> If Maxima could just step aside and let the cpu figure out the result, that
> would be an improvement. This applies to floating point comparisons as
> well as arithmetic results.
"Just stepping aside" has a good chance of working correctly for
all-float calculations.... Though there may be bugs relating to
eq/eql/=/equal and also to things like 0.0*flinf (because Maxima might
assume that 0.0 times anything = 0.0).
However, I suspect it will mess up mixed symbolic and numerical
calculations in some cases, just the way the current INF, UND, ZEROA,
etc. mess up sometimes, e.g. INF-INF => 0.
So it's a bit more work than just using IEEE floats blindly.
-s