Re: pg_am.amowner

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> It seems that access methods nominally have an "owner", but that owner is
> nowhere else referenced. Since there is no user interface for adding
> access methods anyway, would there be any problems with removing that
> field?

Hmm ... offhand I'm having a hard time seeing that it would make sense
to associate protection checks with an access method. The only use
I can see for the owner field is to control who could delete an access
method --- and I don't have much problem with saying "only the
superuser". It's even harder to believe that we'd really want non-
superusers installing access methods.

But the other side of the coin is what harm is it doing? Surely you're
not worried about the space occupied by the column ;-)

regards, tom lane