Date: Tue, 27 Apr 2021 22:44:15 -0700
From: Noah Misch <noah@leadboat.com>
To: Bruce Momjian <bruce@momjian.us>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, neverov.max@gmail.com,
	pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #16939: Plural interval for negative singular
Message-ID: <20210428054415.GA2493304@rfd.leadboat.com>
References: <16939-cceeb03fb72736ee@postgresql.org>
 <20210323160336.GE579@momjian.us>
 <20210425115726.GA2353095@rfd.leadboat.com>
 <20210426163116.GR7629@momjian.us>
 <2266336.1619455534@sss.pgh.pa.us>
 <20210426165435.GT7629@momjian.us>
 <2267469.1619456564@sss.pgh.pa.us>
 <20210426170616.GV7629@momjian.us>
 <20210427061329.GB2435607@rfd.leadboat.com>
 <20210427201019.GY7629@momjian.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20210427201019.GY7629@momjian.us>
User-Agent: Mutt/1.5.24 (2015-08-30)
Archived-At: 
 <https://www.postgresql.org/message-id/20210428054415.GA2493304%40rfd.leadboat.com>
Precedence: bulk

On Tue, Apr 27, 2021 at 04:10:19PM -0400, Bruce Momjian wrote:
> On Mon, Apr 26, 2021 at 11:13:29PM -0700, Noah Misch wrote:
> > On Mon, Apr 26, 2021 at 01:06:16PM -0400, Bruce Momjian wrote:
> > > On Mon, Apr 26, 2021 at 01:02:44PM -0400, Tom Lane wrote:
> > > > Bruce Momjian <bruce@momjian.us> writes:
> > > > > On Mon, Apr 26, 2021 at 12:45:34PM -0400, Tom Lane wrote:
> > > > >> I agree with Noah's opinion that we should stick to the historical
> > > > >> behavior in the interval I/O functions.  There is not enough solidity
> > > > >> in the "this is grammatically wrong" argument to justify taking any
> > > > >> risk of application breakage, and it seems like there is some risk of
> > > > >> that there.
> > > > 
> > > > > Are you saying we should revert the patch and leave the plurals
> > > > > inconsistent in different places?  We were already expressing -1 as
> > > > > singular in most of the places I saw.
> > 
> > Which places did you find?  I used the following to check for code like you
> > were adding, but it probably missed things:
> > 
> > $ git grep '"s"' 5da9868^ | grep abs
> > 5da9868^:src/backend/utils/adt/datetime.c:						(abs(sec) != 1 || fsec != 0) ? "s" : "");
> > 5da9868^:src/interfaces/ecpg/pgtypeslib/interval.c:						(abs(sec) != 1 || fsec != 0) ? "s" : "");
> 
> Yes, I found the same like you did.  Are you saying we should remove the
> abs() calls here so we are consistent?

No, I advise just reverting 5da9868.  The output funcs should keep doing what
they did in v13, inconsistency and all.

I was asking the above question to see which of singular or plural was more
common before $SUBJECT's commit.  I count two examples of singular and seven
examples of plural (one in EVALUATE_MESSAGE_PLURAL(), six modified in commit
5da9868).  Given that, ...

On Mon, Apr 26, 2021 at 01:06:16PM -0400, Bruce Momjian wrote:
> Well, we should then add a comment that this is inconsistent but should
> not be changed.

... adding comments to the two places that use abs() to achieve singular "-1"
sounds fine.