f2cl, cernlib, and maxima (Re: [Maxima] Release plans)

On Wed, 2002-03-20 at 11:15, Richard Fateman wrote:
> Back to technicalities:
> 
> 
> There is a key difference in the numerical software (claims)
> of Mathematica vs this other stuff.  Mathematica numerics
>   are all variable-precision, and so one can crank up the
> accuracy of any of their functions. You should be aware
> of this.   

Good point. I have been quietly worrying about accuracy beyond double
precision in Maxima, but I don't know what to do about it.

I cannot help but slip into a rant about Mathematica at this point. I
know Mathematica claims to have variable (is "arbitrary" a better word?)
precision numerics, but I also remember that in my experience using
Mathematica, it was only true for some undocumented subset of
operations. Someone please correct me if I am wrong. I know Mathematica
will give you 10,000 digits of Pi, but will it give you 50 digits of,
say, BesselJ[7/13]? How about the rest of the special functions it knows
about?

--Jim