RE: [boost] LWG reaction: Header <cstdint>

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-11-01 16:36:20

At 01:05 PM 11/1/2001, Paul A. Bristow wrote:

 >Was there any discussion on the C99 additions to math.h, which
 >one will get by #including <cmath>

No specific discussion.  My understanding is that Bill Plauger (also know 
as PJP) will propose bringing all the C99 library features forward into 
C++, with modifications only proposed where there is an outright clash 
between C99 and C++98.

If you are interested, be sure to look at the publicly available 
documentation: http://www.dinkumware.com/refxc.html

If any Boosters think that there should be changes beyond what PJP is 
likely to propose, those of us on the committee will figure out some way to 
get your concerns addressed early on.

--Beman

 >C99 math.h seems to include a lot of MACROs and some math functions like
 >gamma
 >(as tgamma - note just three versions for float, double and long double)
 >
 >but most important, it adds standard methods of
 >handling floating point infinity, NaN and FP exception handling.
 >
 >For example, MSVC 6 provides int _fpclass( double x );
 >
 >but this is not portable.
 >
 >Should C++ math functions start to use these MACRO definitions,
 >as for example:
 >
 >#define fpclassify(x) <int rvalue>
 >
 >'returning' FP_INFINITE, FP_NAN, FP_NORMAL ...
 >
 >be used,
 >
 >or should fpclassify be templated in some way for C++?
 >(suggestions now?)
 >
 >Paul