Alvaro Herrera <alvherre@commandprompt.com> writes:
> I wrote such a patch some time ago but never applied it ...
> http://archives.postgresql.org/message-id/20091116162531.GA3669@alvh.no-ip.org
> I still wonder if it's useful enough to be applied. Would it solve your
> use case?
Needs to be fixed to behave sanely for the typbyval case (as
implemented, it's depending on the caller to check that).
Otherwise, seems pretty reasonable.
regards, tom lane