From: Tom Lane <tgl@sss.pgh.pa.us>
To: Adrian Klaver <adrian.klaver@aklaver.com>
cc: "Colin 't Hart" <colinthart@gmail.com>, pgsql-general@lists.postgresql.org
Subject: Re: Regression in Postgres 17?
In-reply-to: <6eded364-8072-4ca4-9ba7-1003a120c2b8@aklaver.com>
References: <CAMon-aR07TWHZt2QX9QsZQEniyE-6+FdHJSGwyMwvegJSjjNug@mail.gmail.com> <6eded364-8072-4ca4-9ba7-1003a120c2b8@aklaver.com>
Comments: In-reply-to Adrian Klaver <adrian.klaver@aklaver.com>
	message dated "Tue, 22 Oct 2024 09:03:05 -0700"
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-ID: <1971273.1729613566.1@sss.pgh.pa.us>
Content-Transfer-Encoding: 8bit
Date: Tue, 22 Oct 2024 12:12:46 -0400
Message-ID: <1971274.1729613566@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/1971274.1729613566%40sss.pgh.pa.us>
Precedence: bulk

Adrian Klaver <adrian.klaver@aklaver.com> writes:
>> In Postgres 17 trying to create the function yields an error:
>> 
>> pg17> create function json_test(out value text, out json jsonb)
>> returns record
>> ...
>> ERROR:  syntax error at or near "jsonb"
>> LINE 1: create function json_test(out value text, out json jsonb)
>> 
>> Am I doing something wrong? Or is this a regression?

> Yes you are doing something wrong, naming an argument with a type 
> name(json) is not a good idea.

The actual problem is that the SQL standards committee invented
some bizarre syntax that we couldn't parse without making JSON
a partially-reserved word.  It still works as a type name, but
in this particular syntax where it's not initially clear which
names are type names, you lose.  Double-quote the argument name,
or name it something other than "json".

			regards, tom lane