Author: zly
Date: Sun Mar 12 21:13:49 2017
New Revision: 1786613

URL: http://svn.apache.org/viewvc?rev=1786613&view=rev
Log:
PIG-5180: MergeSparseJoin fails with Spark exec type (Adam via Liyun)

Modified:
    
pig/branches/spark/src/org/apache/pig/backend/hadoop/executionengine/spark/JobGraphBuilder.java

Modified: 
pig/branches/spark/src/org/apache/pig/backend/hadoop/executionengine/spark/JobGraphBuilder.java
URL: 
http://svn.apache.org/viewvc/pig/branches/spark/src/org/apache/pig/backend/hadoop/executionengine/spark/JobGraphBuilder.java?rev=1786613&r1=1786612&r2=1786613&view=diff
==============================================================================
--- 
pig/branches/spark/src/org/apache/pig/backend/hadoop/executionengine/spark/JobGraphBuilder.java
 (original)
+++ 
pig/branches/spark/src/org/apache/pig/backend/hadoop/executionengine/spark/JobGraphBuilder.java
 Sun Mar 12 21:13:49 2017
@@ -59,6 +59,7 @@ import org.apache.pig.impl.PigContext;
 import org.apache.pig.impl.plan.DependencyOrderWalker;
 import org.apache.pig.impl.plan.OperatorKey;
 import org.apache.pig.impl.plan.VisitorException;
+import org.apache.pig.newplan.logical.relational.LOJoin;
 import org.apache.pig.tools.pigstats.spark.SparkPigStats;
 import org.apache.pig.tools.pigstats.spark.SparkStatsUtil;
 import org.apache.spark.api.java.JavaSparkContext;
@@ -121,7 +122,9 @@ public class JobGraphBuilder extends Spa
         if (poMergeJoins.size() > 0) {
             for (POMergeJoin poMergeJoin : poMergeJoins) {
                 String idxFileName = poMergeJoin.getIndexFile();
-                filesForMoreReplication.add(new Path(idxFileName));
+                if (idxFileName != null) {
+                    filesForMoreReplication.add(new Path(idxFileName));
+                }
                 // in spark mode, set as null so that PoMergeJoin won't use 
hadoop distributed cache
                 // see POMergeJoin.seekInRightStream()
                 poMergeJoin.setIndexFile(null);


Reply via email to