[
https://issues.apache.org/jira/browse/SPARK-20202?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15956879#comment-15956879
]
Steve Loughran commented on SPARK-20202:
----------------------------------------
# the ugliness need to inset the spark thrift stuff under the hive thrift stuff
is obsolete, can be cut entirely.
# with the shading of kryo not needed, an unshaded hive *may* work. I forget
which troublespots there were last time, probably the usual suspects: jackson,
guava, etc.
# Hive 1.2.x refuses to work with Hadoop 3; it considers that an unsupported
version. For basic client-side testing, you can build Hadoop 3 with a fake
version (e..g {{mvn install -DskipShade -Ddeclared.hadoop.version=2.11}}, but
as hadoop version is something which NN/DNs care about, not something that's
really going to work in real systems. Presumably later hive versions will
address that.
If hive take over ownership of the spark 1.2.1-spark branch, this could be done
first simply by pulling the spark branch into the Hive repo as a branch,
defining the artifact naming properly and releasing it. If that is done, before
any release of that 1.2.x branch is done, there's a couple of outstanding PRs
to pull in (groovy version for security reasons, ... ).. A quick import &
re-release would be the fast way to get this out as an asf-approved binary
> Remove references to org.spark-project.hive
> -------------------------------------------
>
> Key: SPARK-20202
> URL: https://issues.apache.org/jira/browse/SPARK-20202
> Project: Spark
> Issue Type: Bug
> Components: Build, SQL
> Affects Versions: 1.6.4, 2.0.3, 2.1.1
> Reporter: Owen O'Malley
> Priority: Blocker
>
> Spark can't continue to depend on their fork of Hive and must move to
> standard Hive versions.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
