Re: Function use in query

Ioana Danes wrote

> Hi All,
> Is there any similar syntax that only invokes the procedure once and
> returns all the columns?

Generic, adapt to fit your needs.

WITH func_call AS (
SELECT function_call(...) AS func_out_col
)
SELECT (func_out_col).*
FROM func_call;

Basically you have to execute the function call and leave the result as a
single column (a row type).  Then, in another layer of the query, you expand
that single column into its components using "*".  Because you are expanding
a column and not a table you must put the column name in "()" - otherwise
the parser thinks "func_out_col" is a table and errors out.

This all definitely applies to 9.2 and earlier.  9.3 (with lateral) may
behave differently...

David J.

Hi David,

Thank you for your reply, I haven't thought about it. This works as expected if I don't need to filter the table tmp_Cashdrawer:

select tmp_Cashdrawer.CashdrawerID,
(test1(tmp_Cashdrawer.CashdrawerID)).* from tmp_Cashdrawer where
tmp_Cashdrawer.CashdrawerID in (1);

If I will have to filter the tmp_Cashdrawer table then it executes the function for the all the cash drawers and then filter out the result which again is not efficient...

I might use an aggregate table for this. This way I can use a simple function call to update the aggregate table when a cash drawer is balanced or before executing the report.

Thanks again for your reply,
Ioana