> In the constraints test, a table (primary_tbl) is defined with a
> primary key. Any help would be appreciated.
Uh, yeah. This isn't a "primary key" problem, but is the same symptom we
have been seeing on Linux boxes. Bruce's FreeBSD box does not exhibit
the problem.
It seems to involve a trashed index on the pg_class table. Do you also
see a failure in the create_index regression test? afaik the problem
trying to add a second index to the "onek" table is also index-related.
Check the mhonarc archive for the full history, but we are still trying
to get a handle on it and would welcome any help.
There also is a problem with views/rules on the Linux boxes (at least,
some problems clearly span multiple platforms but no problems appear on
all platforms); do you see a failure when doing "select * from pg_user"
on a clean installation?
- Tom