Re: Math Functions: Basic building blocks...

From: Eric Ford (eford_at_[hidden])
Date: 2001-09-20 17:57:00

> I think a relatively low level templatized interface to the floating
point mathematical functions provided by the standard C library might
be a good idea.

I'd like to see some progress on this front.  I think we should start
very simply with something that can be used as a part of a more
powerful mathematical library to come.  For example, presently the
"special functions" library uses the double version of routines such
as sqrt, sin, log, rather than a version optimal for the arguments
type.  I have in mind something that merely chooses the correct
version of the functions provided by the standard library.  That way
there's no additional runtime costs and minimal compile time costs. 
Eventually, I'd like to provide versions which are capable of
providing some error checking, but I think we should make something
like this simple library first.  

I've uploaded a draft of the documentation to
http://groups.yahoo.com/group/boost/files/standard
_functions/standard_functions.html
.  The header file, test program, and test program output can be
accessed via links for the standard_functions directory in the files
section.

If a few people would take a look at it, I'd appreciate it.  One thing
I know I'll need some help on is making it work on other compilers. 
When the ridciulous #define
MY_COMPILER_GCC_296_DOESNT_SUPPORT_THESE_HOW_SHOULD_I_DEAL_WITH_THEM
is commented out, it compiles and runs fine for me (g++ 2.95 redhat
linux 6.2 i386).  If people with other compilers could verify that all
the functions used with that commented out work for their compiler,
that would be appreciated.  Also, I imagine some compilers could
provide some of these functions.  If anyone wants to let me know which
compilers support which, that could be useful.  I'm not sure whether
it would be better to only provide what nearly all compilers support
or to provide as much as possible on each compiler.  

Thanks,
E