Simple optimizations missing

[Geoff Keating]

dewar@gnat.com writes:

> <<For floating point: 0.0/0.0 might generate a floating point exceptions
> and the result is a NaN.  The same applies for the infinites.
> >>
> 
> Of course this is only required if you are trying to be strict IEEE, I 
> would think it perfectly fine to do this optimization unless -mieee were set.

GCC is strict IEEE by default on platforms where this is reasonable
(like powerpc and anything with soft-float).  IMHO it should be strict
IEEE everywhere by default and people should use -ffast-math (or
something) when they know non-IEEEness is safe.

-- 
- Geoffrey Keating <geoffk@geoffk.org>