[ https://issues.apache.org/jira/browse/SPARK-18859?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16076463#comment-16076463 ]
Anton Okolnychyi commented on SPARK-18859: ------------------------------------------ [~maropu] I think the approached suggested by you is a better option here since it will avoid the exception. However, this behavior should be also documented and clear to users. I can create a PR and we can discuss alternative ways to solve this issue there. What do you think? > Catalyst codegen does not mark column as nullable when it should. Causes NPE > ---------------------------------------------------------------------------- > > Key: SPARK-18859 > URL: https://issues.apache.org/jira/browse/SPARK-18859 > Project: Spark > Issue Type: Bug > Components: Optimizer, SQL > Affects Versions: 2.0.2 > Reporter: Mykhailo Osypov > Priority: Critical > > When joining two tables via LEFT JOIN, columns in right table may be NULLs, > however catalyst codegen cannot recognize it. > Example: > {code:title=schema.sql|borderStyle=solid} > create table masterdata.testtable( > id int not null, > age int > ); > create table masterdata.jointable( > id int not null, > name text not null > ); > {code} > {code:title=query_to_select.sql|borderStyle=solid} > (select t.id, t.age, j.name from masterdata.testtable t left join > masterdata.jointable j on t.id = j.id) as testtable; > {code} > {code:title=master code|borderStyle=solid} > val df = sqlContext > .read > .format("jdbc") > .option("dbTable", "query to select") > .... > .load > //df generated schema > /* > root > |-- id: integer (nullable = false) > |-- age: integer (nullable = true) > |-- name: string (nullable = false) > */ > {code} > {code:title=Codegen|borderStyle=solid} > /* 038 */ scan_rowWriter.write(0, scan_value); > /* 039 */ > /* 040 */ if (scan_isNull1) { > /* 041 */ scan_rowWriter.setNullAt(1); > /* 042 */ } else { > /* 043 */ scan_rowWriter.write(1, scan_value1); > /* 044 */ } > /* 045 */ > /* 046 */ scan_rowWriter.write(2, scan_value2); > {code} > Since *j.name* is from right table of *left join* query, it may be null. > However generated schema doesn't think so (probably because it defined as > *name text not null*) > {code:title=StackTrace|borderStyle=solid} > java.lang.NullPointerException > at > org.apache.spark.sql.catalyst.expressions.codegen.UnsafeRowWriter.write(UnsafeRowWriter.java:210) > at > org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown > Source) > at > org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43) > at > org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:370) > at > org.apache.spark.sql.execution.debug.package$DebugExec$$anonfun$3$$anon$1.hasNext(package.scala:146) > at org.apache.spark.util.Utils$.getIteratorSize(Utils.scala:1763) > at org.apache.spark.rdd.RDD$$anonfun$count$1.apply(RDD.scala:1134) > at org.apache.spark.rdd.RDD$$anonfun$count$1.apply(RDD.scala:1134) > at > org.apache.spark.SparkContext$$anonfun$runJob$5.apply(SparkContext.scala:1899) > at > org.apache.spark.SparkContext$$anonfun$runJob$5.apply(SparkContext.scala:1899) > at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:70) > at org.apache.spark.scheduler.Task.run(Task.scala:86) > at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274) > at > java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142) > at > java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617) > at java.lang.Thread.run(Thread.java:745) > {code} -- This message was sent by Atlassian JIRA (v6.4.14#64029) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org