RE: [boost] Re: Re: Math functions - requirements 'spec'

From: Paul A Bristow (pbristow_at_[hidden])
Date: 2004-11-08 14:16:49

| -----Original Message-----
| From: boost-bounces_at_[hidden] 
| [mailto:boost-bounces_at_[hidden]] On Behalf Of Thorsten Ottosen
| Sent: 07 November 2004 11:13
| To: boost_at_[hidden]
| Subject: [boost] Re: Re: Math functions - requirements 'spec'
| 
|  Do all reasonable desktop compilers actually go for IEEE 754?

Well Dinkumware code used by Microsoft and many others only handles IEEE754.
| 
| Why can't you use std::numeric_limits<F>::is_nan() etc for C++ ?

My understanding is that only std::numeric_limits<F>::has_nan()

and A value std::numeric_limits<F>::quiet_nan()
- interesting but you can't test a NaN by comparing with this!
There are tons of possible NaNs :-(

are provided, but NOT what you really want is_nan().

But as noted by Hubert Holin, and his reference - this is a mess.

Paul

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

PS to Standards groups - where is this on your agendas?