Re: BUG #3913: pg_dump -T option not recognised

OK, just reproduced with 8.2.6 compiled from source on Ubuntu 7.10 (all
updates applied as a couple of hours ago).

Steps to reproduce.  Forgive any typos, this isn't all from bash_history,
but was done on a machine that never had postgresql on it before
(specifically, my laptop).

Command output is shown between ... characters.

** Basic install and clean DB setup:

wget http://mirror/postgresql-8.2.6.tar.gz
tar zxvf postgres-8.2.6.tar.gz
cd postgresql-8.2.6
./configure --prefix=/opt/postgres
make
sudo make install
sudo lndir /opt/postgres/bin /usr/local/bin
sudo mkdir /opt/postgres/data
sudo chown postgres /opt/postgres/data
sudo adduser postgres
sudo su - postgres
initdb -D /opt/postgres/data
postgres -D /opt/postgres/data > logfile 2>&1 &
createdb test
psql test

** Create test tables:

create table a ( col1 date, col2 date );
create table b ( col1 date, col2 date );
create table c ( col1 date, col2 date );

exit psql

** Sanity check psql:

pg_dump -v test > /dev/null

...
pg_dump: creating TABLE a
pg_dump: creating TABLE b
pg_dump: creating TABLE c
...

As expected

** Check --exclude-table option:

pg_dump -v --exclude-table=b test > /dev/null

...
pg_dump: creating TABLE a
pg_dump: creating TABLE c
...

As expected

** Check -T option:

pg_dump -v -T=b test > /dev/null

...
pg_dump: creating TABLE a
pg_dump: creating TABLE b
pg_dump: creating TABLE c
...

Silently ignores -T


I hope I'm an idiot, but that looks like a bug according to pg_dump --help
which says:

...
-T, --exclude-table=TABLE   do NOT dump the named table(s)
...


Hope that helps you reproduce it.

Cheers,

David



On 30/01/2008, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Zdenek Kotala <Zdenek.Kotala@Sun.COM> writes:
> > David Dunwoody wrote:
> >> When running pg_dump, the -T option for table exclusion is ignored, but
> >> --exclude-table does work.
>
> > It works for me. Can you recheck it again?
>
> Works for me too, on Fedora 8.  But could you give a complete example?
> Maybe there's some other contributing factor in the particular case
> you're trying.
>
>                         regards, tom lane
>