[
https://issues.apache.org/jira/browse/SPARK-17818?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15553777#comment-15553777
]
Todd Nemet commented on SPARK-17818:
------------------------------------
I don't believe you can CAST AS NULL in SQL since NullType isn't exposed at the
SQL layer. Not an expert though.
> Cannot SELECT NULL
> ------------------
>
> Key: SPARK-17818
> URL: https://issues.apache.org/jira/browse/SPARK-17818
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 2.0.0, 2.0.1
> Reporter: Todd Nemet
>
> When connecting to Spark SQL via JDBC/thriftserver, running `SELECT NULL`
> returns this error:
> {code}
> 0: jdbc:hive2://localhost:10016/> select cast(NULL as NULL);
> Error: org.apache.spark.sql.catalyst.parser.ParseException:
> DataType null() is not supported.(line 1, pos 20)
> {code}
> This is a regression from Spark 1.x/Hive QL compatibility. In Spark 1.5.2,
> The same query returns NULL:
> {code}
> 0: jdbc:hive2://localhost:10015/> select null;
> +-------+--+
> | _c0 |
> +-------+--+
> | NULL |
> +-------+--+
> 1 row selected (0.658 seconds)
> {code}
> I can select expressions that evaluate to NULL or CAST NULL to different
> types:
> {code}
> 0: jdbc:hive2://localhost:10016/> select null=null
> 0: jdbc:hive2://localhost:10016/> ;
> +----------------+--+
> | (NULL = NULL) |
> +----------------+--+
> | NULL |
> +----------------+--+
> 1 row selected (0.69 seconds)
> 0: jdbc:hive2://localhost:10016/> select cast (NULL as int);
> +--------------------+--+
> | CAST(NULL AS INT) |
> +--------------------+--+
> | NULL |
> +--------------------+--+
> 1 row selected (0.676 seconds)
> 0: jdbc:hive2://localhost:10016/> select cast (NULL as string);
> +-----------------------+--+
> | CAST(NULL AS STRING) |
> +-----------------------+--+
> | NULL |
> +-----------------------+--+
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
