COMMENT ⊗   VALID 00010 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Jim,
C00004 00003	spengler and lisp
C00009 00004	∂19-Jan-81  1028	williams at PARC-MAXC 	Re: spengler and lisp 
C00012 00005	∂20-Jan-81  2118	williams at PARC-MAXC 	Re: spengler
C00013 00006	jim,
C00015 00007	68k lisp
C00018 00008	lisps
C00024 00009	    1. Why the stack limit of 38? That is about a fifth of what one gets with
C00028 00010	∂16-Aug-81  1830	HOLLAN at BBNG 	status   
C00041 ENDMK
C⊗;
Jim,

alas, i don't ssem to be over this stuff yet, so rather than bet on getting
back here 'till monday i'm shipping a draft to you. i'm still getting break-downs
on the parts and pieces, but the outline of what i'm doing and where it's going
is in place. i've shipped the whole plan, from what's up immediately (phase I)
all the way through full-scale development (Phase III). the emphasis is on
an integrated and gradual build-up of systems and personnel, with the
immediate concern of funding the low-cost end for the spring and summer projects.

please let me know which parts-and pieces of this proposal you'd be most
interested in.  if you'd like to call i'm at 408 353-3857 or msgs to  408 353-2227

						john


spengler and lisp

i don't know whether to thank you or complain. i got the abridged  version
of "decline"  and the  first volume  of the  unabridged version.   started
reading the abridged, and  was fascinated except  that some things  didn't
ring true  (e.g.  a  section  called "antiquity  and india"  that  doesn't
mention india!) so compared the unabridged version ...quite a  difference.
so now i'm reading (and re-reading!)  the first two chapters before  going
on with the rest.

it's an amazing book, both for its  content and its style. i find it  slow
going but  almost  impossible  to put  down  for  more than  a  few  hours
(grumble, grumble, grumble; it takes up all of my time now).  the style is
almost poetic  --truly  amazing that  it  comes through  the  translation.
there is a section describing the birth-death cycle of cultures that could
have come out of  Smith's "kamongo", another favorite  book of mine.   The
content, of  course, makes  hofstatder's godel-escher-bach  insights  look
like child's play; i wonder that doug didn't reference spengler.

i'm being seduced by his argument, but since my backgrounnd is not at  all
strong in  philisophy, i'm  having a  colleague who  studied  pre-socratic
greeks, give a second opinion.  at worst, this book will give me a broader
base to hang my arguments on; at best?... well that's a bit early yet.


i've finished hacking papers to the  wccf, one is the bankruptcy of  basic
--religous tract; the  other is  the tutorial  outline: "lisp,  functions,
objects, frames,  and  constraints: functional  flesh,  functional  bones"
these will turn into the santa clara courses and a book.

unfortunately, the money  for the interactive  programming lab is  getting
further away,  and the  info  world review  of  tlc-lisp doesn't  help  my
finances (or mood!)  much. their "lisp expert" doesn't know lisp or how to
read: i'm "pleased" to  discover that my lisp  compiles its programs,  i'm
"perplexed" to discover that i  only have 16-bit integers; i'm  "dismayed"
to learn that the  lisp world still wants  progs even though  do,catch,and
throw are there plus &aux-ies  and let (they could  only find let); and  i
really wonder why one needs strings longer than 256 characters in lisp for
such a tiny machine.

sigh.

to both of you: that  was a very pleasant lunch  the other day; i  reallly
appreciate finding out about spengler.

mike: --do you know if the unabridged version of "decline" is available in
paperback? i'd like it for the class).

jim: --any progress on the "rfp" front?

i'd better go hack course notes before the machine fills up.

john
∂19-Jan-81  1028	williams at PARC-MAXC 	Re: spengler and lisp 
Date: 19 Jan 1981 10:23 PST
From: williams at PARC-MAXC
Subject: Re: spengler and lisp
In-reply-to: JRA's message of 16 Jan 1981 0418-PST
To: John Allen <JRA at SU-AI>
cc: williams
--------------------------------------------
John,

	You force me to display my ignorance.  I've only read/seen the
abridged version.  But do you really think that novice computer scientists will
want to tackle the whole thing, even a substantial part?  You have greater hope
for the world than even I.

		Mike

mike,

i'm sure some novice computer scientists will want read spengler.   that's
one of the hopeful  signs in the  field: it's more  than a pure  technical
exercise like  much  of engineering.   the  difficulty with  the  abridged
version is the removal of much of the mathematically relevant material (as
well as short-changing much of his poetry  --he really is a poet) i  don't
believe a lot will want to read it  but, of course, a lot of people  don't
know what's  good for  them.  consider  the people  who think  "lisp is  a
special purpose list-processing language".

i've bought my own copy of vol 1, and will have to order vol 2 when i  get
some bucks.  his work puts some perspective on my rantings and ravings.  i
want to  be  able  to  talk  accurately about  his  work  for  the  spring
undergrad. course at santa clara --"the art of computer science". the real
challenge will be trying to relate  computing and spengler to the  faculty
this summer; i hope that comes off.

