[Ecls-list] wrong value for (float-sign -0.0)

[Gabriel Dos Reis]

On Sun, Sep 28, 2008 at 7:13 PM, Raymond Toy <toy.raymond at gmail.com> wrote:

>> What do you call `unsigned zero'? And why is it necessary?  We have been
>> successfully writing scientific applications with signed zeros as
>> described by IEEE 754 for decades.  Why is it bad to have signed zero
>> without the other one?  In what concreate situations is it bad to have
>> signed zeros but not `unsigned zero'?
>>
> I am not an expert in this area.  I know Kahan has argued strongly for
> signed zeroes.

Do you have substantial strong reasons to believe that we should disregeard
his arguments -- see reference posted earlier, or visit Prof. W. Kahan's
website.

I have been doing numerics for nearly 14 years (mostly for living); I'm glad
that the C committee finally recognized standard practice.  I'm glad that
ECL maintainers are taking more practical and useful approach to ECL
development and its uses.

>  Some discussions on comp.arch.arithmetic from
> knowledgeable people think signed zeroes without an unsigned zero is
> bad.

Well, hearsays are not very helpful, especially when they are not substantiated.

>  An unsigned zero is a zero exactly on the axis, as opposed to +0
> which is, essentially, above the axis and -0 which is below the axis.

which cannot be represented in IEEE-754, since that standard (widely
adopted by hardware vendors) explicitly defines every floating point
number in terms of its sign, mantissa and exponent.  Including 0.0.
Given that ECL assumes IEEE floating point on all systems, whatever
the hearsays say about things outside IEEE 754 appears to be completely
irrelevant.

-- Gaby