Question

Lets say we have the following:

create view view_1 as
(
  select key, data from some_table_or_view;
);

create table table_1
(
  key
  more_data
);

create view view_2 as
(
  select v1.key, v1.data, t1.more_data 
  from view_1 v1, table_1 t1 
  where v1.key = t1.key
);

create table table_2 as
(
  key
  data
  more_data
);

create view view_3 as
(
  select key, data, more_data from view_2
  union
  select key, data, more_data from table_2
);

So basically, we have joined two data feeds, one that comes from a join, one that is raw data.

I want to be able to do the following.

update view_3 set more_data = 'BLAH_MORE_DATA' where key = 'BLAH_KEY';

Where this somehow updates either table_1 or table_2 depending whether “BLAH_KEY” comes from table_1 or table_2.

Currently the only way I can think of is:

create view view_3 as
(
  select 'TAB1' as source, key, data, more_data from view_2
  union
  select 'TAB2' as source, key, data, more_data from table_2
);

and then doing updates using a PL/SQL function which checks the source column and updates the appropriate table, but this means I have to kept the view and the PL/SQL function in sync.

I’m looking for a nicer way to do this.

Answer 1

You can use an instead of trigger on the view_3

CREATE OR REPLACE TRIGGER view3_trg
INSTEAD OF  UPDATE
ON view_3
FOR EACH ROW
BEGIN
IF :OLD.source = 'TAB1' THEN

  UPDATE table_1 t1
  set t1.data = :NEW.data
  WHERE
    t1.key = :OLD.key
  ;
ELSIF OLD.source = 'TAB2' THEN
   UPDATE table_2 t2
  set t2.data = :NEW.data
  WHERE
    t2.key = :OLD.key
  ;
END IF;
END;
/