Oracle cursor question

Question

Hello,
When using cursors, is it the same thing if i issue:
cursor c is .....
r c%rowtype;
begin
for r in c loop .... end loop;or
loop
open c;
fetch c into r;
exit when c%NOTFOUND;
-- do some actions....
end loop;?
Or is it any difference?
Thanks
Edited by: Roger22 on 23.06.2010 10:59

Answer 1

user13165454 wrote:
I understand both are almost same process to iterate through the result set. One is iterating the cursor result set while the other one is using for loop iteration.There is no such thing as a "+cursor result set+" - this creates the illusion that a physical set of results are created by the cursor and the client is fetching rows from this data set.
Does not work like this. Where would such a result set lives? In database memory? What then about scalability? Such a result set can potentially contain millions of rows. A database server can be required to service 1000's of open cursors at the same time.
Such a result set approach cannot scale as it will make a huge demand on memory.
The results of a cursor is the output of a cursor. The cursor (the set instructions to find the relevant rows) are executed by the server. Rows are found. These are output. And the process repeats with each fetch where the cursor finds the next set of rows (based on Oracle multi version concurrency control model) to output to the client.
As for the difference between a FOR cursor loop and a FETCH cursor loop - simply one of implicit versus explicit cursor variable management. There is no difference in how the server-side SQL cursor is used.
E.g.
-- FETCH loop                 -- FOR loop
open c;                       for varBuffer in (...)  -- cursor is implicitly opened
loop                          loop
  fetch c into varBuffer;          -- n.a. as the fetch is implicitly performed into the buffer variable "varBuffer"
  exit when c%notfound;            -- n.a. as loop will automatically terminated when %notfound
  .. process ..                    .. process ..
end loop;                    end loop;     
close c;                     -- n.a. as cursor is implictly closedWhich one is faster? Neither really. Both do the same things..
a) open a cursor on the Oracle server
b) fetch from the cursor
c) close the cursor
Whether that is implicitly or explicitly done, is irrelevant in terms of performance in this case.
What is important is how the cursor on the client side acts. The FOR loop does an implicit bulk collect of a 100 rows (10g and later). With the FETCH loop one needs to manually code the bulk collect. But there one has more flexibility as one can also use bulk processing (i.e. a FORALL DML). This is not possible in the FOR loop as it does not provide access to the implicit collection used for the bulk fetch.
On the Oracle side.. both will use the VERY SAME CURSOR for the same source SQL statement. It is important to note that "cursor performance" is not really relevant on the Oracle server side. Whether the client uses a ref cursor, DBMS_SQL cursor, implicit cursor, explicit cursors.. the Oracle server side does not know, does not need to know and does not care.
This is a client side method for interacting with the cursor in the server.. and how "fast" that interaction (e.g single fetch vs bulk fetch) is with the cursor on the server is a client issue.