Date: Thu, 16 Mar 2023 16:11:59 -0500
From: Justin Pryzby <pryzby@telsasoft.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Robert Treat <rob@xzilla.net>, Robert Haas <robertmhaas@gmail.com>,
	Andrew Dunstan <andrew@dunslane.net>,
	pgsql-hackers@lists.postgresql.org,
	David Rowley <drowley@postgresql.org>,
	Alvaro Herrera <alvherre@alvh.no-ip.org>
Subject: Re: doc: mentioned CREATE+ATTACH PARTITION as an alternative to
 CREATE TABLE..PARTITION OF
Message-ID: <ZBOGH1qUZmkVA9H7@telsasoft.com>
References: <20220718143304.GC18011@telsasoft.com>
 <1e91af94-2bea-5b35-49f3-ea14859c2a0d@dunslane.net>
 <CABV9wwP1s9Zh+a8XQqH4YzGMdF34q6a3fPNGucOuJ8GWVT3ahw@mail.gmail.com>
 <20220905180403.GN31833@telsasoft.com>
 <CAJSLCQ0JMu2CTa8ka5yV5v3jjFrCT4XtFq3M-jy7q=v0-2t08Q@mail.gmail.com>
 <CA+TgmoZxY_YpqCmNBYgQJTWLG4XF_0QjLwJ61W0_XK-_eig9vg@mail.gmail.com>
 <CAJSLCQ2oSki4nAKUmE2EybkBAWGNRtdwbFunsv3MD_fZrj9bWg@mail.gmail.com>
 <20230119215832.GH13860@telsasoft.com>
 <3975824.1678999927@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <3975824.1678999927@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/ZBOGH1qUZmkVA9H7%40telsasoft.com>
Precedence: bulk

On Thu, Mar 16, 2023 at 04:52:07PM -0400, Tom Lane wrote:
> Justin Pryzby <pryzby@telsasoft.com> writes:
> > On Thu, Jan 19, 2023 at 04:47:59PM -0500, Robert Treat wrote:
> >> I think all of that feedback is useful, I guess the immediate question
> >> becomes if Justin wants to try to proceed with his patch implementing
> >> the change, or if adjusting the documentation for the current
> >> implementation is the right move for now.
> 
> > The docs change is desirable in any case, since it should be
> > backpatched, and any patch to change CREATE..PARTITION OF would be for
> > v17+ anyway.
> 
> Right.  Pushed with a little further effort to align it better with
> surrounding text.

Thanks.

      It is possible to use <link linkend="sql-altertable"><command>ALTER
      TABLE ATTACH/DETACH PARTITION</command></link> to perform these
      operations with a weaker lock, thus reducing interference with
      concurrent operations on the partitioned table.

Note that in order for DETACH+DROP to use a lower lock level, it has to be
DETACH CONCURRENTLY.  ATTACH is implicitly uses a lower lock level, but for
DETACH it's only on request.

-- 
Justin