Message-ID: <43D13C2F.1030203@zg.htnet.hr>
Date: Fri, 20 Jan 2006 20:38:23 +0100
From: Rikard Pavelic <rikard.pavelic@zg.htnet.hr>
User-Agent: Thunderbird 1.5 (Windows/20051201)
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: [PERFORMANCE] Stored Procedures
References: <43D130EF.3010205@zg.htnet.hr>
	<20060120191031.GO20182@pervasive.com>
In-Reply-To: <20060120191031.GO20182@pervasive.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Jim C. Nasby wrote:
> My college professor said it, it must be true! ;P
>
>   
The famous joke ;)
> My understanding is that in plpgsql, 'bare' queries get prepared and act
> like prepared statements. IE:
>
> SELECT INTO variable
>     field
>     FROM table
>     WHERE condition = true
> ;
>
>   
Unfortunately I don't know enough about PostgreSQL, but from responses 
I've been reading I've
come to that conclusion.
> Ok, so post some numbers then. It might be interesting to look at the
> cost of preparing a statement, although AFAIK that does not store the
> query plan anywhere.
>
> In most databases, query planning seems to be a pretty expensive
> operation. My experience is that that isn't the case with PostgreSQL.
>   

I didn't think about storing query plan anywhere on the disk, rather 
keep them in memory pool.
It would be great if we had an option to use prepare statement for 
stored procedure so it
would prepare it self the first time it's called and remained prepared 
until server shutdown or
memory pool overflow.

This would solve problems with prepare which is per session, so for 
prepared function to be
optimal one must use same connection.