDING -->
> FAILED
> org.apache.drill.exec.work.foreman.ForemanException: Unexpected exception
> during fragment initialization: AND(AND(NOT(IS TRUE($7)), IS NOT NULL($3)),
> IS NOT NULL($4))
> at org.apache.drill.exec.work.foreman.Foreman.run(Foreman.java:197)
> [drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> at
> org.apache.drill.exec.work.WorkManager$RunnableWrapper.run(WorkManager.java:303)
> [drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
> [na:1.7.0_71]
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
> [na:1.7.0_71]
> at java.lang.Thread.run(Thread.java:745) [na:1.7.0_71]
> Caused by: java.lang.AssertionError: AND(AND(NOT(IS TRUE($7)), IS NOT
> NULL($3)), IS NOT NULL($4))
> at org.eigenbase.rel.FilterRelBase.<init>(FilterRelBase.java:56)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at org.eigenbase.rel.FilterRel.<init>(FilterRel.java:50)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at org.eigenbase.rel.CalcRel.createFilter(CalcRel.java:212)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> org.eigenbase.sql2rel.SqlToRelConverter.convertWhere(SqlToRelConverter.java:840)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> org.eigenbase.sql2rel.SqlToRelConverter.convertSelectImpl(SqlToRelConverter.java:497)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> org.eigenbase.sql2rel.SqlToRelConverter.convertSelect(SqlToRelConverter.java:474)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> org.eigenbase.sql2rel.SqlToRelConverter.convertQueryRecursive(SqlToRelConverter.java:2657)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> org.eigenbase.sql2rel.SqlToRelConverter.convertQuery(SqlToRelConverter.java:432)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> net.hydromatic.optiq.prepare.PlannerImpl.convert(PlannerImpl.java:186)
> ~[optiq-core-0.9-drill-r18.jar:na]
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.convertToRel(DefaultSqlHandler.java:163)
> ~[drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> at
> org.apache.drill.exec.planner.sql.handlers.DefaultSqlHandler.getPlan(DefaultSqlHandler.java:126)
> ~[drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> at
> org.apache.drill.exec.planner.sql.DrillSqlWorker.getPlan(DrillSqlWorker.java:145)
> ~[drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> at
> org.apache.drill.exec.work.foreman.Foreman.runSQL(Foreman.java:515)
> [drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> at org.apache.drill.exec.work.foreman.Foreman.run(Foreman.java:188)
> [drill-java-exec-0.8.0-SNAPSHOT-rebuffed.jar:0.8.0-SNAPSHOT]
> ... 4 common frames omitted
> 2015-02-12 22:47:11,747 [2b22d290-315e-4450-8b3f-9b3590eb20c3:foreman] ERROR
> o.a.drill.exec.work.foreman.Foreman - Error
> 8f0bb8dd-deac-4846-9608-e941da9035e8: AssertionError: AND(AND(NOT(IS
> TRUE($7)), IS NOT NULL($3)), IS NOT NULL($4))
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
