From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | "D'Arcy J(dot)M(dot) Cain" <darcy(at)druid(dot)net> |
Cc: | pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: pg_dump: could not find namespace with OID |
Date: | 2003-03-09 17:50:49 |
Message-ID: | 19910.1047232249@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
"D'Arcy J.M. Cain" <darcy(at)druid(dot)net> writes:
> Somehow I created two rows for each type in the pg_opclass table, one
> correct and one with NULL entries in opcnamespace and opcowner.
> Should those fields be NOT NULL to protect against this?
They are. But 7.3 has a bug that keeps it from enforcing NOT NULL on
certain system catalogs. Did you make the bogus row by means of a
manual INSERT into pg_opclass? If so, you're just seeing that bug.
If you were able to cause CREATE OPERATOR CLASS to make such a broken
entry, I'd be real interested to see how.
regards, tom lane
PS: you'd best take another look at the RESTRICT and JOIN entries for
your operators...
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2003-03-09 18:25:50 | Re: regression failure in CVS HEAD |
Previous Message | Neil Conway | 2003-03-09 17:38:02 | Re: regression failure in CVS HEAD |