why are the float math functions unimplemented or defined ?

why are the float math functions unimplemented or defined ?

Post by Kurt Klingbe » Sat, 22 Apr 1995 04:00:00



Is there any particularly good reason that expf et al   i.e. the
float math functions are just left unimplemented and undefined ?

I remember the bad_old_early_days of HPUX where porting was always
an adventure since there was so much stuff unaccountably missing from
the header files.  Recently it's been deja vue all over again with
Linux:  snmp stuff wouldn't build because ip-fragmentation-related
calls are not supported, fping/Satan won't build because the OOTB
ip/udp headers are broken, and now even basic math library stuff
is only semi-implemented :-(  

kk

 
 
 

why are the float math functions unimplemented or defined ?

Post by H. Peter Anv » Sun, 23 Apr 1995 04:00:00




In newsgroup: comp.os.linux.development.system

Quote:

> Is there any particularly good reason that expf et al   i.e. the
> float math functions are just left unimplemented and undefined ?

> I remember the bad_old_early_days of HPUX where porting was always
> an adventure since there was so much stuff unaccountably missing from
> the header files.  Recently it's been deja vue all over again with
> Linux:  snmp stuff wouldn't build because ip-fragmentation-related
> calls are not supported, fping/Satan won't build because the OOTB
> ip/udp headers are broken, and now even basic math library stuff
> is only semi-implemented :-(  

On the i386 (i387 really) architecture, there is no benefit from using
"float" except for saving storage since all operations are done on
80-bit quantities no matter what.  Hence you can just #define expf(X)
exp(X) etc if you want.

        /hpa

--
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah