Re: Patch to fix search_path defencies with pg_bench

Em Qua, 2009-05-06 às 13:49 -0700, Joshua D. Drake escreveu:
> On Wed, 2009-05-06 at 17:42 -0300, Dickson S. Guedes wrote:
> > Em Qua, 2009-05-06 às 16:27 -0400, Tom Lane escreveu:
> > > Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
> > > > I think it would be better that the schema is specified on the command
> > > > line.
> > >
> > > Surely that's more work than the issue is worth. It's also inconvenient
> > > to use, because you'd have to remember to give the switch both for the
> > > -i run and the normal test runs.
> >
> > So, in my opinion, the Joshua alternative is a good little change that
> > let "pgbench" runs in a more flexible way.
> >
> > But, there is the possibility that someone are using an automated script
> > that could be broken by this change?
>
> Only if the role pgbench is using as an explicit search_path set.

So, in a way to avoid the scenario where a ROLE has an explicit
search_path set to schemes that already have tables named same as the
pgbench's tables, doesn't makes sense also create a "pgbench_" suffix
for them?

--
Dickson S. Guedes
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br