[SPARK-18058] AnalysisException may be thrown when union two DFs whose struct fields have different nullability

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.6.2, 2.0.1
• Fix Version/s: 2.0.2, 2.1.0
• Component/s: SQL
• Target Version/s: 2.0.2, 2.1.0
• Assignee: Nan Zhu
• Reporter: Cheng Lian
• Created: 21/Oct/16 22:30
• Updated: 28/Nov/16 16:40
• Resolved: 23/Oct/16 17:45

Description

The following Spark shell snippet reproduces this issue:

spark.range(10).createOrReplaceTempView("t1")
spark.range(10).map(i => i: java.lang.Long).toDF("id").createOrReplaceTempView("t2")
sql("SELECT struct(id) FROM t1 UNION ALL SELECT struct(id) FROM t2")

org.apache.spark.sql.AnalysisException: Union can only be performed on tables with the compatible column types. StructType(StructField(id,LongType,true)) <> StructType(StructField(id,LongType,false)) at the first column of the second table;
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$class.failAnalysis(CheckAnalysis.scala:40)
  at org.apache.spark.sql.catalyst.analysis.Analyzer.failAnalysis(Analyzer.scala:57)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$checkAnalysis$1$$anonfun$apply$11$$anonfun$apply$12.apply(CheckAnalysis.scala:291)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$checkAnalysis$1$$anonfun$apply$11$$anonfun$apply$12.apply(CheckAnalysis.scala:289)
  at scala.collection.immutable.List.foreach(List.scala:381)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$checkAnalysis$1$$anonfun$apply$11.apply(CheckAnalysis.scala:289)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$checkAnalysis$1$$anonfun$apply$11.apply(CheckAnalysis.scala:278)
  at scala.collection.immutable.List.foreach(List.scala:381)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$checkAnalysis$1.apply(CheckAnalysis.scala:278)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$checkAnalysis$1.apply(CheckAnalysis.scala:67)
  at org.apache.spark.sql.catalyst.trees.TreeNode.foreachUp(TreeNode.scala:132)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$class.checkAnalysis(CheckAnalysis.scala:67)
  at org.apache.spark.sql.catalyst.analysis.Analyzer.checkAnalysis(Analyzer.scala:57)
  at org.apache.spark.sql.execution.QueryExecution.assertAnalyzed(QueryExecution.scala:49)
  at org.apache.spark.sql.Dataset$.ofRows(Dataset.scala:61)
  at org.apache.spark.sql.SparkSession.sql(SparkSession.scala:573)
  ... 50 elided

The reason is that we treat two StructType incompatible even if their only differ from each other in field nullability.

Issue Links

• links to: [Github] Pull Request #15595 (CodingCat)
• links to: [Github] Pull Request #15602 (CodingCat)
• links to: [Github] Pull Request #16041 (hvanhovell)