From: Tom Lane <tgl@sss.pgh.pa.us>
To: ab@principiamentis.com
cc: pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #16022: to_json on arrays with unusual lower bound is not
 intuitive
In-reply-to: <16022-5c864f8a572d5b5b@postgresql.org>
References: <16022-5c864f8a572d5b5b@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Wed, 25 Sep 2019 12:23:11 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <23021.1569422056.1@sss.pgh.pa.us>
Date: Wed, 25 Sep 2019 10:34:16 -0400
Message-ID: <23022.1569422056@sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> I would expect the array indices of a JSON array to match with the
> subscripts of a postgres array when converting them back and forth.
> However, Postgres lets the JSON array begin at the `lower_bound` of the
> array, not at subscript 1 (= index 0).
> For example, `to_json('[-2:5]={-2,-1,0,1,2,3,4,5}'::int[])` results in
> `[-2,-1,0,1,2,3,4,5]` where I would have expected `[1,2,3,4,5]` (or an
> error, since JSON arrays must not have negative indices).

I can see no reason whatever for either of those definitions to be
better than the established one.  If you want some other conversion
rule, write your own function that behaves the way you want.

			regards, tom lane