Message-ID: <f6c98da81f11aca47e6f407a1c76c8265cdb1c2a.camel@cybertec.at>
Subject: Re: Command Line option misunderstanding
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: punch-hassle-guise@duck.com, "pgsql-novice@lists.postgresql.org"
	 <pgsql-novice@lists.postgresql.org>
Date: Tue, 03 Dec 2024 07:12:54 +0100
In-Reply-To: <A65F772B-97BF-4F1A-B97C-2E31B3A31E3C.1@smtp-inbound1.duck.com>
References: <487DB217-EA37-4139-AB97-B61B04ECAEA7.1@smtp-inbound1.duck.com>
	 <cd1068a8f07f98936bc2964dc7c22c109191801b.camel@cybertec.at>
	 <493C622D-D3B6-4662-A617-EBCCDE5AA4DF.1@smtp-inbound1.duck.com>
	 <B1918EE7-CABD-40D3-8002-B3958F145152.1@smtp-inbound1.duck.com>
	 <30948e6771500c0e9d8b587f4e34165aadb1cc0b.camel@cybertec.at>
	 <AEA5E111-EE09-4BF3-9009-6A6B12F9F215.1@smtp-inbound1.duck.com>
	 <A65F772B-97BF-4F1A-B97C-2E31B3A31E3C.1@smtp-inbound1.duck.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.52.4 (3.52.4-2.fc40) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/f6c98da81f11aca47e6f407a1c76c8265cdb1c2a.camel%40cybertec.at>
Precedence: bulk

On Mon, 2024-12-02 at 18:49 -0500, punch-hassle-guise@duck.com wrote:
> The problem seems to be, as alluded to by others attempting to help me=C2=
=A0
> that the problem only exists when using -c on the same line as -v.=C2=A0
> I am guessing here, that my original question should have been:
>=20
> "Why is it okay to use a psql variable declared on a command line in a ps=
ql session
> that was started from said command line and not with a session that was c=
reated and
> destroyed with a -c command switch?"

"Why" questions tend to be difficult to answer.
The simple answer is: because somebody implemented it that way.

I dug into the commit history, and the current behavior seems to stem from =
commit
a45195a191 from 1999 that introduced psql variables and has the lapidary co=
mmit
message "Major psql overhaul by Peter Eisentraut".
I couldn't find a pertinent discussion on the list.

If you dislike the current behavior enough, you could write a patch that ch=
anges
it.  There might be some resistence, however, because a behavoir change lik=
e that
could break existing scripts and persent a backward incompatibility.

Yours,
Laurenz Albe