Quite right, "create sequence" occuring after the "create table" is not the
cause of the problem. Thanks, Tom. (Now I'm suspecting a character
translation problem, most likely with the line feed character.)
-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Thursday, April 21, 2005 9:06 AM
To: Garris, Nicole
Cc: 'pgsql-admin@postgresql.org'
Subject: pg_dump not in object dependency order for sequences used for pri
mary keys
"Garris, Nicole" <Nicole.Garris@dof.ca.gov> writes:
> I am running PostgreSQL 8.0.1 on Max OS X. Rather than using the serial
data
> type for a table's primary key, we use integer and default it to the
nexval
> of a sequence with the name of <tablename>_seq (this syntax seems to work
> best with Apple's WebObjects Enterprise Object Modeler). Therefore, the
> table depends on the sequence, and in a reload of the database, the
sequence
> must be created first. However, pg_dump lists the "create sequence" after
> the "create table". Of course I cannot use the pg_dump file as-is to
reload
> the database, it fails on the "create table" statement when referencing
the
> (as-yet) non-existent sequence.
How so? The nextval only references the sequence as a string; it's not
going to trigger any failure at create time. Please show a specific
test case instead of assuming you know what the problem is.
regards, tom lane