∂20-Jan-81  2118	williams at PARC-MAXC 	Re: spengler
Date: 20 Jan 1981 17:02 PST
From: williams at PARC-MAXC
Subject: Re: spengler
In-reply-to: Your message of 20 Jan 1981 0457-PST
To: John Allen <JRA at SU-AI>
cc: williams

John,

	From the sound of it I may want to attend a few of your lectures.  You
are making me think more of spengler and the intellectual revolution underway
today.  Maybe I'll even tackle the unabridged spengler.

		Mike

jim,

i agree, apollo looks interesting. i corresponded with d. nelson
about a year ago and he expressed interest in lisp as a possible
language for his machine. i didn't pursue it then 'cause they had no
machines, i had no money and with the conference, had no time.

compucolor (i, think) has announced a 68k system too, and appears to
look reasonable. i understand they're essentially without software.

... and there's rumors coming out of apple, and cromemco, and ... ;sigh.

i'll try running down rumors and details.

*** it might help me get some leverage on thses companies if i could
*** mention the NPRDC connection. would that be ok?

indeed, please call or net mail me if you have troubles/questions.
i'm here every morning but wed, from about 4am 'tll about noon, at
santa clara (408 984-4358) in afternoons, and home 7-9pm (408-353-3857)
with msgs to 408 353-2227.

on the other front: spengler has been a real win. makes me feel quite
inadequate to handle the santa clara faculty workshop.

					john
68k lisp

al,

i've got a few questions about the arrangements.

1. it bothers me to specify a cost of an as yet unspecified product.
will your auditor really be satisfied with such generalities?
i'm willing to say that i can develop a version of tlc-lisp for the
68k in the 90-100k range (machine and labor) --and that's a bargin. but 
if it's specified that labor is $50-60k then there's no reason for me trying
to deal for a machine with apollo. i just end up getting paid less and have
an additional committment to apollo.  if it's feasible it may be better for me to 
get a lump sum and work out an arrangement for a machine. right now
every nickel counts!

2. can we specify the arrangement such that the target code ends up
in the public domain, but the source remains tlc's?

i hate to be over cautious, but i have to come out of this in a position
to do something with the results. if i end up broke, and with the code
available to all, tlc is dead. my competitors are quite well off with
staff and pr budgets --most
people i talk with claim tlc's lisp is far superior, but it doesn't pay bills.
i have no money to  advertize, and no money to pay people to develop products
or extend the existing version; the effect is that  mu-lisp and now several
more micro lisps, are  doing rather well while tlc languishes.

what bothers me is the similarity to the cromemco contract: machines plus
money got me a product but put me badly in debt. cromemco doesn't advertise
it and i've only been able to afford one $300 ad. in the 
cromemco  case, however,
i did retain rights to the software; here i'm afraid the worse of both could
happen.

indeed, i will do the 68k lisp. i just want to
make sure that as many avenues are examined as possible in order
to come out of this alive!
lisps

jim,

i'm sorry too that things have gone so slowly. it's unfortunate that the
hardware took so long to come, and then wasn't the same as mine. it was
with shock, boardering on hysteria, that i happened to  find out that  bbn
was using 5", not 8" drives; this after sending them my version of cp/m
without solving  their problem.  so far i've absorbed my costs on this
business: free single-bank lisp, fed express bills, phones, and free
multi-bank lisp as a "beta test-site" (bbn is holding a p.o. on the multi-bank,
but i don't want to accept money until this thing works). i'm anxious to have
the multi-bank version finished; i thought you knew that it was under test
at bbn, not a product yet; unfortunately it was not until april/may that
hardwar/os configurations have been stable enough to begin testing. the
n-1 th message i had from larry stead was that they'd discuss shipping a
flopppy disc controller and a 5" disc loaner to me so i'd be compatible with
them; i heard nothing, assumed the worst (i.e. they'd given up), but
when i called them  discovered that they were up and running. i'd like to
be able to buy a 5" and controller myself to resolve this faster, but i'm
broke and in debt! on the n-th call to al, it wasn't at all clear that they
thought the multi-bank was worth the extra work. we reaally DO need better
communications.

