Re: OT: Languages and type systems (was RE: Settlers ...)

[Johann Hibschman <johannh@...>]

On Sun, 28 Nov 2004 16:46:33 +0100, Adrián Pérez <copylight@iespana.es> wrote:
> 
> Yeah, "dynamicity" started with Lisp a long, long time ago...
> (but Lisp is not compiled to machine code).

Now you've done it.  You've angered the Secret Lisp Masters.  As a
long-time lisp junkie and medium-long Common Lisp fan, I have to tell
you that all commercial Common Lisps are compiled to rather efficient
machine code.

In fact, in a test run on an Ultra 10, CMUCL generated code that was
10% slower than Fortran and 20% faster than C, for a
reasonably-optimized Bessel function calculation.  (This was just my
one test, so it's clearly meaningless, but it shows that CMUCL is in
the ballpark of C and Fortran.)

The implementations of the Common Lisp object system are also compiled
to machine language and are about as dynamic (especially with the MOP)
as you can possibly get.  Multiple-dispatch methods are wonderful for
linear algebra systems.

This has been a public-service announcement.

Cheers,

-Johann