Re: pgbench - allow to create partitioned tables

On Thu, Oct 3, 2019 at 1:53 PM Fabien COELHO <coelho(at)cri(dot)ensmp(dot)fr> wrote:
>
>
> Hello,
>
> > As partitions is an integer type variable, the maximum value it can
> > hold is "2147483647". But if I specify partitions as "3147483647",
> > atoi function returns a value lesser than zero and pgbench terminates
> > with an error. However, if the value for number of partitions
> > specified is something like "5147483647", atoi returns a non-negative
> > number and pgbench creates as many number of partitions as the value
> > returned by atoi function.
> >
> > This seems like a problem with atoi function, isn't it?
>
> Yes.
>
> > atoi functions has been used at several places in pgbench script and I
> > can see similar behaviour for all. For e.g. it has been used with
> > scale factor and above observation is true for that as well. So, is
> > this a bug or you guys feel that it isn't and can be ignored? Please
> > let me know your thoughts on this. Thank you.
>
> I think that it is a known bug (as you noted atoi is used more or less
> everywhere in pgbench and other commands) which shoud be addressed
> separately: all integer user inputs should be validated for syntax and
> overflow, everywhere, really. This is not currently the case, so I simply
> replicated the current bad practice when developing this feature.
>

Okay, I think we should possibly replace atoi with strtol function
call for better error handling. It handles the erroneous inputs better
than atoi.

> There is/was a current patch/discussion to improve integer parsing, which
> could address this.
>

It seems like you are trying to point out the following discussion on hackers,

https://www.postgresql.org/message-id/flat/20190724040237.GB64205%40begriffs.com#5677c361d3863518b0db5d5baae72bbe

--
With Regards,
Ashutosh Sharma
EnterpriseDB:http://www.enterprisedb.com