Hi
Thanks. The unique key with nulls does indeed work. Apparently postgresql
treats every null value as a separate unique value, great.
The second problem though is more persistant. I have created a user
'tst1003' with userid 1003 in my linux system. After that I used 'create
user tst1003;' in postgresql. But I the same error pops up if I type \dt or
\di
Do you have any other suggestions?
-----Original Message-----
From: Vadim Mikheev <vadim@krs.ru>
To: Wim Ceulemans <wim.ceulemans@pophost.eunet.be>
Cc: pgsql-general@postgreSQL.org <pgsql-general@postgreSQL.org>
Date: Friday, August 14, 1998 4:14 AM
Subject: Re: [GENERAL] Unique key with nulls and copying databases across
computers
>Wim Ceulemans wrote:
>>
>> Hi
>>
>> I have a few questions:
>>
>> 1. Is there a simple way in postgresql 6.3.2 to implement a unique index
on
>> one
>> field with null's allowed. So if the field is null it's ok else the value
>> must be unique?
>
>This is exactly what postgresql does!
>Did you see any problems?
>
>>
>> 2. I have copied the contents of a database ( ../data/base/name/* ) from
one
>> computer to another computer. Is this in fact the way to copy databases?
>> After copying everything works except for example \dt , \di which gives
the
>> following
>> error:
>>
>> ERROR: cache lookup for userid 1003 failed
>>
>> What does this mean?
>
>Create user (createuser) with id 1003 in new database.
>
>Vadim