Re: pg_upgrade and relkind filtering

Robert Haas <robertmhaas@gmail.com> writes:
> On Mon, Dec 5, 2011 at 5:06 PM, Bruce Momjian <bruce@momjian.us> wrote:
>> Pg_upgrade has the following check to make sure the cluster is safe for
>> upgrading:
>> 
>> What types, other than views, can we skip in this query?

> It's not obvious to me that anything other than a table or index would matter.

You'd better complain about composite types too, since one of them could
be a column in a table.  (Unless you want to test to see whether it
actually is stored anywhere, but that seems like way overkill for this.)

			regards, tom lane