Re: ISO C99: Macros MATH_ERRNO, MATH_ERREXCEPT, math_errhandling

[Andreas Jaeger <aj at suse dot de>]

>>>>> Ulrich Drepper writes:

Uli> Andreas Jaeger <aj@suse.de> writes:
>> If we'd like to be on the safe side, we could set math_errhandling to
>> 0 - but that's lazyness, I do think we can do better ;-).
>> 
>> What's the right value to use?

Uli> We (rth, me) discussed this already.  We'll hopefully have some gcc
Uli> support for this since there is such an option already in gcc.  This
Uli> will decide about the value.  If anything thereafter has to be fixed
Uli> in glibc so be it.

We need to define MATH_ERRNO and MATH_ERREXCEPT as macros when
<math.h> is included.  I do think we've got to change glibc for this.

Getting support for this in gcc seems to be the right thing - but we
do need also program accordingly in glibc (if math_errhandling is 3,
all math functions have to honour this).

What option is already in gcc for this?  Richard, are you going to
handle the gcc part?

Uli, please add this to your glibc 2.2 ToDo list.

Thanks,
Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.rhein-neckar.de