[ https://issues.apache.org/jira/browse/PIG-4456?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
liyunzhang_intel updated PIG-4456: ---------------------------------- Description: Sort the leaves by SparkOperator.operatorKey to guarantee that sparkOperator with smaller operatorKey is executed before those with larger operatorKey if both are the leaves of spark plan, for example, sparkOperator(scope-65) and sparkOperator(scope-67) are all leaves of spark plan. SparkOperator(scope-65) should be executed before sparkOperator(scope-67) because the output of scope-65 maybe is used by scope-67. (was: Sort the leaves by SparkOperator.operatorKey to guarantee that sparkOperator with smaller operatorKey is executed before those with larger operatorKey if both are the leaves of spark plan, for example, sparkOperator(scope-65) and sparkOperator(scope-67) are all leaves of spark plan. SparkOperator(scope-65) should be executed before sparkOperator(scope-67) because the output of scope-65 maybe is used by scope-67. This patch fixs following unit test failures: org.apache.pig.test.TestLimitVariable.testLimitVariable1 org.apache.pig.test.TestLimitVariable.testLimitVariable2 org.apache.pig.test.TestLimitVariable.testLimitVariable3 org.apache.pig.test.TestLimitVariable.testLimitVariable4 org.apache.pig.test.TestSample.testSample_VariableNone org.apache.pig.test.TestSample.testSample_VariableSome org.apache.pig.test.TestScalarAliases.testScalarAliasesSplitClause) > Sort the leaves by SparkOperator.operatorKey in SparkLauncher#sparkOperToRDD > ---------------------------------------------------------------------------- > > Key: PIG-4456 > URL: https://issues.apache.org/jira/browse/PIG-4456 > Project: Pig > Issue Type: Sub-task > Components: spark > Reporter: liyunzhang_intel > Assignee: liyunzhang_intel > Attachments: PIG-4456.patch > > > Sort the leaves by SparkOperator.operatorKey to guarantee that sparkOperator > with smaller operatorKey is executed before those with larger operatorKey if > both are the leaves of spark plan, for example, sparkOperator(scope-65) and > sparkOperator(scope-67) are all leaves of spark plan. SparkOperator(scope-65) > should be executed before sparkOperator(scope-67) because the output of > scope-65 maybe is used by scope-67. -- This message was sent by Atlassian JIRA (v6.3.4#6332)