>
> > Barton> Since 1.0 * %i --> %i, numeric contagion has a problem:
> > Hmm. That is a problem. I suppose we could turn off the conversion
> > of 1.0*%i to %i.
> I struck this when I was implementing Airy functions. The conversion
> of 1.0*%i to %i interferes with the numerical evaluation of airy(%i).
>
Yes, I think I mentioned this a long time ago, either on the mailing list or
in a bug report.
Maxima treats 1.0 as equivalent to 1 in simplification, and so
1.0*xxxsimplifies to xxx. This removes the information that the
result is
approximate, not exact.
Interestingly, 0.0*x correctly gives 0.0, not 0.
So we have 1.0*x - 1*x => 0 but (1.0 - 1)*x => 0.0.
-s