[ https://issues.apache.org/jira/browse/HIVE-2239?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ashutosh Chauhan updated HIVE-2239: ----------------------------------- Attachment: hive-2239.patch Ok, at runtime we don't want to include any jar file which hadoop anyway will include in its classpath. For that we simply need to delete the existing jar file from ${hive-root}/lib and we will be good. However, doing so will make hive fail to compile. So, this patch retrieves the jar and put it in the build/ivy/lib/default so that it could be used at compile time (and also while running tests). This I think is the cleanest approach, since hive doesn't force any version and whichever hadoop is running it will use its version of jar file. I upped the version in ivy/libraries.properties since we need to up it one day in any case, but strictly that is not required. After applying this patch, please necessarily delete lib/common-codecs-1.3.jar > Up the commons-codec version from 1.3 to 1.4 > -------------------------------------------- > > Key: HIVE-2239 > URL: https://issues.apache.org/jira/browse/HIVE-2239 > Project: Hive > Issue Type: Task > Components: Contrib > Reporter: Ashutosh Chauhan > Assignee: Ashutosh Chauhan > Attachments: commons-codec-1.4.jar, hive-2239.patch > > > Hadoop upgraded from 1.3 to 1.4. Since, hive is launched from hadoop scripts, > it gets 1.4 from there and then it includes 1.3 itself. Having two different > versions of same jar causes unwanted grief. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira