MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Date: Fri, 25 Sep 2020 14:46:00 +0900
From: Katsuragi Yuta <btkatsuragiyu@oss.nttdata.com>
To: legrand legrand <legrand_legrand@hotmail.com>
Cc: pgsql-hackers@postgresql.org
Subject: Re: [PATCH] Add features to pg_stat_statements
In-Reply-To: <1600941306200-0.post@n3.nabble.com>
References: <0d9f1107772cf5c3f954e985464c7298@oss.nttdata.com>
 <1600500942767-0.post@n3.nabble.com>
 <16a262a87492b4f4afc4c5010f8fa631@oss.nttdata.com>
 <1600924523318-0.post@n3.nabble.com>
 <da43a8a55f92622569b9ce4c57a0df41@oss.nttdata.com>
 <1600941306200-0.post@n3.nabble.com>
Message-ID: <dcb63417e592d5966c19c04f90db0aef@oss.nttdata.com>
User-Agent: Roundcube Webmail/1.1.9
Precedence: bulk

On 2020-09-24 18:55, legrand legrand wrote:
> Not limited to 3, Like an history table.
> 
> Will have to think if data is saved at shutdown.

Thanks.

This design might introduce some additional complexity
and things to be considered.
For example, the maximum size of "history table",
how to evict entries from the history table and how to manage
the information maintained by evicted entries.

Also, providing a history table looks similar to logging.

Providing the original view (# of dealloc and last_dealloc ts)
and optional logging looks a simple and better way.

Regards,
Katsuragi Yuta