Message-ID: <10260.125.24.233.43.1179814225.squirrel@webmail.xs4all.nl>
In-Reply-To: <4652133D.2080103@rogers.com>
References: <4651216B.2060707@rogers.com>
	<20070521055759.848496006EA@longblack.object-craft.com.au>
	<4652133D.2080103@rogers.com>
Date: Tue, 22 May 2007 13:10:25 +0700 (ICT)
Subject: Re: PQgetvalue failed to return column value for
	non-text data in binary format
From: "Jeroen T. Vermeulen" <jtv@xs4all.nl>
To: "Jeff Lynn" <jmlynn@rogers.com>
Cc: "Andrew McNamara" <andrewm@object-craft.com.au>,
	pgsql-interfaces@postgresql.org
User-Agent: SquirrelMail/1.4.8
MIME-Version: 1.0
Content-Type: text/plain;charset=iso-8859-1
Content-Transfer-Encoding: 8bit

On Tue, May 22, 2007 04:46, Jeff Lynn wrote:

> This is interesting position to be taken by PostgreSQL.  For simple
> query, particularly with regards to reporting, one can go about using
> text for all column.  But a lot of time, I deal with large dataset
> ingestions and extracts that tends to take many hours to perform.  So I
> usually prefer binary form data for performance reason.  To that end, I
> found there is a significant different in the time it takes to ingest or
> extract data from one database to another database of totally different
> vendor.  Most of our company's works are done in Java nowadays, but for
> large data ingestions/extract/conversion, a C program out performs Java
> program fair and square, no debate there (based on actual run-time
> data).  For clarification, we deal with various business partners that
> we have to import/export large amount of data between heterogeneous
> databases that the wonderful technology of database replication does not
> apply for our use.

For that kind of thing, may I suggest you try using the COPY API?  That
may buy you more speed than transferring data in binary form
would—particularly for insertions.


Jeroen