Brook Milligan <brook@trillium.NMSU.Edu> writes:
> pg_dump won't drop stuff before trying to create it; this makes
> dropping a single table (say) and recreating it difficult to automate
> since it is subject to error if all the stuff doesn't get dropped
> properly. The following patch causes pg_dump to emit DROP ...
> statements prior to emitting CREATE ... statements.
I think that ought to be driven by a switch to pg_dump ...
much of the time, I would *want* pg_dump's script to fail
if there's already an existing item of the given name.
I see your concern, I just don't think it's the only scenario.
regards, tom lane