[
https://issues.jboss.org/browse/TEIID-4705?page=com.atlassian.jira.plugin...
]
Ramesh Reddy commented on TEIID-4705:
-------------------------------------
[~ejlp12] See the confusing part is, you mention CDH as the system, that is Clodera's
distribution, which is recommend to be used with Impala JDBC Driver and impala translator.
When your system is Open Source Apache Hive or Horton Works Distribution then it is
recommended to use the Hive JDBC driver and Hive translator in Teiid. You seem to be
using both interchangeably. The exception is logged against impala, but logs indicate
behavior of "hive" translator. The impala translator's nvl(x, y) function is
converted to isnull (x, y), where as hive translator converts nvl(x,y) to coalesce(x, y).
So, the queries are being converted correctly per translators. I have *not* seen the case
when
{code}
select nvl(e1, '') from table1, table2 where table1.e1 = table2.e1
{code}
to
{code}
select isnull(e1) from table1, table2 where table1.e1 = table2.e1
{code}
in fact the code converts to
{code}
select isnull(e1,'') from table1, table2 where table1.e1 = table2.e1
{code}
so, I advise you check what is mis-configuration step you did in your setup?
correct the nvl pushdown for the impala translator
--------------------------------------------------
Key: TEIID-4705
URL:
https://issues.jboss.org/browse/TEIID-4705
Project: Teiid
Issue Type: Bug
Components: Misc. Connectors
Reporter: Johnathon Lee
Assignee: Steven Hawkins
queries consisting of the nvl(x, y) function will fail since JDV translates to
isnull(x,y)
--
This message was sent by Atlassian JIRA
(v7.2.3#72005)