From: pilsl@goldfisch.at
Date: Mon, 4 Nov 2002 01:11:06 +0100
To: Doug McNaught <doug@mcnaught.org>
Cc: pgsql-general@postgresql.org
Subject: Re: 7.2.3: tuple is too big (max 8136)
Message-ID: <20021104011105.A11131@goldfisch.at>
References: <20021101143348.A7426@goldfisch.at>
	<m37kfx9z5t.fsf@varsoon.wireboard.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <m37kfx9z5t.fsf@varsoon.wireboard.com>
User-Agent: Mutt/1.3.23i

On Fri, Nov 01, 2002 at 08:50:54AM -0500, Doug McNaught wrote:
>
> It's possible you are hitting the maximum column limit rather than the
> tuple size limit.  The FAQ list the max columns in a table as:
> 
> "250-1600 depending on column types"
> 

I already created the table with 1600 cols of varchar(30) so this
should not be the limit. And the errormessage is clearly talking about
the tuples size. (which should not have such a limit in 7.2.3).

> So try a tuple of the same size but fewer columns and see if that's
> the issue.
>

I created smaller tables (200-300 cols of varchar(1000)) and inserted
rows with long text and the error didnt occure. I cant tell if the
size of this tuple is then below the 8k-limit cause I dont know how to
calculate the size of the tuple from a given insert-command.

(The only idea about the tuple-size is when I get the error where the
actual tuple-size is printed. There seems to be no relation between
the length of the supplied command and the tuple-size reported in the
error)

To check the problem it would be very helpful to me to learn more
about tuple-size. It seem like postgres would use compression, cause a
supplied insert-command of length 620000 lead to the same tuplesize of
20200 than a insert-command of length 20000.

> 
> I'm sure you're right but it's always good to double-check.  ;)
>

I did double and triple.

thnx,
peter
 

-- 
mag. peter pilsl
IT-Consulting
tel: +43-699-1-3574035
fax: +43-699-4-3574035
pilsl@goldfisch.at