Horton

Re: Restore replication settings when modifying a field type

Peter Eisentraut <peter.eisentraut@2ndquadrant.com>

View thread View on postgr.es
From: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
To: Euler Taveira <euler.taveira@2ndquadrant.com>
Cc: Quan Zongliang <quanzongliang@foxmail.com>, Kyotaro Horiguchi <horikyota.ntt@gmail.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2020-03-13T12:32:27Z
Lists: pgsql-hackers 
On 2020-03-10 14:16, Euler Taveira wrote:
> On Thu, 5 Mar 2020 at 09:45, Peter Eisentraut 
> <peter.eisentraut@2ndquadrant.com 
> <mailto:peter.eisentraut@2ndquadrant.com>> wrote:
> 
>     On 2020-02-11 00:38, Quan Zongliang wrote:
>      > new patch attached.
> 
>     I didn't like so much how the updating of the replica identity was
>     hacked into the middle of ATRewriteCatalogs().  I have an alternative
>     proposal in the attached patch that queues up an ALTER TABLE ...
>     REPLICA
>     IDENTITY command into the normal ALTER TABLE processing.  I have also
>     added tests to the test suite.
> 
> LGTM. Tests are ok. I've rebased it (because 
> 61d7c7bce3686ec02bd64abac742dd35ed9b9b01). Are you planning to backpatch 
> it? IMHO you should because it is a bug (since REPLICA IDENTITY was 
> introduced in 9.4). This patch can be applied as-is in 12 but not to 
> other older branches. I attached new patches.

Thanks.  This has been committed and backpatched to 9.5.

-- 
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Preserve replica identity index across ALTER TABLE rewrite

  2. Prevent reindex of invalid indexes on TOAST tables