Re: [boost] Math Constants Library formal review results

From: Rainer Deyke (root_at_[hidden])
Date: 2001-04-25 14:06:44

----- Original Message -----
From: "Jens Maurer" <Jens.Maurer_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Wednesday, April 25, 2001 11:18 AM
Subject: Re: [boost] Math Constants Library formal review results

> Having binary precision (i.e. some handcrafted hex value coerced
> into a float) does not inhibit constant folding per se; the
> implementor may have such optimization in its compiler.  Besides,
> constant-folding surely is an optimization thing and thus a
> quality-of-implementation issue anyway.

I still don't see why handcrafted hex values would even be necessary.  All
floating point values can be represented exactly as a finite length decimal
number.  Given that, a compiler would not have to do any rounding when
converting such a decimal number back into binary.  Are there really
compilers that still get it wrong?

--
Rainer Deyke (root_at_[hidden])
Shareware computer games           -           http://rainerdeyke.com
"In ihren Reihen zu stehen heisst unter Feinden zu kaempfen" - Abigor