[
https://issues.apache.org/jira/browse/HIVE-6380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13892810#comment-13892810
]
Jason Dere commented on HIVE-6380:
----------------------------------
Actually I want to start with something even simpler than what is in the UDF
doc - the CREATE FUNCTION will just reference the paths of the files used:
CREATE FUNCTION udfname AS 'my.udf.class'
USING JAR '/path/to/myjar.jar', FILE '/path/to/file'
USING ARCHIVE '/path/to/archive.tgz';
This is similar to the earlier proposed syntax of using a resource defined in
Hive (and doesn't preclude that functionality from being added later on); but
is a bit easier to start with as there are less database objects to create and
no management of resource files on the part of Hive.
If possible, Hive should try to make ensure that these referenced files are
using the same filesystem as the warehouse directory, with the idea that these
UDF resources should be accessible to any Hive clients. The function creator
should also be responsible for making sure the file remains at the path
specified at function create time.
> Specify jars/files when creating permanent UDFs
> -----------------------------------------------
>
> Key: HIVE-6380
> URL: https://issues.apache.org/jira/browse/HIVE-6380
> Project: Hive
> Issue Type: Sub-task
> Components: UDF
> Reporter: Jason Dere
> Assignee: Jason Dere
>
> Need a way for a permanent UDF to reference jars/files.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
