question

All Oracle statements use cursors to manipulate rows, so every SQL statement that doesn't have a named (explicit) cursor has a default (implicit) cursor.

UPDATE a set b = 1;
DELETE a where b = 1;
SELECT * from a where b = 1;

all use an implicit cursor to do their work.

The implicit cursor can be referenced by SQL%. There are some special attributes defined by Oracle that help you to work with implicit cursors, like SQL%ROWCOUNT, which tells you how many rows were affected by the last implicit cursor:

UPDATE a set b = 1;
dbms_output (SQL%ROWCOUNT);

and you'd get a message with the number of rows the UPDATE affected.

See these links for good explanations:

Cursors

More Cursors

Now, when you want to manage the cursor yourself, you have to explicitly set it up with a name of your choosing and specify the query to run. This link has a pretty good explanation:

Explicit Cursors

Then there's the "cursor for loop" which is as shorthand way of setting up an explicit cursor and is rather neat...

Here's a simple one, where "cur" is my name for a cursor on a select from t1...:

-- there exist tables t1 and t2 with columns c1, c2 and c3 in each... for cur (select c1, c2, c3 from t1) loop -- defines implicit cursor 'cur' within for loop -- inside the 'for' loop, you have access to cur.c1, cur.c2 and cur.c3 select t2.c1, t2.c2, cur.c3 || t2.c3 from t2 where t2.c1 = cur.c1 and t2.c2 = cur.c2; end loop; -- implicit cursor desroyed outside scope of loop -- do other things...

I believe I got all the syntax right but I didn't actually test it.

There's another example, here: Example Cursor