Re: pgbench - allow to create partitioned tables

On Thu, Sep 19, 2019 at 10:25 AM Fabien COELHO <coelho(at)cri(dot)ensmp(dot)fr> wrote:
> Hello Amit,
>
> >> Yes, on -i it will fail because the syntax will not be recognized.
> >
> > Maybe we should be checking the server version, which would allow to
> > produce more useful error messages when these options are used against
> > older servers, like
> >
> > if (sversion < 10000)
> > fprintf(stderr, "cannot use --partitions/--partitions-method
> > against servers older than 10");
> >
> > We would also have to check that partition-method=hash is not used against v10.
> >
> > Maybe overkill?
>
> Yes, I think so: the error detection and messages would be more or less
> replicated from the server and would vary from version to version.
>

Yeah, but I think Amit L's point is worth considering. I think it
would be good if a few other people can also share their suggestion on
this point. Alvaro, Dilip, anybody else following this thread, would
like to comment? It is important to know others opinion on this
because this will change how pgbench behaves with prior versions.

> I do not think that it is worth going this path because the use case is
> virtually void as people in 99.9% of cases would use a pgbench matching
> the server version.

Fair enough, but there is no restriction of using it with prior
versions. In fact some people might want to use this with v11 where
partitioning was present. So, we shouldn't ignore this point.

> One thing we could eventually do is just to check pgbench version against
> the server version like psql does and output a generic warning if they
> differ, but franckly I do not think it is worth the effort:
>

Yeah and even if we want to do something like that, it should not be
part of this patch.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com