On Wed, Oct 12, 2005 at 07:28:37PM -0400, Tom Lane wrote:
> Oliver Elphick <olly@lfix.co.uk> writes:
> > On Wed, 2005-10-12 at 17:45 -0400, Tom Lane wrote:
> >> Hm. Could we see the actual pg_attribute data for both this table and
> >> its parent?
>
> > Here you are:
>
> Thanks. Nothing particularly strange-looking there though. Do you want
> to try tracing through COPY with a debugger to see where it's getting
> the bogus column order from?
Do you get the same bogosity with the attached test case? It's
based on the same table definition and data from your first message;
about the only differences from what you're doing are that the table
isn't inherited and it uses "COPY foo FROM stdin" instead of reading
the data from an external file. If this test works correctly, what
happens if you use "COPY FROM stdin" on the real table?
--
Michael Fuhr