I wrote:
> If you're really intent on making it work this way, my vote is to
> expose namespace.c's isOtherTempNamespace() as a SQL-callable function,
> and add a test on that to the info-schema views, rather than relying on
> is_visible or explicit knowledge of the temp-schema naming convention.
I've done the above and now withdraw my complaints about this patch.
I notice however that the patch seems to have touched only about half a
dozen of the information_schema views ... shouldn't more of them have
similar filters?
regards, tom lane