From: Tom Lane <tgl@sss.pgh.pa.us>
To: jian he <jian.universality@gmail.com>
cc: Francisco Olarte <folarte@peoplecall.com>,
        Michael Bondarenko <work.michael.2956@gmail.com>,
        pgsql-bugs@lists.postgresql.org, dgrowleyml@gmail.com,
        Peter Eisentraut <peter.eisentraut@enterprisedb.com>
Subject: Re: BUG #18348: Inconsistency with EXTRACT([field] from INTERVAL);
In-reply-to: 
 <CACJufxGFuZsj+31a204fFpxv_RCHA-u1h6WEBZ6qpjBVoio5dQ@mail.gmail.com>
References: <18348-b097a3587dfde8a4@postgresql.org>
 <CAApHDvr_-05ce6FaLM0P=hJBdV=Ht3=sdk_Eq__7YOf+fs5x9g@mail.gmail.com>
 <CACJufxHYQSKNnt8rS7GxvK04wf75nw_ugMaibFVVfrcdVMmF1Q@mail.gmail.com>
 <CAF_O+z1FU5GN=LnW50QjANGVB8+6VxyK1JHa8fE-Xs205MFuqw@mail.gmail.com>
 <CA+bJJbw6q=6no1Hf67FraYGMo12asWCa7yw=J0zg_ELbivX0gw@mail.gmail.com>
 <3709307.1708201825@sss.pgh.pa.us>
 <CACJufxGB4MD6vRz3jVGK6kza-6fuDgfW2jC_1S2+MwcQYGkgSA@mail.gmail.com>
 <3749306.1708222748@sss.pgh.pa.us>
 <CACJufxHL0g73FPaRS+OGSYWJ0YbP6nEdVK79=wjL-TKU0_heDw@mail.gmail.com>
 <222870.1708458150@sss.pgh.pa.us> <270582.1708462564@sss.pgh.pa.us>
 <CACJufxFOjGpAQJ5EJ0=a1geERgdHW77SEdRyK8A_qwB3CVdwhQ@mail.gmail.com>
 <166456.1715117228@sss.pgh.pa.us>
 <CACJufxGFuZsj+31a204fFpxv_RCHA-u1h6WEBZ6qpjBVoio5dQ@mail.gmail.com>
Comments: In-reply-to jian he <jian.universality@gmail.com>
	message dated "Wed, 08 May 2024 09:03:56 +0800"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <476282.1715177448.1@sss.pgh.pa.us>
Date: Wed, 08 May 2024 10:10:48 -0400
Message-ID: <476283.1715177448@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/476283.1715177448%40sss.pgh.pa.us>
Precedence: bulk

jian he <jian.universality@gmail.com> writes:
> but in real life, for week, we generally begin with 1?
> like "the first week", "second week"
> so should
> select extract(week from interval '1 day');
> return 1
> ?

Hmm, I read it as being "the number of (whole) weeks in the
interval".  Starting with week 1 is what happens in the timestamp
case, true, but I don't find that appropriate for interval.
By analogy,

regression=# select extract(day from interval '23 hours');
 extract 
---------
       0
(1 row)

There's no such thing as "day 0" in the timestamp case,
but that doesn't make this wrong.

In any case, I'm starting to wonder why this issue is on the v17
open items list.  These are hardly new bugs in 17.  If there's
still differences of opinion about what the definition should be,
I think cramming in a change post-feature-freeze is not appropriate.
Let's just queue the issue for the next commitfest (already done
at [1]) and take it off the open items list.

			regards, tom lane

[1] https://commitfest.postgresql.org/48/4979/