Re: pg_class -> reltuples?

From: Neil Conway <nconway(at)klamath(dot)dyndns(dot)org>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: pg_class -> reltuples?
Date: 2002-03-08 00:44:55
Message-ID: 1015548295.19014.31.camel@jiro
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Thu, 2002-03-07 at 17:51, Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > Neil Conway wrote:
> >> Is there a reason why the reltuples column of pg_class is stored as a
> >> "real", rather than one of the integer data types?
>
> > That is an excellent question. I assume it is related to having > 4
> > billion rows, but we have int8 for that.
>
> 1. We support tables > 4G rows.

I agree we should try to support very large tables -- so why waste space
on storing floating point? And am I missing something, or is a "real"
only 4 bytes?

> 2. int8 is not available on all platforms.

I have no problem making restrictions on data types for portability, but
at least we should be consistent:

% grep -rI 'long long' * | wc -l
37
% grep -rI 'int64' * | wc -l
191

On all the platforms I tested (x86, SPARC, PPC, PA-RISC, Alpha), a 'long
long' is supported, and is 8 bytes. Which platforms don't have this, and
are we actively supporting them?

Cheers,

Neil

--
Neil Conway <neilconway(at)rogers(dot)com>
PGP Key ID: DB3C29FC

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2002-03-08 00:47:54 Re: Bad Build
Previous Message Bruce Momjian 2002-03-07 23:28:27 Re: Bad Build