Question

I have a table like this:

A
ID_A    ID_1   ID_2
1       1      (null)
2       3      (null)
3       7      (null)

B
ID_B    ID_1   ID_2
1       (null) 2
2       (null) 4
3       (null) 6

REF     
ID_A    ID_B
1       2
3       1

According to the ref table, the object with an ID_A of 2 is the same object as an ID_B of 1.

Therefore, I should be able to update the table this way:

A
ID_A    ID_1   ID_2
1       1      4
2       3      (null)
3       7      2

Indeed, you get that result if you do this query:

select
  A.ID_A, B.ID_1, C.ID_2
from
  A, B, REF
where
  A.ID_A = REF.ID_A
  AND REF.ID_B = B.ID_B

(actually you lose the null row because it’s an inner join, but that’s not the point.)

What I am completely unable to do is update A with this new information! I either get “single-row subquery returns more than one row” with an update or the lovely result that my query is non-deterministic with a merge.

Given that I really do have the three tables as I’ve shown, how can I write a query to update id_2 correctly?
;

Answer 1

merge into A w
using(select a.id_a
           , b.id_2
       from a 
       join rf on (a.id_a = rf.id_a)
       join b   on (b.id_b = rf.id_b) 
     ) q
  on (q.id_a = w.id_a )
when matched then
  update 
     set w.id_2 = q.id_2

SQL Fiddle