Hi all,
I'm investigating adding geospatial user-defined functions and types to
Spark SQL in Spark 2.0.x. That is going rather well; I've seen how to
add geospatial UDT and UDFs (and even UDAFs!).
As part of the investigation, I tried out the Thrift JDBC server, and I
have encountered two general issues.
First, there are a few places in the Hive Thrift Server module which
aren't general. In two places, a SessionState is cast to a
HiveSessionState. Additionally, in SparkExecuteStatementOperation, some
serialization code for handling DataTypes doesn't consider
UserDefinedTypes.
For those issues, would a JIRA and pull request be appropriate?
My second issue was connecting existing JDBC code to the Hive JDBC
connector. The code calls connection.getMetaData.getTables(). I tried
various options to see that call work, and I was unable to get the list
of tables back.
It looks like these tickets are related:
https://issues.apache.org/jira/browse/SPARK-9686
https://issues.apache.org/jira/browse/SPARK-9333
Has there been any more work on the JDBC metadata issues?
Thanks in advance,
Jim
---------------------------------------------------------------------
To unsubscribe e-mail: dev-unsubscr...@spark.apache.org
