XMLTable issue in 10.2.0.3 or privs problem?

Question

I stole the following SQL from
Help with a simple XMLTable query
I have been scratching my head trying to figure out why the SQL statement I coded up with a XMLTable would not run on the 10.2.0.3 DB I'm connected to. I finally found the following SQL statement to use as a known good test case
select *
from xmltable
'//OrderMetrics/OrderItemMetric'
passing xmltype(
'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<echoems:ECHOMetrics xmlns:echoems="http://myURL/myFilename.xsd"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://myURL/myFilename.xsd
http://myURL/myFilename.xsd">
<OrderMetrics>
<OrderItemMetric>
<requestor>user1</requestor>
<product>123</product>
</OrderItemMetric>
<OrderItemMetric>
<requestor>0.0.0.0</requestor>
<product>ABC</product>
</OrderItemMetric>
<OrderItemMetric>
<requestor>user3</requestor>
<product>XYZ</product>
</OrderItemMetric>
</OrderMetrics>
</echoems:ECHOMetrics>')
columns
prod varchar2(20) path 'product'
)This executes cleanly on 10.2.0.1 but returns an ORA-01780: string literal required on 10.2.0.3
What am I missing, in terms of the SQL or setup/rights/grants/etc in 10.2.0.3?

Answer 1

SQL> select * from xmltable ( '//OrderMetrics/OrderItemMetric' passing xmltype( '<?xml version="1.0"
encoding="UTF-8" standalone="yes"?> <echoems:ECHOMetrics xmlns:echoems="http://myURL/myFilename.xsd
" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://myURL/myFilename.
xsd http://myURL/myFilename.xsd"> <OrderMetrics> <OrderItemMetric> <requestor>user1</requestor> <pro
duct>123</product> </OrderItemMetric> <OrderItemMetric> <requestor>0.0.0.0</requestor> <product>ABC<
/product> </OrderItemMetric> <OrderItemMetric> <requestor>user3</requestor> <product>XYZ</product> <
/OrderItemMetric> </OrderMetrics> </echoems:ECHOMetrics>') columns prod varchar2(20) path 'product'
  2  /
PROD
123
ABC
XYZ
SQL> show parameter cursor
NAME                                 TYPE        VALUE
cursor_sharing                       string      EXACT
cursor_space_for_time                boolean     FALSE
open_cursors                         integer     300
session_cached_cursors               integer     20
SQL> alter session set cursor_sharing='SIMILAR';
Session altered.
SQL> select * from xmltable ( '//OrderMetrics/OrderItemMetric' passing xmltype( '<?xml version="1.0"
encoding="UTF-8" standalone="yes"?> <echoems:ECHOMetrics xmlns:echoems="http://myURL/myFilename.xsd
" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://myURL/myFilename.
xsd http://myURL/myFilename.xsd"> <OrderMetrics> <OrderItemMetric> <requestor>user1</requestor> <pro
duct>123</product> </OrderItemMetric> <OrderItemMetric> <requestor>0.0.0.0</requestor> <product>ABC<
/product> </OrderItemMetric> <OrderItemMetric> <requestor>user3</requestor> <product>XYZ</product> <
/OrderItemMetric> </OrderMetrics> </echoems:ECHOMetrics>') columns prod varchar2(20) path 'product'
  2  /
PROD
123
ABC
XYZ
SQL> alter session set cursor_sharing='force';
Session altered.
SQL> select * from xmltable ( '//OrderMetrics/OrderItemMetric' passing xmltype( '<?xml version="1.0"
encoding="UTF-8" standalone="yes"?> <echoems:ECHOMetrics xmlns:echoems="http://myURL/myFilename.xsd
" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://myURL/myFilename.
xsd http://myURL/myFilename.xsd"> <OrderMetrics> <OrderItemMetric> <requestor>user1</requestor> <pro
duct>123</product> </OrderItemMetric> <OrderItemMetric> <requestor>0.0.0.0</requestor> <product>ABC<
/product> </OrderItemMetric> <OrderItemMetric> <requestor>user3</requestor> <product>XYZ</product> <
/OrderItemMetric> </OrderMetrics> </echoems:ECHOMetrics>') columns prod varchar2(20) path 'product'
  2  /
PROD
123
ABC
XYZAs you can see all three variations of cursor sharing work in the 10.2.0.4 patch set. So whatever is causing this in 10.2.0.3 is fixed now. If you want to try and have Oracle determine the change open an SR with your testcase and we can file a bug to see if we can determine where this was fixed.
Regards
Coby D. Adams Jr.
Edited by: mdrake on Feb 15, 2010 7:35 AM