From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | Greg Stark <gsstark(at)mit(dot)edu>, tomas(at)tuxteam(dot)de, Jeff Davis <pgsql(at)j-davis(dot)com>, Scott Bailey <artacus(at)comcast(dot)net>, hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Range types |
Date: | 2009-12-15 15:06:28 |
Message-ID: | 603c8f070912150706x2e1620c2g38a1eaa991079c20@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, Dec 15, 2009 at 9:58 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Greg Stark <gsstark(at)mit(dot)edu> writes:
>> In fact, as I only recently found out, one of the design goals of IEEE
>> floats was specifically that they sort lexicographically and use every
>> bit pattern. So you can alwys get the "next" float by just
>> incrementing your float as an 64-bit integer. Yes that raises your
>> value by a different amount, and it's still useful.
>
> There are certainly some low-level numerical analysis situations where
> you want to get the "next" float value, but that hardly constitutes
> an argument for treating ranges of floats as discrete rather than
> continuous. Normal users of a range datatype aren't going to be
> interested in dealing with that sort of inherently machine-specific
> behavior.
Yeah, I don't think we want to base this feature on something that
arcane. I also have to say that I'm very skeptical of the argument
that there is only a small list of types people will want this for. I
don't think it's going to turn out to be all that small.
...Robert
From | Date | Subject | |
---|---|---|---|
Next Message | Florian G. Pflug | 2009-12-15 15:15:19 | Re: Compiling HEAD with -Werror int 64-bit mode |
Previous Message | Robert Haas | 2009-12-15 15:03:54 | Re: Adding support for SE-Linux security |