Date: Fri, 30 Jan 1998 22:40:44 -0800 (PST)
Message-Id: <199801310640.WAA09939@adit.ap.net>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: Thomas Demmer <demmer AT LSTM DOT Ruhr-UNI-Bochum DOT De>, djgpp AT delorie DOT com
From: Nate Eldredge <eldredge AT ap DOT net>
Subject: Re: x/0, and a problem with realloc
Cc: viking AT caverock DOT net DOT nz
Precedence: bulk

At 12:12  1/30/1998 +0100, Thomas Demmer wrote:
>> One question, to A. Sinan Unur, what are Extended reals? You mentioned
them in
>> reference to dividing by zero in an article (29-Jan-1998).
>a real is usually 4 byte long, a double 8 byte, and an extended real 10
>bytes.
>They differ in signficant digits and maximum/minimum  exponent.
>An extended real is the "native" data representation for the x87.
I think two different things called "extended real" are being talked about
here. One seems to be a conceptual point of numerical analysis, and the
other is an FPU implementation issue. Seems to me the one we're interested
in here is the one that's "the set of real numbers plus positive and
negative infinity". To me at least, it's not clear how one would actually
get this to work.

Nate Eldredge
eldredge AT ap DOT net