On Thu, 16 May 2002 15:21:15 +0200, Mark Weyer
<wey### [at] frege mathematik uni-freiburg de> wrote:
> 2.01*0.5+2.01*0.5 = 1.01+1.01 = 2.02
> depending on how you round 1.005.
Each of components is integer (or should be becouse everything is float
internally during parsing). If integers are represented with very small
floating point accuracy then such problem appear for original version in
macros as well so my optimizations are still valid.
ABX
Post a reply to this message
|