On 7/17/09, Preben Randhol <rand...@pvv.org> wrote:
> (float)*w/(float)*h != (float) (w/h) in most cases also...
this expression won't even compile: dereference is only allowed for
pointers division for numbers
if you meant
(float)w/h != (float)(w/h)
with int w,h, then you are right, the former is floating point
division the later one is integer
but this is totally unrelated to the problem i was describing
my problem was that it seems that c and ieee754 does not even require
that the arithmetic operations are well defined: same operation on
same numbers with same type does not necessary give the same result
(even on same architecture with same compiler)
for example
(double)atoi("1")/atoi("3") == (double)atoi("1")/atoi("3")
gives false with gcc on my x86 machine eventhough the two expressions
are semantically equivalent
