spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Takeshi Yamamuro (Jira)" <j...@apache.org>
Subject [jira] [Created] (SPARK-30279) Support 32 or more grouping attributes for GROUPING_ID
Date Tue, 17 Dec 2019 03:36:00 GMT
Takeshi Yamamuro created SPARK-30279:
----------------------------------------

             Summary: Support 32 or more grouping attributes for GROUPING_ID 
                 Key: SPARK-30279
                 URL: https://issues.apache.org/jira/browse/SPARK-30279
             Project: Spark
          Issue Type: Bug
          Components: SQL
    Affects Versions: 3.0.0
            Reporter: Takeshi Yamamuro


This ticket targets to support 32 or more grouping attributes for GROUPING_ID. In the current
master, an integer overflow can occur to compute grouping IDs;
https://github.com/apache/spark/blob/e75d9afb2f282ce79c9fd8bce031287739326a4f/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/basicLogicalOperators.scala#L613

For example, the query below generates wrong grouping IDs in the master;
{code}
scala> val numCols = 32 // or, 31
scala> val cols = (0 until numCols).map { i => s"c$i" }
scala> sql(s"create table test_$numCols (${cols.map(c => s"$c int").mkString(",")},
v int) using parquet")
scala> val insertVals = (0 until numCols).map { _ => 1 }.mkString(",")
scala> sql(s"insert into test_$numCols values ($insertVals,3)")
scala> sql(s"select grouping_id(), sum(v) from test_$numCols group by grouping sets ((${cols.mkString(",")}),
(${cols.init.mkString(",")}))").show(10, false)
scala> sql(s"drop table test_$numCols")

// numCols = 32
+-------------+------+
|grouping_id()|sum(v)|
+-------------+------+
|0            |3     |
|0            |3     | // Wrong Grouping ID
+-------------+------+

// numCols = 31
+-------------+------+
|grouping_id()|sum(v)|
+-------------+------+
|0            |3     |
|1            |3     |
+-------------+------+
{code}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message