Re: pgbench - allow to create partitioned tables - Mailing list pgsql-hackers

From Alvaro Herrera
Subject Re: pgbench - allow to create partitioned tables
Date
Msg-id 20190913125502.GA7281@alvherre.pgsql
Whole thread Raw
In response to Re: pgbench - allow to create partitioned tables  (Amit Kapila <amit.kapila16@gmail.com>)
List pgsql-hackers
On 2019-Sep-13, Amit Kapila wrote:

> On Fri, Sep 13, 2019 at 1:50 PM Fabien COELHO <coelho@cri.ensmp.fr> wrote:
> > >>> Is there a reason why we treat "partitions = 0" as a valid value?
> > >>
> > >> Yes. It is an explicit "do not create partitioned tables", which differ
> > >> from 1 which says "create a partitionned table with just one partition".
> > >
> > > Why would anyone want to use --partitions option in the first case
> > > ("do not create partitioned tables")?
> >
> > Having an explicit value for the default is generally a good idea, eg for
> > a script to tests various partitioning settings:
> >
> >    for nparts in 0 1 2 3 4 5 6 7 8 9 ; do
> >      pgbench -i --partitions=$nparts ... ;
> >      ...
> >    done
> >
> > Otherwise you would need significant kludging to add/remove the option.
> > Allowing 0 does not harm anyone.
> >
> > Now if the consensus is to remove an explicit 0, it is simple enough to
> > change it, but my opinion is that it is better to have it.
> 
> Fair enough, let us see if anyone else wants to weigh in.

It seems convenient UI -- I vote to keep it.

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services



pgsql-hackers by date:

Previous
From: Alvaro Herrera
Date:
Subject: Re: [PATCH] Speedup truncates of relation forks
Next
From: Fujii Masao
Date:
Subject: Re: [PATCH] Speedup truncates of relation forks