From: Tom Lane <tgl@sss.pgh.pa.us>
To: Dean Rasheed <dean.a.rasheed@gmail.com>
cc: Bruce Momjian <bruce@momjian.us>, Zhihong Yu <zyu@yugabyte.com>,
        Daniel Gustafsson <daniel@yesql.se>,
        Bryn Llewellyn <bryn@yugabyte.com>,
        Justin Pryzby <pryzby@telsasoft.com>,
        Isaac Morland <isaac.morland@gmail.com>,
        John W Higgins <wishdev@gmail.com>,
        PostgreSQL-development <pgsql-hackers@postgresql.org>
Subject: Re: Have I found an interval arithmetic bug?
In-reply-to: 
 <CAEZATCVMXYR4-GEr4UXzZVdS1CLwFweO7P=vZrepcMu6VrQMUQ@mail.gmail.com>
References: <20210508024204.GB27048@momjian.us>
 <CALNJ-vQLukGOX5HKa7Z5tKYfNOyB6u0dU2+SuRFXExoJ_7m8QA@mail.gmail.com>
 <076E1871-8446-405C-9561-29FA03F46A4A@yesql.se>
 <20210630163516.GA3304@momjian.us>
 <CALNJ-vTD709opMzOyfT=KVtXaAvLmpqPGGEqqqXMSZwF8aQH1A@mail.gmail.com>
 <CALNJ-vQfTHNbH3VCeg3ChT42y1swnWtV-6iztthO8QfOE8wdwg@mail.gmail.com>
 <20210720041419.GA5882@momjian.us>
 <CALNJ-vSc22BFW__bUqJk0USvn8_w+fWLWJjNqH=6WSa1-OJRTQ@mail.gmail.com>
 <20210720225350.GB27025@momjian.us>
 <CALNJ-vQXQGPQiKt1SubT2fd4xGuvoN9UUtJyVr8JRfPxBfQHgg@mail.gmail.com>
 <20210721024834.GA8395@momjian.us>
 <CAEZATCVMXYR4-GEr4UXzZVdS1CLwFweO7P=vZrepcMu6VrQMUQ@mail.gmail.com>
Comments: In-reply-to Dean Rasheed <dean.a.rasheed@gmail.com>
	message dated "Wed, 21 Jul 2021 09:23:18 +0100"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <677936.1626861539.1@sss.pgh.pa.us>
Date: Wed, 21 Jul 2021 05:58:59 -0400
Message-ID: <677937.1626861539@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/677937.1626861539%40sss.pgh.pa.us>
Precedence: bulk

Dean Rasheed <dean.a.rasheed@gmail.com> writes:
> Hmm, looking at this whole thread, I have to say that I prefer the old
> behaviour of spilling down to lower units.

> For example, with this patch:

> SELECT '0.5 weeks'::interval;
>  interval
> ----------
>  4 days

> which I don't think is really an improvement. My expectation is that
> half a week is 3.5 days, and I prefer what it used to return, namely
> '3 days 12:00:00'.

Yeah, that is clearly a significant dis-improvement.

In general, considering that (most of?) the existing behavior has stood
for decades, I think we need to tread VERY carefully about changing it.
I don't want to see this patch changing any case that is not indisputably
broken.

			regards, tom lane