[ https://issues.apache.org/jira/browse/SPARK-11803?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15010315#comment-15010315 ]
Xiao Li commented on SPARK-11803: --------------------------------- I believe this must be an urgent issue. This is my first time to read this Dataset implementation. This might be a quick fix for you. I should not block your current progress. Just want to share my idea. Detecting the conflicting attributes in the joinWith function and then reassign the new expression ids to the other.logicalPlan. This should be similar as how the Analyzer does for self joins for handling the duplicate expression ids. Then, we can use the new `other` for joining with `this`. > Dataset self join returns incorrect result > ------------------------------------------ > > Key: SPARK-11803 > URL: https://issues.apache.org/jira/browse/SPARK-11803 > Project: Spark > Issue Type: Sub-task > Components: SQL > Reporter: Reynold Xin > > See the test case in https://github.com/apache/spark/pull/9789 > {code} > ignore("self join") { > val ds = Seq("1", "2").toDS().as("a") > val joined = ds.joinWith(ds, lit(true)) > checkAnswer(joined, ("1", "1"), ("1", "2"), ("2", "1"), ("2", "2")) > } > {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org