About alternatives to Matlab

[Carl Banks]

sturlamolden wrote:
> Carl Banks wrote:
>
> > > Ok. Perhaps starting a Python JIT in something like MetaOCaml or Lisp/Scheme
> > > would be a good student project?
> >
> > ...and finishing would be a good project for a well-funded team of
> > experienced engineers.
>
> I think this is a good idea. We could use the AST from the CPython
> compiler, translate on the fly to ANSI Common Lisp, and pass it off to
> SBCL (which is very fast and supports native threads). It would even
> solve the issue of Python's crappy GIL and reference counting (as
> opposed to a modern garbage collector).

I doubt it would work too well.  Lisp's semantics differ from Python's
in subtle ways, and covering the edge cases usually means you have to
do all the stuff you thought you would avoid by using another
dynamically-typed language.   Translating Lisp integers to Python ints
would work probably 99 percent of the time, but in the end you'd
essentially have to reimplement the Python int type.

If you're going to go through all that work, you might as well
translate it to C or directly to machine code.


Carl Banks