Re: pg_dump versus rules, once again

I wrote:
> We've talked before about replacing this _RETURN-rule business with
> CREATE OR REPLACE VIEW, ie the idea would be for pg_dump to first emit
> a dummy view with the right column names/types, say
> CREATE VIEW vv AS SELECT null::int AS f1, null::text AS f2;
> and then later when it's safe, emit CREATE OR REPLACE VIEW with the view's
> real query.

Here's a proposed patch for that. It turns out there are some other
kluges that can be gotten rid of while we're at it: no longer need the
idea of reloptions being attached to a rule, for instance.

The changes in pg_backup_archiver.c would have to be back-patched
into all versions supporting --if-exists, so that they don't fail
on dump archives produced by patched versions. We could possibly
put the rest only into HEAD; I remain of two minds about that.

regards, tom lane