On (13 Mar 97) Shamim Islam wrote to Jerry Coffin...
 SI> No compiler or interpreter is possible to be written at the basest
 SI> level within the same language framework, without a basic engine
 SI> written in a pre-existing language.
Rather the contrary - the interpreter is entirely possible.  It just
can't be run on a machine.  Some languages have been put to considerable
use without EVER being implemented on a real machine.  For instance, Ada
Lovelace invented quite a bit of the art of programming, but Babbage's
engine was never completed. (or not in her lifetime - somebody's written
a simulation quite recently...)
More recently (and perhaps more significantly) Messrs. Simon, Newel and
Shaw did a lot of the early research into AI, mostly writing programs in
a language called IPL (Information Processing Language).  The fact that
IPL V was never implemented didn't prevent it's being put to quite a bit
of use for expressing algorithms.
Oddly enough, IPL's demise was almost entirely due to John McCarthy's
invention of LISP...  Odd how all this fits together, isn't it? 
 SI> Even if it was done in assembly first, it was done in something else
 SI> before it was done in the 'language'. C is compiled in C. C++ can be
 SI> compiled in C++. Or in C. I'm sure LISP didn't just exist. Therefore
 SI> it stands to reason that LISP, although can be used to write LISP
 SI> interpreters, was not done that way the first time.
Quite true.  OTOH, later on MIT (as well as a couple of commercial
vendors) made LISP machines that implemented LISP in hardware, so LISP
actually WAS the assembly language.
[ the LISP defnition ]
 SI> I'd be interested in getting that definition. Do you have it lying 
 SI> around somewhere?
I thought I had it around, but I can't seem to find it right now.  If I
run across it, I'll send you a copy.  I should add that as interpreters
go, it's _horribly_ inefficient - nearly the first intepreter I ever
wrote was by transliterating it into Turbo Pascal, and _boy_ was it
slow.  Then again, considering that it was runing on a 6502 at 1 MHz,
with something like 20K of free memory, I guess that's not too
suprising.  One of these days I might have to try it again on a modern
CPU with more memory...  There's nothing like an inefficient program to
make you appreciate a fast CPU... 
    Later,
    Jerry.
... The Universe is a figment of its own imagination.
--- PPoint 1.90
---------------