Re: pg_dumpall behavior in 9.1beta1

oops! sorry for the noise

it was an innocuous 'NOTICE' not ERROR as claimed.
thanks for your time.

regds
mallah.

On Fri, May 13, 2011 at 7:59 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Rajesh Kumar Mallah <mallah(dot)rajesh(at)gmail(dot)com> writes:
>> We migrated to 9.1beta1 from 9.0.x in our development environment.
>> we shall report if any significant problems or issue arises out of this.
>
>> one small thing that we observed rite now regarding pg_dumpall output is that
>
>> the ALTER USER is referring to some db objects which are not created yet
>> and this raises an ERROR this may cause some inconvenience where restoration
>> is being carried out in a txn .
>
>> example of relevant sql lines is given below
>
>> CREATE ROLE pdns;
>> ALTER ROLE pdns WITH NOSUPERUSER INHERIT NOCREATEROLE NOCREATEDB LOGIN
>> NOREPLICATION PASSWORD 'md56dxxxxxxxxxxxxxxxxxxxxxxxxx' ;
>> ALTER ROLE pdns SET search_path TO pdns;
>
>> in above line the 'pdns' namespace does not exists yet.
>
> When I do that, I get a harmless NOTICE, not an ERROR.  Are you sure you
> are getting an error?
>
> regression=# CREATE ROLE pdns;
> CREATE ROLE
> regression=# ALTER ROLE pdns WITH NOSUPERUSER INHERIT NOCREATEROLE NOCREATEDB LOGIN
> regression-# NOREPLICATION PASSWORD 'md56dxxxxxxxxxxxxxxxxxxxxxxxxx' ;
> ALTER ROLE
> regression=# ALTER ROLE pdns SET search_path TO pdns;
> NOTICE:  schema "pdns" does not exist
> ALTER ROLE
> regression=#
>
>                        regards, tom lane
>