[ https://issues.apache.org/jira/browse/SPARK-17735?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15534069#comment-15534069 ]
Herman van Hovell commented on SPARK-17735: ------------------------------------------- It it better to completely do this on the driver side. UDFs are executed on executors, which does not know anything about Hive. You could create a Seq and turn that into a table once you have determined the isTemp & isCached flags. > Cannot call sqlContext inside udf > --------------------------------- > > Key: SPARK-17735 > URL: https://issues.apache.org/jira/browse/SPARK-17735 > Project: Spark > Issue Type: Bug > Components: Spark Core, SQL > Affects Versions: 1.6.0 > Environment: Linux > Reporter: Saif Addin Ellafi > Priority: Minor > > Hello, I know it is a strange use case but I just wanted to append is cache > to hive tables and turned out into this > {noformat} > def inCacheUdf = org.apache.spark.sql.functions.udf {name: String => > sqlContext.isCached(name)} > val ts = sqlContext.tables > ts.withColumn("foo", inCacheUdf(ts.col("tableName"))).show > {noformat} > .. will throw a Null Pointer Exception. > I am not sure, since in my program I was getting: > A master url must be set in your configuration > or NullPointer at getConf as well. > Thoughts? -- 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