Re[2]: [CM] Problems with Clisp

Frantisek Stafek
Thu, 6 Apr 2006 20:27:48 +0200

6. dubna 2006, 20:06:09, Bill:

>> What are your plans for the future of CLM? 

BS> Gah!  I try not to look ahead at all.  I no longer enjoy working on the
BS> Common Lisp version -- the various FFI's are horrible, and debugging
BS> is impossible.  Whenever I'm goofing around with it, I use Snd and
BS> Guile/Scheme, even for with-sound stuff (it's only a factor of 2 or pi
BS> or something-like-that slower than the compiled CL version, and 
BS> I don't think fast anymore anyway).  I'll certainly keep the CL version
BS> going even if Fernando and Chris decide to use something else
BS> for the introductory classes here.  Rick suggested making the current
BS> CL-CLM a pure lisp program, but that slows it down by a factor of
BS> 30 to 50; adding all the type declarations (much worse in CL than
BS> in C because every damned expression has to be wrapped in "the")

Interesting. I'm just CL newbie, sometimes I read discussions on
newsgroups about CL optimizations, where different people claim one can
also use CL for fast code. Problem with unreadable code with
declarations is often mentioned as one that is possible to overcome by
some simple macros. Your complaints sort of support image of CL better
suited for high-level tasks here...

BS> makes the code unreadable, so why use Lisp at all at that point?
BS> This sounds more negative than I mean for it to -- no big changes
BS> are in the works, as far as I can plan ahead.

BS> If it had worked out as I originally hoped it would, composers would
BS> be developing complicated instruments using new synthesis methods,
BS> but the energy now seems to be back in what I'd call music concrete!
BS> Maybe it's a big circle...

BS> _______________________________________________
BS> Cmdist mailing list

Frantisek Stafek.