[Prev][Next][Index][Thread]

Re: Java / Java on the Mac in January / JavaScript




On Fri, 1 Dec 1995, John C. Mallery wrote:

> At 6:36 PM 12/1/95, Andrew Moreno wrote:

> >One of the things I'm programming in Lisp is an agent that can work like 
> >Eliza and store like Boris Katz's START system. Would Katz's START system 
> >be workable through Java like it is through CL-HTTP? (Assuming you have 
> >tried the START system.)
> 
> Yea, I have written my own natural language system. See my home page. No 
> chance that it will be rewritten in Java.  

Can you clarify? No chance that it will be rewritten in Java because you 
or other's won't do it or because it's impossible to write that system with
Java.

The reason I mentioned Katz's work is because I think he's on to 
something with his "reverse transformations." Does your system use that?

> Here, we are looking forward for major qualitative
> advances in technology.  That means, programming environments better than
> the Lisp machine -- and that is 1980 technology. Most of what everyone 
> else is using is 1960s technology. We are only four years from the 
> twenty first century.

Most programmers use C and C++.
I've read that there are things Lisp programmers can do with Lisp that are 
impossible to do with C or C++. Is C or C++ (a fancy term for Macro-11) 
and their programming environment's what you mean by 1960's technology?

CL-HTTP is a 21st century programming environment?

> I guess one can say that Java gives garbage collection to the masses. But,
> it is hard enough to program sophisticated applications in Lisp.  Current 
> java is many steps backward from lisp or scheme.  

I'm supposed to find a copy of SICP by Abelson and Sussman but I can't find
a copy. Supposedly, if I work through the exercises in the book, I'll be able
to generate well-abstracted code that is extensible, readable and debuggable.
Unfortunately, I can't find a copy in Vancouver.

My point is, programmers now don't seem to value writing extensible, 
readable and debuggable code that is efficient. Interpreted Java programs 
would need to be efficient and tightly written and that seems to go against
the trend of programmers not writing tight code.

> the useful thing about it is that you can tell the client what to do.  A 
> Scheme or Lisp plug-in to Netscape on all platforms would be a better 
> target, but you work with what everyone has got.

I saw something about a Scheme project that will let Scheme applets run 
over the web. Is this connected to the CL-HTTP project?

Thanks for answering my novice questions, I've read most of what's on the
MIT AI server. Good information.

Andrew Moreno





Follow-Ups: References: