MIME-Version: 1.0
References: <CAEoWx2n6ShLMOnjOtf63TjjgGbgiTVT5OMsSOFmbjGb6Xue1Bw@mail.gmail.com>
In-Reply-To: <CAEoWx2n6ShLMOnjOtf63TjjgGbgiTVT5OMsSOFmbjGb6Xue1Bw@mail.gmail.com>
From: Robert Treat <rob@xzilla.net>
Date: Thu, 1 Jan 2026 21:54:06 -0500
Message-ID: <CABV9wwMPmo-dg=oV+NaJGa_==uo5+TTvk58aNtjVQZBtMcRVBw@mail.gmail.com>
Subject: Re: DOC: fixes multiple errors in alter table doc
To: Chao Li <li.evan.chao@gmail.com>
Cc: Postgres hackers <pgsql-hackers@lists.postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CABV9wwMPmo-dg%3DoV%2BNaJGa_%3D%3Duo5%2BTTvk58aNtjVQZBtMcRVBw%40mail.gmail.com>
Precedence: bulk

On Thu, Dec 18, 2025 at 2:22=E2=80=AFAM Chao Li <li.evan.chao@gmail.com> wr=
ote:
>
> Hi Hacker,
>
> While working on a patch these days, my eyes are on the =E2=80=9Calter ta=
ble=E2=80=9D doc, and found multiple errors:
>
> 1. Several sub-commands are missed in the top =E2=80=9Caction=E2=80=9D li=
st:
>
>    * ALTER COLUMN SET <sequence-option>
>    * ALTER COLUMN RESTART

I believe these are covered by the line above them:

    ALTER [ COLUMN ] column_name { SET GENERATED { ALWAYS | BY DEFAULT
} | SET sequence_option | RESTART [ [ WITH ] restart ] } [...]

>    * RENAME

This is covered by the 4th line

ALTER TABLE [ IF EXISTS ] name
    RENAME TO new_name

>    * SET SCHEMA

5th option...

ALTER TABLE [ IF EXISTS ] name
    SET SCHEMA new_schema

>    * ATTACH PARTITION
>    * DETACH PARTITION
>    * MERGE PARTITION
>    * SPLIT PARTITION
>

And the above are the 7,8,9,10th options at the top.


> 2. In sub-command details section, "ADD COLUMN [ IF NOT EXISTS ]=E2=80=9D=
 missed =E2=80=9C[]" with =E2=80=9CCOLUMN=E2=80=9D, which is misleading, be=
cause =E2=80=9CCOLUMN=E2=80=9D is actually optional.
>

Seems technically correct and potentially useful, and I see you
handled this for the DROP COLUMN variant as well, so I could see a +1
on this one.

> 3. For all =E2=80=9Calter column=E2=80=9D sub-commands, "ALTER [ COLUMN ]=
=E2=80=9D are omitted, which is also confusing, because none of other sub-c=
ommands omit their prefix part.
>

Hmm... I'm curious what you find confusing about this. Is the
confusion in trying to find or understand the information presented,
or confusing as to why it isn't all documented the same way? The
downside of your "fix" is that this introduces a lot of extra text
that is more or less noise, especially for folks trying to skim the
documents looking for very specific command references.  And while I
agree that we aren't 100% consistent on this within the ALTER TABLE
subcommands, we use this same mixed pattern of omission on other pages
(see ALTER TYPE for instance). If we were to insist on making this
consistent here, I think we'd probably need to look at other pages as
well and evaluate or update them too. I'm not sure that would be an
improvement though.


Robert Treat
https://xzilla.net