Re: functional index search path issue.

On Saturday, July 5, 2025, Kirill Reshke <reshkekirill@gmail.com> wrote:

>
> Is this a postgres fault? One thing is with pg_upgrade failing for a
> valid database, but maybe users should just re-create their indices
> with fully-qualified names.
>

Yes, functions you intend for the server to execute during dump/restore or
other administrative routines where it sets up a safe searh_path need to be
able to resolve all identifiers without the benefit of a session-defined
search_path beyond pg_catalog.  Public is not part of the safe search_path.

David J.