On 08/25/2010 03:29 AM, Andrew Dunstan wrote:
>
>
> I've just been taking another look at this suggestion. I think it will
> work quite cleanly. As long as we add the enums in the correct order
> it should just do the Right Thing (tm).
>
> To answer your other question, Oid wraparound will not be a problem.
>
> Will get coding.
>
>
Revised patch with pg_dump and pg_upgrade support is attached.
cheers
andrew