To: pgsql-odbc@postgresql.org
From: Marc Herbert <Marc.Herbert@continuent.com>
Subject: Re: BLOB handling compatibility with PostgreSQL > 7.4
Date: Fri, 09 Dec 2005 11:46:01 +0100
Lines: 38
Message-ID: <87zmnah7km.fsf@meije.emic.fr>
References: <438B792D.6030802@bnl.gov> <200512081055.12263.peter_e@gmx.net>
	<871x0nr8p0.fsf@meije.emic.fr> <200512081911.30002.peter_e@gmx.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4 (Jumbo Shrimp, linux)
Cancel-Lock: sha1:ycJJyhn0OqwXCWx7kr5eS7FMluY=

Peter Eisentraut <peter_e@gmx.net> writes:

> Marc Herbert wrote:
>> We are developping a middleware and are not free to choose our data
>> types.
>
> I was thinking the ODBC driver should provide the appropriate mapping.

We aim to be as transparent as possible and avoid "mappings".

 <http://www.continuent.org/>


>> As far as i know, the implementation of bytea and large objects in
>> the database are totally different and I suspect there are good
>> reasons to use either one of them, depending on the actual data and
>> application involved.
>
> There is no real reason to use large objects for anything instead of 
> bytea.

<http://www.postgresql.org/docs/7.4/static/jdbc-binary-data.html>
<http://jdbc.postgresql.org/documentation/head/binary-data.html>

 PostgreSQL provides two distinct ways to store binary data. Binary
 data can be stored in a table using the data type bytea or by using
 the Large Object feature which stores the binary data in a separate
 table in a special format and refers to that table by storing a value
 of type oid in your table.
 In order to determine which method is appropriate you need to
 understand the limitations of each method....


Moreover, the respective interfaces have very different semantics
(pointer or not). This can be a reason in itself to prefer one or the
other type, whatever the underlying implementation is.