Re: BUG #7792: pg_dump does not treat -c flag correctly when using tar format

On Mon, Jan 7, 2013 at 1:32 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> emesika(at)redhat(dot)com writes:
>> 1) pg_dump -F t -U postgres -f test.tar test
>> 2) tar xvf test.tar to any directory
>> 3) vi restore.sql
>> * restore.sql includes DROP statements for each object even tough -c flag
>> was not given
>
> I believe this is intentional - at least, pg_backup_tar.c goes out of
> its way to make it happen. (The forcible setting of ropt->dropSchema
> in _CloseArchive is the cause, and it's hard to see why that would be
> there unless the author intended this effect.) Perhaps we should remove
> that, but it would be an incompatible change. Arguing for or against
> it really requires a model of what people would be doing with the
> restore.sql script. I'm not entirely convinced that it should be
> considered equivalent to what you'd get from a plain dump run.

tar archives are more like custom archives ,arne't they? The idea
being that since you use pg_restore to restore them, you should use -c
on the pg_restore commandline, and not on the pg_dump one? And if we
didn't include it in the dump, that wouldn't even be possible.

I definitely think it's doing the right thing, but perhaps it's
something that needs to be clarified in the documentation if people
find it confusing.

--
Magnus Hagander
Me: http://www.hagander.net/
Work: http://www.redpill-linpro.com/