[
https://issues.apache.org/jira/browse/SPARK-6503?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael Armbrust resolved SPARK-6503.
-------------------------------------
Resolution: Won't Fix
Shim that compile differently are no longer part of our story for hive
compatibility. Instead use {{spark.sql.hive.metastore.version}}.
> Create Jenkins builder for testing Spark SQL with Hive 0.12.0
> -------------------------------------------------------------
>
> Key: SPARK-6503
> URL: https://issues.apache.org/jira/browse/SPARK-6503
> Project: Spark
> Issue Type: Sub-task
> Components: SQL
> Affects Versions: 1.2.0, 1.3.0
> Reporter: Cheng Lian
>
> Currently, from the perspective of Spark SQL, the {{dev/run-tests}} script
> does the following steps to build and test Spark SQL:
> # Builds Spark SQL against Hive 0.12.0 to check for compilation errors
> # Cleans the build
> # Builds Spark SQL against Hive 0.13.1 to check for compilation errors.
> # Runs unit tests against Hive 0.13.1
> Apparently, Spark SQL with Hive 0.12.0 is not tested.
> Two improvements could be done here:
> # When executed, {{dev/run-tests}} should always build and test a single
> version of Hive. The version could be passed in as environment variable.
> # Separate Jenkins builders should be set up to test Hive 0.12.0 code paths.
> We probably only want the PR builder run against Hive 0.13.1 to minimize
> build time, and make the master builder take care of both Hive versions.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
