Date: Mon, 1 Oct 2018 00:34:23 -0300
From: Alvaro Herrera <alvherre@2ndquadrant.com>
To: Michael Paquier <michael@paquier.xyz>
Cc: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>,
	"Yotsunaga, Naoki" <yotsunaga.naoki@jp.fujitsu.com>,
	'Postgres hackers' <pgsql-hackers@postgresql.org>
Subject: Re: automatic restore point
Message-ID: <20181001033423.xtc7jckr74good2n@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20180930124510.GB1763@paquier.xyz>
User-Agent: NeoMutt/20180716
Precedence: bulk

On 2018-Sep-30, Michael Paquier wrote:

> On Fri, Sep 28, 2018 at 09:13:17PM +0200, Peter Eisentraut wrote:
> > I think this should be done using event triggers.  Right now, you just
> > have it hardcoded to TRUNCATE and DROP TABLE, which seems somewhat
> > arbitrary.  With event triggers, you have the full flexibility to do
> > what you want.  You can pick which commands to apply it to, you can log
> > the LSN, you can create restore points, etc.
> 
> I still unfortunately don't see what this patch brings more that you
> cannot do.  Event triggers are particularly useful in this prospective,
> so I am marking the patch as rejected.

I don't see it as clear cut as all that ... particularly considering
that a useful event trigger runs *after* the DDL command in question has
already written all its WAL, so such a restore point would be completely
useless.  (Or are ddl_command_start event triggers useful enough?)

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services