[jira] [Updated] (HIVE-6593) Create a maven assembly for hive-jdbc

["Szehon Ho (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HIVE-6593?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Szehon Ho updated HIVE-6593:
----------------------------

    Attachment: HIVE-6593.patch

I made a sub-project of 'packaging' called 'jdbc-packaging' that makes the following assembly when hive builds with -Pdist profile.  The contents are like so:

{noformat}
./lib/commons-logging-1.1.3.jar
./lib/hive-exec-0.14.0-SNAPSHOT.jar
./lib/hive-jdbc-0.14.0-SNAPSHOT.jar
./lib/hive-metastore-0.14.0-SNAPSHOT.jar
./lib/hive-serde-0.14.0-SNAPSHOT.jar
./lib/hive-service-0.14.0-SNAPSHOT.jar
./lib/libfb303-0.9.0.jar
./lib/libthrift-0.9.0.jar
./lib/log4j-1.2.16.jar
{noformat}

Let me know if this works.

> Create a maven assembly for hive-jdbc
> -------------------------------------
>
>                 Key: HIVE-6593
>                 URL: https://issues.apache.org/jira/browse/HIVE-6593
>             Project: Hive
>          Issue Type: Improvement
>          Components: Build Infrastructure
>    Affects Versions: 0.12.0
>            Reporter: Mark Grover
>            Assignee: Szehon Ho
>         Attachments: HIVE-6593.patch
>
>
> Currently in Apache Bigtop we bundle and distribute Hive. In particular, for users to not have to install the entirety of Hive on machines that are just jdbc clients, we have a special package which is a subset of hive, called hive-jdbc that bundles only the jdbc driver jar and it's dependencies.
> However, because Hive doesn't have an assembly for the jdbc jar, we have to hack and hardcode the list of jdbc jars and it's dependencies:
> https://github.com/apache/bigtop/blob/master/bigtop-packages/src/rpm/hive/SPECS/hive.spec#L361
> As Hive moves to Maven, it would be pretty fantastic if Hive could leverage the maven-assembly-plugin and generate a .tar.gz assembly for what's required for jdbc gateway machines. That we can simply take that distribution and build a jdbc package from it without having to hard code jar names and dependencies. That would make the process much less error prone.



--
This message was sent by Atlassian JIRA
(v6.2#6252)