Gaussian RNG?, numerics in general...



On Mon, 2002-02-04 at 14:31, Andreas.Eder@t-online.de wrote:
> Rather than get all the different FFIs to work and incorporate Fortran
> code into Maxima, I'd rather transfer the code to CL via f2cl and have
> it in lisp. 

The GSL code is C, not Fortran. The GSL codebase is also roughly twice
as big as Maxima. It isn't just a few functions. I don't find the
prospect of adding another 200,000 lines of code to Maxima enticing,
even (especially?) if it is automatically converted code.

> Sure, you would have to redo it on bugfixes to the original code, but
> then it is much easier to maintain lisp code than C or Fortran. And it
> is a lot easier to look for the bugs.

The point is to have the people who wrote, understand and use the code
fix it. We don't have to fix it at all, except to update to the new
version.

--Jim