> Oh, this is bad news. The problem we have is that rules don't
> distinguish the UPDATE on the underlying tables of the rule from other
> updates that may appear in the query.
>
> If we go with Tom's idea and total just UPDATE's, we will get the right
> answer when there is only one UPDATE in the ruleset.
As long as the rules don't overlap (1 row is handled by 1 instead statement,
another row by a different one), it is ok. Again, you can create "non instead"
rules or triggers for the other work needed.
I am still in favor of not distinguishing the different tags. The dba needs to
take responsibility anyway (as long as we don't autogenerate the rules for simple
cases).
Andreas