[ https://issues.apache.org/jira/browse/SPARK-6123?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14344278#comment-14344278 ]
Yin Huai commented on SPARK-6123: --------------------------------- To workaround this issue, users need to load the existing data (the one with containsNull=false) and write the data to a new dir. > Parquet reader should use the schema of every file to create converter > ---------------------------------------------------------------------- > > Key: SPARK-6123 > URL: https://issues.apache.org/jira/browse/SPARK-6123 > Project: Spark > Issue Type: Bug > Components: SQL > Reporter: Yin Huai > > For two parquet files for the same table having an array column, if values of > the array in one file was created when containsNull was true and those in > another file was created when containsNull was false, the containsNull in the > merged schema will be true and we cannot correctly read data from the table > created with containsNull=false. -- 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