Re: pg_dump's restore gives "operator does not exist: public.iprange = public.iprange" but copy paste works - Mailing list pgsql-general

"David G. Johnston" <david.g.johnston@gmail.com> writes:
> IIUC, functions can force a search_path even during dump/restore by being
> created with one specified as part of the create function command.  Since
> the issue is with stored objects moreso than queries generically is it
> feasible to approach the view solution by adding a "SET" clause to CREATE
> VIEW as well?

Don't see the point.  The issue here is what search_path applies at
view definition time, and you have syntax to control that today:

    SET search_path = whatever;
    CREATE VIEW ... ;
    RESET search_path;

So the problem is not lack of a server feature, it's persuading pg_dump
to emit something other than what it does now.

            regards, tom lane