Thread: Fwd: Detecting renamed columns via pgouput in logical replication ?
Hello,
re-posting question to the dev mailing list, since no reply received in the general mailing list. This mailing list also seems a better place for this question. See below for the original mail.
thx
---------- Forwarded message ---------
From: Andreas Andreakis <andreas.andreakis@gmail.com>
Date: Thu, Apr 16, 2020 at 9:22 PM
Subject: Detecting renamed columns via pgouput in logical replication ?
To: <pgsql-general@lists.postgresql.org>
From: Andreas Andreakis <andreas.andreakis@gmail.com>
Date: Thu, Apr 16, 2020 at 9:22 PM
Subject: Detecting renamed columns via pgouput in logical replication ?
To: <pgsql-general@lists.postgresql.org>
Hello,
when using Postgres 10 or higher, it seems that pgoutput can be used as an output plugin for logical replication.
Does this allow to detect column renames ? Or is there a ticket for adding support if the feature does not exist (if it is feasible to implement) ?
https://www.postgresql.org/docs/10/protocol-logicalrep-message-formats.html should be the format spec of pgoutput and it does not seem to contain sufficient information for detecting renames. Also checked for Postgres version 11 and 12.
What I was hoping to find is the ordinal position of columns and use that to infer column renames. As new columns always get a higher ordinal position and renamed columns keep their position. Hence, a column rename could be detected if different column names are received for the same ordinal position, by tracking the column name per ordinal position at the consumer. (Please let me know if any of that is incorrect)
cheers
On Sunday, April 26, 2020, Andreas Andreakis <andreas.andreakis@gmail.com> wrote:
Hello,re-posting question to the dev mailing list, since no reply received in the general mailing list. This mailing list also seems a better place for this question. See below for the original mail.
Actually this is probably the worse possible list to choose. It would be better to simply ping the original post.
David J.
Hi Andreas, On Sun, Apr 26, 2020 at 10:40:08PM -0700, Andreas Andreakis wrote: > re-posting question to the dev mailing list, since no reply received in the > general mailing list. This mailing list also seems a better place for this > question. See below for the original mail. pgsql-committers is used as a mailing list to send automatically posts when a commit to the central repository happens. What you may be looking for here is pgsql-hackers: https://www.postgresql.org/list/pgsql-hackers/ -- Michael