Re: Have I found an interval arithmetic bug?

On Tue, Jul 27, 2021 at 04:01:54PM -0400, Tom Lane wrote:
> Bruce Momjian <bruce(at)momjian(dot)us> writes:
> > I went ahead and modified the interval multiplication/division functions
> > to use the same logic as fractional interval units:
>
> Wait. A. Minute.
>
> What I think we have consensus on is that interval_in is doing the
> wrong thing in a particular corner case. I have heard nobody but
> you suggesting that we should start undertaking behavioral changes
> in other interval functions, and I don't believe that that's a good
> road to start going down. These behaviors have stood for many years.
> Moreover, since the whole thing is by definition operating with
> inadequate information, it is inevitable that for every case you
> make better there will be another one you make worse.

Bryn mentioned this so I thought I would see what the result looks like.
I am fine to skip them.

> I'm really not on board with changing anything except interval_in,
> and even there, we had better be certain that everything we change
> is a case that is certainly being made better.

Well, I think what I had before the multiply/divide changes were
acceptable to everyone except Bryn, who was looking for more
consistency.

> BTW, please do not post patches as gzipped attachments, unless
> they're enormous. You're just adding another step making it
> harder for people to look at them.

OK, what is large for you? 100k bytes? I was using 10k bytes.

--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com

If only the physical world exists, free will is an illusion.