From: JensRie@gmx.de   
      
   Hi Ed,   
      
   > Jens, can you describe the actual concepts behind "a" and "b" in   
   > your situation?  I suspect that Joe is right, but that it would   
   > be easier to understand when applied to a specific example.   
      
   table "a" and "b" have no specific constraint or similar.   
   In the current case we fill several tables during a data import.   
   If we want repeat the import we have to delete all data that has been   
   stored during the first run.   
   Most of the tables have a field which refers to a specific import batch,   
   but some miss this field. But there are some relations where one can   
   find out which data belongs to the import by joining them with a table   
   containing the import ID.   
   So we use something like   
      
   delete from table_a   
   where relation_id in   
      (select some_id from table_b where import_id = 25)   
      
   Because it is allowed that table_a contains data which has no relation   
   to table_b there is no constraint or cascade-delete construct we could use.   
   The only thing I'm concerned about is that the execution of the   
   statement could slow down the complete DB if the sub-query delivers   
   thousands ore more IDs for the Where-In-condition.   
      
   Regards   
   Jens   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)