At what point is PL/SQL cursor data retrieved?

Question

Hi
This may sound like a silly question but since I am not entirely sure, I thought I'd ask.
In Oracle PL/SQL, do you know at what stage the query in a cursor is actually executed?
Is it at the DECLARE stage or when you first OPEN / LOOP on it?
I suspect upon the OPEN or the LOOP.
I.e. if I had a cursor declared but did not loop (using FOR .. LOOP) through it for some hours, would the query results be as at the time of declare or as at the time of the OPEN / FOR...LOOP?
E.g.
CURSOR invoices
  IS
    SELECT invoice_number
    FROM unpaid_invoices
    WHERE due_date < TRUNC (SYSDATE)
BEGIN
-- Some time consuming processing that will take an hour
-- Start Loop through cursor of invoices.
  <<overdue_invoices>>
  FOR inv IN invoices LOOP
-- Some processing on the invoice
  END LOOP overdue_invoices;
END;Would the cursor details for invoices be as at the CURSOR declaration or when the FOR inv... is reached?
Thanks
Martin

Answer 1

When you use a for loop like that it implicitly opens the cursor at the start and closes it at the end of the loop.
But the SQL query in that cursor is actually "run" when the first implicit or explicit fetch is issued, and retrieves new rows for every subsequent fetch.
Of course, before rows can be retrieved there needs to be several other tasks.
There's a good description of the steps of SQL processing in this part of the documentation.
http://download.oracle.com/docs/cd/E11882_01/appdev.112/e10471/adfns_sqlproc.htm
PL/SQL implicitly declares a cursor for all SQL data manipulation statements, including queries that return only one row. For queries that return multiple rows, you can explicitly declare a cursor to process the rows individually.
A cursor is a handle to a specific private SQL area. In other words, a cursor can be thought of as a name for a specific private SQL area. A PL/SQL cursor variable enables the retrieval of multiple rows from a stored subprogram.Edited by: fsitja on Dec 15, 2009 1:03 AM