Message-ID: <97d29fa4f3738ce6869dc30fd58f2a3136abb2e2.camel@cybertec.at>
Subject: Re: Feature Proposal: Column-Level DELETE Operation in SQL
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: Abhishek Hatgine <hatgineabhishek99@gmail.com>, 
	pgsql-general@postgresql.org
Date: Tue, 22 Apr 2025 15:39:58 +0200
In-Reply-To: <CAD2voTQMUmNCKBok-_HvYLqFNECRWNkvqPruq0NcsFgd6g-Odw@mail.gmail.com>
References: 
	<CAD2voTQMUmNCKBok-_HvYLqFNECRWNkvqPruq0NcsFgd6g-Odw@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.54.3 (3.54.3-1.fc41) 
MIME-Version: 1.0
Archived-At: <https://www.postgresql.org/message-id/97d29fa4f3738ce6869dc30fd58f2a3136abb2e2.camel%40cybertec.at>
Precedence: bulk

On Mon, 2025-04-21 at 22:23 +0530, Abhishek Hatgine wrote:
> I'd like to propose a new feature for consideration in future versions of=
 SQL =E2=80=94 the ability
> to perform a column-level DELETE operation, allowing removal of specific =
column values
> without affecting the entire row.

In PostgreSQL, it affects the entire row anyway...

> Proposal Summary
> Currently, SQL provides two core commands:
> =C2=A0*=20
> =C2=A0=C2=A0=C2=A0DELETE =E2=80=93 to remove entire rows.
> =C2=A0*=20
> =C2=A0=C2=A0=C2=A0UPDATE =E2=80=93 to change or nullify column values.
>
> However, there=E2=80=99s no specific, expressive way to delete the value =
of a column directly.
> The typical workaround is to use:
>
> UPDATE Customers SET Address =3D NULL WHERE CustomerID =3D 103;
>
> While this works fine, it doesn't semantically express that the developer=
 intends to
> remove the value =E2=80=94 not just update it.
>
> Proposed Syntax Examples
> Here are some ideas for possible new syntax:
>=20
> DELETE Address FROM Customers WHERE CustomerID =3D 103;
> -- or
> REMOVE COLUMN Address FROM Customers WHERE CustomerID =3D 103;
> And even:
> DELETE Address, PostalCode FROM Customers WHERE Country =3D 'India';
>
> These would act as a shortcut or expressive alias for setting one or more=
 column values
> to NULL.
>
> Why This Matters
> =C2=A0* Improved readability and code clarity.
> =C2=A0* More intuitive for developers coming from languages or NoSQL syst=
ems where fields can be
>    "deleted" from an object/document.
> =C2=A0* Emphasizes intent: deleting a value is conceptually different fro=
m updating it to NULL.
> =C2=A0* Opens doors for potential enhancements in tooling and IDE support=
.
> I understand this would require careful consideration within the SQL stan=
dards, but I
> believe it could make SQL more expressive and beginner-friendly while pre=
serving its power.

My immediate gut reaction is "no, thank you".
For anybody who knows SQL, the code would become *less* readable.

But I'd say that PostgreSQL is the wrong place to propose this change.  We =
are unlikely
to implement SQL syntax that deviates from the standard like that.
You should try to convince the SQL standard committee to accept that new sy=
ntax, then
we'd feel more motivated to implement it.

Apart from my strong gut reaction, I have some techical problems with your =
proposed
syntax:

- If you REMOVE or DELETE a column in a row, will it always become NULL or =
should it
  become the DEFAULT value?  With an UPDATE, that is clear: either you say
  "SET col =3D NULL" or "SET col =3D DEFAULT".  So the UPDATE syntax is act=
ually clearer.

- Also, the proposed syntax could easily be confused with "ALTER tab DROP c=
ol", which
  actually removes the column from a table.  SQL users would be confused by=
 your syntax,
  because they would expect that if you REVOVE or DELETE a column, it would=
 no longer
  be there.

Yours,
Laurenz Albe