I was specifically speaking of stored SQL procedures. This restricts you to a sequence of queries (no control structures).
I can pass in table names to PL/pgSQL but haven't figured out exactly what I'm allowed to do afterwards.
I can probably refer to them by declaring them to be of type RECORD?
That's how NEW and OLD are passed to triggers. The semantics once you get the parameters passed in and declared correctly are there, obviously, since triggers work. The question is whether or not I can define such magic at the user level.
I should play with this, maybe I will this afternoon. Reducing the use of EXECUTE was what I had in mind, indeed. I have some things I have to do today, so I won't be playing with this for awhile ...
