[plt-scheme] Improving mzc output

At 12:48 PM 1/10/2005, Eli Barzilay <eli at barzilay.org> wrote:
>On Jan 10, Blake McBride wrote:
> > Is there any plan to improve mzc?  As you are all probably aware,
> > the output of mzc offers little optimization when compared to
> > chicken or bigloo.  MzScheme is a great development platform and has
> > a lot of really nice features.  It would be nice to be able to
> > compile it to a form more comparable to C programs as does Chicken
> > and Bigloo.
>
>The current plan is to work on a more proper JIT compiler, instead of
>going through C.  How fast will that be compared to other compiler is
>a question, but it will most likely be an improvement of the current
>state (and hopefully an improvement over mzc-compiled code too).
>
>--
>           ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:


Why go through all the work of writing a JIT compiler when they are
entirely CPU specific?  Doesn't it make more sense to spend all that
effort to generate portable optimized C?  Who cares how fast the interpreter
works (within reason) so long as a fast compiler is available?  Portability
is an important issue.

I know JIT compilers are in vogue but an optimizing compiler through C makes
much more sense.

Thanks.

Blake McBride
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.racket-lang.org/users/archive/attachments/20050110/d19ab6c9/attachment.html>