Content-Type: text/plain;
  charset="iso-8859-1"
From: "D'Arcy J.M. Cain" <darcy@druid.net>
Reply-To: pgsql-interfaces@PostgreSQL.org
To: Joel Mc Graw <jmcgraw@databill.com>
Subject: Re: Python interface and Money?
Date: Wed, 26 Feb 2003 08:53:18 -0500
User-Agent: KMail/1.4.3
References: <3BD45B33.C217E6FF@databill.com>
In-Reply-To: <3BD45B33.C217E6FF@databill.com>
Cc: pgsql-interfaces@PostgreSQL.org
MIME-Version: 1.0
Message-Id: <200302260853.18997.darcy@druid.net>
Content-Transfer-Encoding: quoted-printable

On Monday 22 October 2001 13:45, Joel Mc Graw wrote:
> I have a table that has a column of type money.  A query such as "select
> sum(document_amount) from foo" returns the correct result when executed
> from psql, yet the same query returns the wrong answer (it appears to be
> a rounding issue--something like "1234.9999999999...") when executed
> through PygreSQL.  Is there a workaround or a fix?

The problem is that Python doesn't (yet) have a native decimal type and so =
the=20
closest thing we can convert to is float.  There have been attempts at a=20
decimal type and at some point I will convert to one of them, hopefully the=
=20
same one that Python chooses as a standard.

> While browsing the mailing list for an answer to this problem I found
> several references to the money type being deprecated.  Is that the
> problem?

There are people who want to remove it but I still think it serves a useful=
=20
purpose.  I think that it needs to move to a 64 bit int and handle a lot mo=
re=20
type conversions but I moved some stuff to numeric due to the limits and=20
found a noticeable slowdown when working with lots of aggregates on it.  Th=
e=20
int type is just more efficient.  I plan on making a contrib version of the=
=20
type before it gets deleted if that is what is decided.

--=20
D'Arcy J.M. Cain <darcy@{druid|vex}.net>   |  Democracy is three wolves
http://www.druid.net/darcy/                |  and a sheep voting on
+1 416 425 1212     (DoD#0082)    (eNTP)   |  what's for dinner.