Robert Haas <> writes:
> On Sun, Jan 5, 2014 at 3:45 PM, Tom Lane <> wrote:
>> pg_dump creates extensions before tables, no?  So what dump/restore
>> hazard?

> Creating the extension doesn't guarantee that the shared library will
> always be loaded.

No, but unless the plan is that no validation happens at all (which
I gather is not your desire) then there must be some mechanism for
figuring out which extension owns a given reloption and asking it
to validate the value.  This might be more complicated than a passive
hook, but I don't feel bad about demanding that it work like that.

                        regards, tom lane

Sent via pgsql-hackers mailing list (
To make changes to your subscription:

Reply via email to