GSL, FFI, GCL, Windows; was: find_root with bfloats

  On 9/29/2010 8:19 AM, Robert Dodier wrote:
> On 9/29/10, Raymond Toy<toy.raymond at gmail.com>  wrote:
>
>> You mean having maxima use gsl to implement special functions?
>>
>> This (and related) issues has come up before.  The main blocking point
>> is that it doesn't work with all of the Lisps we currently support and
>> gcl in particular since that is still the primary lisp for distribution.
>>   (I think; I don't build distribution binaries.)
There is another issue, which is not a disqualification for what GSL 
does, but for what it
doesn't do.  What one presumably would like is a library that fits the 
computer-algebra
model of what can or should be computed in a sometimes-exact, 
sometimes-high-precision,
or sometimes-symbolic context.

   For example...

f(<rational>)   should do something useful, perhaps even returning an 
exact rational.
f(<bigfloat>) similarly.

  Perhaps it should be
f(<bigfloat>, desired_error_bound) --> bigfloat

perhaps singularities and infinities should somehow work in concert with
the operations in Maxima.  Not that they are necessarily right, currently.:)

and probably
f(<double-float>).   which is likely what GSL does, mostly.



RJF