[
https://issues.apache.org/jira/browse/SPARK-5117?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Cheng Hao resolved SPARK-5117.
------------------------------
Resolution: Won't Fix
This IS NOT a bug of Spark SQL.
Hive changed the LPAD implementation since Hive 0.13, previously(Hive 0.12) the
LPAD was implemented as the UDF, and the UDF Bridge will cast the un-match
argument type when necessary. However, in Hive 0.13, LPAD was re-implemented as
Generic UDF, which means the function itself has to take care of the arguments
type, and will throws exception if un-matched argument type found.
I've also test that in Hive 0.13, it says:
{panel}
hive> SELECT LPAD("test",5, 0) FROM src LIMIT 1;
FAILED: ClassCastException org.apache.hadoop.io.IntWritable cannot be cast to
org.apache.hadoop.io.Text
hive>
{panel}
>From the document of Hive:
{panel}
lpad(string str, int len, string pad)
{panel}
The third parameter should be String-based type, the end-user should take care
of the new usage for this function.
> Hive Generic UDFs don't cast correctly
> --------------------------------------
>
> Key: SPARK-5117
> URL: https://issues.apache.org/jira/browse/SPARK-5117
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 1.2.0
> Reporter: Michael Armbrust
> Assignee: Cheng Hao
> Priority: Blocker
>
> Here's a test cast that is failing in master:
> {code}
> createQueryTest("generic udf casting",
> """SELECT LPAD("test",5, 0) FROM src LIMIT 1""")
> {code}
> This appears to be a regression from Spark 1.1
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
