Re: pg_dump: CREATE TABLE + CREATE RULE vs. relreplident

From: Andres Freund <andres(at)2ndquadrant(dot)com>
To: Marko Tiikkaja <marko(at)joh(dot)to>
Cc: Pgsql Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: pg_dump: CREATE TABLE + CREATE RULE vs. relreplident
Date: 2015-03-12 14:14:48
Message-ID: 20150312141448.GC20199@awork2.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On 2015-03-12 14:25:24 +0100, Marko Tiikkaja wrote:
> My colleague Per Lejontand brought to my attention that when dumping views
> with circular dependencies from a postgres version older than 9.4 using a
> recent pg_dump, the SQL looks something like the following:
>
> create table qwr();
> create rule "_RETURN" as on select to qwr do instead select;
>
> In this case the relreplident column in pg_class for the view ends up being
> 'd', instead of the 'n' normally used for views. Patch to update
> relreplident when turning a table into a view is attached; this makes sure
> that the identity is NOTHING regardless of how the view was created.

I think that's a good idea.

> I consider this a bug fix, and suggest back patching to 9.4.

I agree on backpatching it. Arguably we could additionally avoid
emitting the ALTER TABLE ... REPLICA IDENTITY for views that have
already been created with identity set like this. But I doubt it's worth
it.

Greetings,

Andres Freund

--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2015-03-12 14:15:12 Re: logical column ordering
Previous Message Tom Lane 2015-03-12 14:08:54 Re: pg_dump: CREATE TABLE + CREATE RULE vs. relreplident