Gaussian RNG?, numerics in general...



So, it looks like our options boil down to this (someone correct me if
I'm wrong):

1.  Fix the numerical code in Maxima, and live without the extra 
    features offered by GSL and the like.

2.  Impliment FFIs for all necessary cases, doing what we need to to 
    maintain them on different lisps, so we can use GSL and other such
    libraries as drop in added functionality.

3.  Impliment a few FFIs for some key functions, and otherwise try to 
    reimpliment functionality in Maxima.

All have pros and cons.  Personally I would tend to favor 2, but that's
just me and I don't have a good knowledge of what it would take to do
this.  If I may, I'll insert one or two ignorant annoying newbie
questions:

What are the actual differences between the FFIs?  What are the actual
issues which we would need to deal with?

Depending on the answer to the previous question, would it be possible
to do something like this:

  - Deside what the best FFI style is to use
  - Write interfaces in that style
  - Then, for each lisp, include files which teach that implimentation 
    of lisp what it doesn't already know about the good FFI style.
  - That way, we could have one set of code we could update to reflect
    changes in the external code, and one to handle local quirks of 
    the various lisps.

I'm sure that's impractical for some reason, but maybe knowing why it
won't work would help focus our thinking.  (Or mine, anyway :-)

CY

__________________________________________________
Do You Yahoo!?
Great stuff seeking new owners in Yahoo! Auctions! 
http://auctions.yahoo.com