Re: Support ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION ... syntax - Mailing list pgsql-hackers

From Amit Kapila
Subject Re: Support ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION ... syntax
Date
Msg-id CAA4eK1JV4S8b6vvHPQWy965oPnkGU5jhy7RO2PwHRrGjc5q=jA@mail.gmail.com
Whole thread Raw
In response to Support ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION ... syntax  (japin <japinli@hotmail.com>)
Responses Re: Support ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION ... syntax
Re: Support ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION ... syntax
List pgsql-hackers
On Tue, Jan 26, 2021 at 9:18 AM japin <japinli@hotmail.com> wrote:
>
>
> When I read the discussion in [1], I found that update subscription's publications
> is complicated.
>
> For example, I have 5 publications in subscription.
>
>     CREATE SUBSCRIPTION mysub1 CONNECTION 'host=localhost port=5432 dbname=postgres'
>     PUBLICATION mypub1, mypub2, mypub3, mypub4, mypub5;
>
> If I want to drop "mypub4", we should use the following command:
>
>     ALTER SUBSCRIPTION mysub1 SET PUBLICATION mypub1, mypub2, mypub3, mypub5;
>
> Also, if I want to add "mypub7" and "mypub8", it will use:
>
>     ALTER SUBSCRIPTION mysub1 SET PUBLICATION mypub1, mypub2, mypub3, mypub5, mypub7, mypub8;
>
> Attached implement ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION ... syntax, for the above
> two cases, we can use the following:
>
>     ALTER SUBSCRIPTION mysub1 DROP PUBLICATION mypub4;
>
>     ALTER SUBSCRIPTION mysub1 DROP PUBLICATION mypub7, mypub8;
>
> I think it's more convenient. Any thoughts?
>

While the new proposed syntax does seem to provide some ease for users
but it has nothing which we can't do with current syntax. Also, in the
current syntax, there is an additional provision for refreshing the
existing publications as well. So, if the user has to change the
existing subscription such that it has to (a) add new publication(s),
(b) remove some publication(s), (c) refresh existing publication(s)
then all can be done in one command whereas with your new proposed
syntax user has to write three separate commands.

Having said that, I don't deny the appeal of having separate commands
for each of (a), (b), and (c).

-- 
With Regards,
Amit Kapila.



pgsql-hackers by date:

Previous
From: Dilip Kumar
Date:
Subject: Re: Is Recovery actually paused?
Next
From: "kuroda.hayato@fujitsu.com"
Date:
Subject: RE: ECPG: proposal for new DECLARE STATEMENT