Nested Subselect in UPDATE statement

Question

Hi everybody,
I have the following problem: There are two tables FOO and BAR that both have columns X and Y and a primary key. I want to fill the values X and Y of FOO with the values X and Y from BAR where FOO.key = BAR.key . There may be multiple rows from BAR with the same key so I simply select the first one. The query would look like
UPDATE foo f
   SET ( x, y ) = ( SELECT *
                      FROM ( SELECT b.x, b.y
                               FROM bar b
                              WHERE f.key = b.key )                    
                     WHERE ROWNUM = 1 );However, this doesn't work because the alias F is not known inside the nested subquery. If I remove the outer subquery ( the one with ROWNUM=1 ) it works fine. But I need to limit the result to a single row ( in the real application I sort the innermost query by date ). Why does Oracle forget the meaning of alias F within the nested subquery?
Pat

Answer 1

Hi, Pat,
A subquery can be correlated only to its immediate parent query, not its grandparent.
You can get the results you want using MERGE instead of UPDATE.
You can also get the same results with only one sub-query, something like this:
UPDATE foo f
   SET ( x, y ) = ( SELECT MAX (x) KEEP (DENSE_RANK LAST ORDER BY a_date)
                           ,        MAX (y) KEEP (DENSE_RANK LAST ORDER BY a_date)
              FROM   bar
              WHERE  key     = f.key
            );I suspect MERGE will be more efficient.
Edited by: Frank Kulash on Mar 9, 2010 11:02 AM