COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	larry masinter  said  something  about  talking  with  you  concerning  my
C00006 ENDMK
C⊗;
larry masinter  said  something  about  talking  with  you  concerning  my
proposal for an interlisp subset on 820-like machines.

after stuffing a  healthy maclisp into  a z-80, it  seems (a)possible  (b)
useful, and (c) slightly  insane, with (a) and  (b) somewhat ahead of  (c)
for the moment.

the thing that drove me to the z-80 is that it leads to cheap system,  and
since i want to destroy the  basic/pascal disease that is being  inflicted
on high-school (and younger) children, it was important that lisp  systems
be inespensive (i've got several long harrangues dealing with how to teach
mathematics and cs to h.s. students using a lisp base, and how to teach  a
"no-bullshit" literacy  course based  in  lisp to  college  undergraduates
--the  latter done at santa clara university)...anyway, the 820 looks like
a reasonable start, and  if it gets bit-mapped  graphics, it will be  even
better; then of course, there's the issue of color.... a bit further  down
the line  is the  z-800 (not  8000) --the  3-to-5x z-80A  with an  on-chip
mapper to a  4MB address  space. I'm  ready for  that one  too, now  using
bank-switching on the z-80  to get 32K-32-bit lisp  objs, plus room for  a
lot more non-lisp stuff.  the bank-swithing is embedded in one macro;  the
mapper-chip hack will probably be trivial.

the reason for raising  the issue now  is that if  enhancemnts are in  the
workds for the  820, then i'd  like get  in the picture  early (e.g.   i'm
guessing that they didn't  consider bank-swithcing or extended  addressing
in the design  --"why would anyone  want a thing  like that?"  the  answer
obviously is LISP!) but who in their righ mind would do lisp on a z-80?

that, of course is issue (c):  insanity. clearly a better vehicle for lisp
(or almost anything  else!)  is  a larger  address space  than these  dink
machines, but the  response is  still cost. until  the ≥16-bitters  become
cheaper, there's life.  of,  course if there's a  cheap 16-bit 8xy in  the
near horizon ...

i mentioned two issues in a note to pahlavan (1)820 "lisp evaluation kits"
for prospective 100 customers.  (2) possibility of networking to an  1100.
the issue  that interestes  me  (and perhaps  you)  is the  potential  for
inexpensive lisp-based aaplications on these machines.