Is there a faster form of this query?

Question

I received a table of corrections that have to be 'propagated' to a master table, i.e. the corrections table is composed of a few columns and every value of every column of every row must be used to update the corresponding row-column in the master table. Both tables have id columns that can be used to match rows from the two. The query I was given to do this was:
update master a
set a.calculated_error = ( select calculated_error from corrections b where b.id = a.id ),
... additional columns to update follow with the same syntax.
The columns being updated are not indexed in either table.
The query takes a very long time to run. I suppose I can see why, lots of queries on columns that are not indexed so each query involves a table scan.
Is there a faster form of this query?
Thanks,
-=b

Answer 1

In additon to Herald's suggestion, assuming he is correct in his assumption that corrections is small in relation to master, I might also consider updating all of the columns in one shot. something along the lines of:
update master a
set (a.calculated_error, a.first_add, a.second_add) = (select calculated_error, first_add, second_add
                                                      from corrections b where b.id = a.id )
WHERE a.id in (SELECT id from corrections)Alternatively, if corrections.id is unique, I might also try creating a unique index on it and using an updateable join view like:
UPDATE (SELECT a.calculated_error, a.first_add, b.second_add,
               b.calculated_error new_calc, b.first_add new_first,
               b.second_add new_second
        FROM master a, corrections b
        WHERE a.id = b.id)
SET calculated_error = new_calc,
    first_add = new_first,
    second_add = new_secondIf you need to deal with NULLs in corrections, then use something like:
SET first_add = NVL(new_first, first_add)John