>>>>> "Vadim" == Vadim V Zhytnikov <vvzhy@mail.ru> writes:
Vadim> C Y wrote:
>> --- U-E59264-Osman F Buyukisik <absd00t@c1186.ae.ge.com> wrote:
>>
>> > GCL 2.4.0 does not build on hpux anymore :-( HPUX support was
>> > nonexitent anyways. Bill did not know and the initial port was done
>> > by
>> > someone who is not involved with hpux anymore. 2.3.8 works ok.
>> >
>> > I am in favor of maintaining/strengthening CLISP build. Seems to
>> > work
>> > on any platform. If you need speed, maybe you'll need to buy a
>> > commercial common lisp implementation or be lucky cmucl/sbcl
>> > working on your platform.
>>
>> I've wondered about this point - CLISP is slow, granted, but maybe we
>> could work with them to find and remove the bottlenecks? There seems
>> to be much to recommend CLISP if it weren't for the speed issue. We
>> have gcl and cmucl to refer to when suggesting to them what the
>> bottlenecks are to remove.
>>
Vadim> Clisp programs are slower since clisp's compiler do not produce
Vadim> native machine code but rather some special portable code (.fas files).
Vadim> This approach is more flexible but obviously less efficient.
Vadim> So this is design concept and can't be easily changed.
However, if there is a particular Common Lisp function or set of
Common Lisp functions that are the bottle neck it might be possible to
get the Clisp folks to implement them in C as is done for many, many
Common Lisp functions now.
I doubt there is such a thing, though, or they are already implemented
in C.
Ray