RE: [boost] Formal Review for Interval Library- gettingintervalsofconstants

From: Paul A. Bristow (boost_at_[hidden])
Date: 2002-09-08 06:19:45

I think we should stick to radix 2 for starters
unless I hear some shouts for any other radix.

Itty Bitty Machines always were a bit odd!

Paul

Dr Paul A Bristow, hetp Chromatography
Prizet Farmhouse, Kendal, Cumbria, LA8 8AB  UK
[phone]  Mobile [phone]
mailto:pbristow_at_[hidden]

> -----Original Message-----
> From: boost-bounces_at_[hidden]
> [mailto:boost-bounces_at_[hidden]]On Behalf Of Gabriel Dos Reis
> Sent: Saturday, September 07, 2002 1:03 AM
> To: boost_at_[hidden]
> Subject: Re: [boost] Formal Review for Interval Library-
> gettingintervalsofconstants
>
>
> "Paul A. Bristow" <boost_at_[hidden]> writes:
>
>
> One of my previous points is that it is unwise to assume that just
> because you're using IEEE-754 system means that the long double
> datatype is 80 bits wide.  If you take an IEEE-754 plateform like a
> SPARC, you get a 128 bits.  Some IEEE-754 plateforms use 96 bits.
> That means that a scalable way to tackle the issue it to parameterize
> on the relevant FP system description as given by numeric_limits<>.

WE are all agreed on this and a switch seems the most attractive.
I can provide the intervals for any value of digits == number of significand
bits.
Are there any cases where numeric_limits<>::digits is unreliable.

> | (Are there really any machines using radix other than 2 that we need to
> | cater for?)
>
> If you have users with some IBM machines with radix = 16, then you'll
> have to add the support.  I know I did use such a system for
> computations a year ago. However, I can't tell you whether I'll use it
> in the near future or not -- I'm not yet graduated in Madame Soleil
> magic :-) :-)
>
> -- Gaby