Message-Id: <3.0.1.32.19990630115109.00ba55d0@ori-ind.com>
Date: Wed, 30 Jun 1999 11:51:09 -0700
To: pgsql-interfaces@postgreSQL.org
From: Byron Nikolaidis <byronn@insightdist.com> (by way of "Ken J. Wright"
	<ken@ori-ind.com>)
Subject: Re: [INTERFACES] Use Declare/Fetch option
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

An old posting for JT...


Hiroshi Inoue wrote:

> Hello all,
>
> I have a question about psqlodbc driver.
> Now the default setting for "Use Declare/Fetch" is false.
>
> In MVCC thanks to Vadim,readers block no writers.
> So we can change the [Use Declare/Fetch] option to true ?
> And if so,it's effective ?
>

All you say is correct.   It probably could go back to true.  But I
think some people would probably prefer it stay false, because there is
more odbc functionality provided with this setting.  For example, when
the option is false, the entire SQLExtendedFetch functionality is
supported (prev,first,last,next,absolute,relative,bookmark).  When the
Declare/Fetch option is true, only fetch "next" is supported.  There are
more things like this.

If you have extremely large result sets, then setting the option to true
would prevent the driver from sucking down all your memory.   And when
the MVCC is available, the problem of locking rows won't be a problem
anymore.  For small to medium result sets, setting the option to false
will probably result in faster performance.

A future enhancement I have in mind for the driver is using both memory
and files for caching large result sets and/or using a combination of
both kinds of tuple access.

Byron


> Thanks.
>
> Hiroshi Inoue
> Inoue@tpf.co.jp