You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafodion.apache.org by "Hans Zeller (JIRA)" <ji...@apache.org> on 2016/09/28 00:01:45 UTC
[jira] [Updated] (TRAFODION-2248) WITH clause with CTE used in
subquery gives error
[ https://issues.apache.org/jira/browse/TRAFODION-2248?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Hans Zeller updated TRAFODION-2248:
-----------------------------------
Description:
Here is a simple example for a failing query:
{noformat}
with cse1 as (select * from (values (1)) T(a))
select count(*)
from cse1
where a > (select avg(a) from cse1);
*** ERROR[3288] A syntax error occurred. WITH clause redefined. WITH name CSE1 .
*** ERROR[8822] The statement was not prepared.
{noformat}
There are really two things going on here:
* We hit an assert in RelRoot::transformNode() that assumes there are no subqueries above the RelRoot.
* The error we see is different, however. That's because the initial assert triggers AQR (auto-query retry). When we compile the query a second time, it still has the CTE from the first time, and that causes the error 3288. So, we also need to initialize the parser list of CTEs before a recompilation.
This is the original assert we hit:
{noformat}
// We are currently assuming that no subqueries have been introduced above me;
// any new subquery parent would just silently be ignored!
CMPASSERT( this == locationOfPointerToMe ); // Genesis 10-970828-6025
{noformat}
was:
Here is a simple example for a failing query:
{noformat}
with cse1 as (select * from (values (1)) T(a))
select count(*)
from cse1
where a > (select avg(a) from cse1);
*** ERROR[3288] A syntax error occurred. WITH clause redefined. WITH name CSE1 .
*** ERROR[8822] The statement was not prepared.
{noformat}
There are really two things going on here:
* We hit an assert in RelRoot::transformNode() that assumes there are no subqueries above the RelRoot.
* The error we see is different, however. That's because the error triggers AQR (auto-query retry). When we compile the query a second time, it still has the CTE from the first time, and that causes the error 3288. So, we also need to initialize the parser list of CTEs before a recompilation.
This is the original assert we hit:
{noformat}
// We are currently assuming that no subqueries have been introduced above me;
// any new subquery parent would just silently be ignored!
CMPASSERT( this == locationOfPointerToMe ); // Genesis 10-970828-6025
{noformat}
> WITH clause with CTE used in subquery gives error
> -------------------------------------------------
>
> Key: TRAFODION-2248
> URL: https://issues.apache.org/jira/browse/TRAFODION-2248
> Project: Apache Trafodion
> Issue Type: Bug
> Components: sql-cmp
> Affects Versions: 2.0-incubating
> Reporter: Hans Zeller
>
> Here is a simple example for a failing query:
> {noformat}
> with cse1 as (select * from (values (1)) T(a))
> select count(*)
> from cse1
> where a > (select avg(a) from cse1);
> *** ERROR[3288] A syntax error occurred. WITH clause redefined. WITH name CSE1 .
> *** ERROR[8822] The statement was not prepared.
> {noformat}
> There are really two things going on here:
> * We hit an assert in RelRoot::transformNode() that assumes there are no subqueries above the RelRoot.
> * The error we see is different, however. That's because the initial assert triggers AQR (auto-query retry). When we compile the query a second time, it still has the CTE from the first time, and that causes the error 3288. So, we also need to initialize the parser list of CTEs before a recompilation.
> This is the original assert we hit:
> {noformat}
> // We are currently assuming that no subqueries have been introduced above me;
> // any new subquery parent would just silently be ignored!
> CMPASSERT( this == locationOfPointerToMe ); // Genesis 10-970828-6025
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)