CLISP, Linux (NT?) and life?

[Marcus Daniels <marcus@sysc.pdx.edu>]

>>>>> "Phil" == Phil Perucci <pperucci@access.digex.net> writes:

Phil> Will 1) CLISP development continue
Phil> (the sunsite.unc.edu binary is from summer of '94)

I'm not sure what you are driving at.  CLISP development is not
funded; I don't know that it ever was.  Generally, features are added
to CLISP whenever someone is willing and able to implement said
features.  Ports are done when someone decides to enhance CLISP so
that it runs on their platform.

That a Linux binary exists on sunsite should only lead you to believe
that a CLISP developer used Linux.  ma2s2.mathematik.uni-karlsruhe.de
remains the official distribution site for CLISP.

Phil> and will 2) Unix continue to be a "preferred"
Phil> platform? 

My spin is that Unix is preferred because Unix is a more
development-friendly environment (than NT).  That is, Unix being
preferred for CLISP is really an effect rather than a cause.  Note
that only one component of "developer-friendly" is technical
superiority.  Another reason for preferring Unix is that there are
several freely available Unix implementations.  Note also that CLISP
is GPL'ed.

Phil> I dumped my NT coding efforts in Visual C++, thinking
Phil> AI researchers still preferred Lisp and Unix.

If you find C++ suitable to your research, use it!
It might well be the case that if your interest was expert-systems,
the C++ world is as well or better equipped to address your needs.
Still, "AI researchers" seems like a impossibly broad category to me --
does that include all of optimization, machine learning, and evolutionary
computation?  If so, I doubt a recommendation w.r.t. to OS's or
programming languages would make sense.

Phil>   Are even researchers and universities gravitating to the
Phil> not-so-free world of Microsoft?

Whether they are or not, I don't see how it pertains to CLISP.  :-)
Keep in mind that this perception of "better" is contextual.  The
context is the development of a programming tools, not user
applications (an entirely artificial distinction, IMO).  What I see
is that proprietary systems inhibit the development of programming
tools.  For CLISP, the situation is especially pronounced, since CLISP
tries to work closely with the C compiler (usually GCC), and closely
with OS features (like mmap).  Compiler bugs and OS bugs can be
crippling if a developer is not empowered to correct matters.  Few
developers have such pull with Microsoft.  As an example, Bruno
Haible actually influenced the design and implementation of Linux's
mmap -- not to a small extent out of insights provided by his work on
CLISP!

At one level, CLISP is just a body of code.  At another level, the
users perspective, CLISP is system.  Together, these facts keep me
going in the direction of using freely available retargetable tools,
instead of switching to proprietary tools.  Using GCC and Cygnus'
win32 DLL generator and libraries provides the means for doing nearly
100% of the work of NT/95 maintenance -- all on Unix.

Phil>   Currently I use Patrick Henry Winston's books "Artificial
Phil> Intelligence" and "Lisp" for my fundamental approaches.  Is
Phil> there a more "hip" AI guru these days?

Peter Norvig has a good intro/reference book.  Grab the Lisp FAQ
posted to comp.lang.lisp and comp.lang.scheme 
(or rtfm.mit.edu:/pub/usenet/comp.lang.lisp) if you haven't already.
Lots of stuff on texts.  

Some AI-related sites that might help hook you up:
http://www.cs.washington.edu/research/jair/home.html
http://www.comlab.ox.ac.uk/archive/comp/ai.html#sites