Function Call returning old SQL Query

Question

Hello All,
I have a Pipeline Function which creates a SQL within (Dynamic SQL that gets stored in a LONG variable) based on the parameter STRING passed to the function. Inside this function, once the SQL is built, I am inserting this SQL into a log table, for logging purpose.
Note: my function has only one parameter which is a string. This string accepts a name:value pairs with a delimiter which I breakdown inside the function. But this functionality is working fine.
Issue:
When I run the function with parameter with a STRING say (Age = 20, Gender = M) for the first time, it works.
<code>SELECT * FROM TABLE (
PIPE_FUN_SEARCH_PKG.get_search_records ('EMP_AGE:20|EMP_GENDER:M'));
</code>
When I change the parameters to (Age = 20, Gender = F), it gives me the results of the earlier function call.
<code>SELECT * FROM TABLE (
PIPE_FUN_SEARCH_PKG.get_search_records ('EMP_AGE:20|EMP_GENDER:F'));
</code>
When I open the logs, I see the SQL being built is the earlier one.
As a test I closed the session and ran (Age = 20, Gender = F) first. It works fine. When I run a different parameter string, it always mimics the earlier function call.
Is CACHING in play here. I tried both the following:
<code> dbms_result_cache.bypass(FALSE);
dbms_result_cache.flush;
</code>
I tried multiple tests, with different parameters and only the first one runs fine and second one copied the earlier. However, when I open two sessions on two different windows it doesn't happen.
Also, in the Logging table I am capturing the input string as a confirmation, which is coming correctly. But the SQL being build mimics the earlier call.
I tried to set the variable which hold the SQL Statement to empty (v_sql := '';) at the beginning and also at the end. Still no use.
Kindly help if I am over looking anything.
Regards,
Aj

Answer 1

Aj09 wrote:
I have a Pipeline Function which creates a SQL within (Dynamic SQL that gets stored in a LONG variable) based on the parameter STRING passed to the function. The LONG data type has been replaced by the LOB data type. Oracle specifically recommends not using the old LONG data type.
Issue:
When I run the function with parameter with a STRING say (Age = 20, Gender = M) for the first time, it works.
<code>SELECT * FROM TABLE (
PIPE_FUN_SEARCH_PKG.get_search_records ('EMP_AGE:20|EMP_GENDER:M'));
</code>
When I change the parameters to (Age = 20, Gender = F), it gives me the results of the earlier function call.
<code>SELECT * FROM TABLE (
PIPE_FUN_SEARCH_PKG.get_search_records ('EMP_AGE:20|EMP_GENDER:F'));
</code>The tag is ** - not *<code>*.
Why a pipeline function? Why dynamic SQL? Are you using +DBMS_SQL+ to create the dynamic cursor? If not, why not? Only +DBMS_SQL+ allows dynamic binding in PL/SQL. Without that, your code will burn a lot of additional CPU on hard parsing and trash and fragment Shared Pool memory.
When I open the logs, I see the SQL being built is the earlier one.
How do you record the current SQL? Are you using a static variable to capture the SQL statement generated?
From what you have described - this is yet another horribly flawed approach in all respects. To data modelling. To relational databases. To Oracle. To SQL.
Reinventing the SQL language for data retrieval as a pipeline function using a funky parameter interface - sorry, I just don't get that. It is an insane approach.