[Shootout-list] Two notes: Pascal and CMUCL

Brent Fulgham bfulg@pacbell.net
Sat, 09 Oct 2004 23:11:37 -0700


On 2004-09-30 10:07:45 -0700 kerochan ii <kerochan2@gmail.com> wrote:

> sense. The compiled FASL code _needs_ the lisp
> runtime. Of course you can tell GCC to link your app
> dynamically so it needs a (runtime?) library too.
> However, you cannot get CMUCL to "link" the code
> statically... See?
> I think a different category might be added like
> "compiles to standalone code".

It's a reasonable question, and one that is being debated
on the list, but I don't think we are fine-grained enough to
describe this difference on the site right now.

What you are pointing out is more of a "startup time" issue
than a question of whether the code is interpreted or not.
CMUCL is clearly not producing bytecode -- it is compiling
to straight machine code, which does require the existance
of the Lisp runtime but is not "interpreted."

> Shootout ;) I would be happy to port/fix/extend the
> Pascal code to work on Linux with FPC/GPC and see you
> surprised when it beats C++ :>
> (But even if it doesn't, I am sure it won't be one of
> "those at the end".)

If you provide the tests I'll add them to the benchmark,
assuming the compiler is freely available and can be
installed on my machine.  GPC certainly fits this requirement,
I'm not sure about the other.

Thanks,

-Brent