> > I might mention that the lisp output would be significantly easier
> > to parse and maintain.
>
> Are you saying that aldor should not be translated to C or Fortran
> or whatever other target there will be in the future?
A point which is probably orthogonal to this discussion but is
something to consider is why we would want to target Aldor towards
something other than lisp, in the first place.
Axiom's core is currently based on Lisp, and I personally think this is
a Good Thing. If we output Aldor to something other than Lisp, we face
the problem of communicating between the main Lisp environment and
whatever happens to the non-Lisp code Aldor has exported. To me, this
seems to invite a world of compatibility issues, cross-platform
problems, FFI issues, etc. Aren't we better off using other libraries
which have been developed specifically to solve numerical problems
quickly, and gradually working on optimizing our Lisp code?
Am I missing something?
Exporting to C is a good think, if Aldor can create library in C other program could use it easily. I am not sure if maintaining a Fortran back end is still relevant.
I do not see has a problem not to be able to mix C Aldor and Lisp Aldor.
And FFI functionality should be back end independent. For example I should be able to use GMP in all cases.