on the bugs/sources: indeed i have agreed to sell the sources, and will help
resolve these multi-bank problems. the system is  very close to debugged, but
its been a case of priorities for me: if i don't get some funding soon, tlc
will no longer exist; i made about $9K last year (before taxes!) with about
$2K from tlc; it's pretty stupid to continue at that rate. so i've been spending
the majority of my time trying to convince prospective investors that lisp is
a good thing, that lisp/education is a good thing, that lisp/applications are
viable. unfortuantely with zero funds, i cannot advertize, cannot hire people,
basically cannot to much but bleed to death. that gets very tiresome.
if you/bbn seriously want the bank-switched version, then we should set down some
ground rules: if i spend my time doing z-80 lisp (to the exclusion of pursuit/work
on 16-32 bit lisps) i'd like to know that when it's finished i won't find you/bbn
going to another lisp supplier for next  product. in my current position, i can't
pursue both, and i think we all know that the larger address space machines are 
the ultimate lisp markets. i've watched lmi and ssmbolic eat the high end market
and now logo inc take the low end --all this happening while tlc has insufficient
funds to move-- the area that's left is the middle ground, and i'd be royally
bummed to find someone take that while i piss it away on z-80's.

				john
p.s. on a somewhat happier note, the course went rather well, but unforutnately
will have to sit out a year at santa clara at least. turns out the
engineering dean doesn't like "philosophy taught in his school" --philosophy equals
lisp, logo, and smalltalk, with a little "social relevance" thrown in, and
the humanities school is "very interested in the couurse" but just went through
a re-organization so "we'll set up a committee this fall to study the matter".
so i'm looking for a new home for the course; several schools are interested.
    1. Why the stack limit of 38? That is about a fifth of what one gets with
    regualr tlc. That is too shallow to even do a nice pp. If the stack cannot
    be made substantially deeper multi-bank lisp is essentially useless for us.

this sounds like a  growth in the size of  cpm, what i'll do is look through
the lisp library for  functions that can be flushed into lisp without
adversely affecting performance. i'll also check the size of my stack

    2. How can one establish a reasonable link between mince and multi-bank
    lisp? 
bruce roberts suggested that he might hack that; also scott layson at the
mince co., and a tlc user at LLL have said likewise. if their not progressing
i'll do it.

    3. There seem to be a number of missing i/o functions with multi-bank lisp
    (e.g. autoload facilities, type files, etc).
i know i diminished the size of the kernel for debugging purposes; perhaps these
functions didn't get back in.

    4. Why does multi-bank tlc insist on the printer being on? We just recently
    found this quite by accident.
that sounds like a cpm problem. i don't even have a printer.

    5. Why doesn't the third argument to prin0 work? We get a bad parameter print.
lisp bug, will fix  -probably wrong  bank bug.

    6. What is going on with exit? We get a BDOS err on p:select.
hat sounds like cpm; it exits on mine.

    7. Why is a garbage collect required before doing any load after the first
    one? If one forget you get an illegal character message?
lisp bug; someone's in wrong bank.

    The stack depth and reasonable mince interface are of course the major
    problems that restrict us. I would appreciate your reading of this and
    an estimate of how major the fixes would have to be. How long etc.

none of those problems are majorprojects, just frustrating for you i know.
the class preparation has been 3am 'til 10pm the last ten days, but
that's coming to an end. i've borrowed a printer so i can get listings
and get this stuff pinned down. i expect to have time this weekend to
get thsee bugs out. i'll polk the mince people about their interfcae progress.

∂16-Aug-81  1830	HOLLAN at BBNG 	status   
Date: 16 Aug 1981 2129-EDT
From: HOLLAN at BBNG
Subject: status
To:   jra at SU-AI
cc:   hollan, stevens

John,
	I am very sorry that things seem to be going so poorly for you.
The TLC experience has been quite troublesome I guess for all of us. We had
really hoped to bring-up a number of training system using the bank-switched
version on Z-80s and then move on to the 68000. Chris Riesbeck has been
working with us the summer and I had hoped he would be able to do a lot
of Z-80 work for us. Unfortunately, that has not been possible due to various
bank-switched bugs.

	You seem to be very naive about contracting and dealing with other
people. What you really seem to want is someone to support you to do exactly
what you want to do: tell the world about the joys of lisp and the beauty
of computation. All of your early messages sought funding for machines and
instruction. I tried to be clear that what we needed was a lisp on a portable
machine. The bank-switched lisp looked promising and it looked like it would
not be too major a project to move it to a 68000. The Steamer project is
a large project and we are under tremendous pressure to deliver working
systems. We needed some small system running early to be able to get some
user feedback about the systems we are developing. That is the role I saw
for bank-switched lisp. If we got that going, movement to the 68000 would
have been very natural.  The use of TLC in such a large and visible project
would have perhaps been the best advertising you could get for TLC.

	In the BBN contract I specified support for doing the 68000 lisp.
The whole BBN contract is for over 2 million dollars and it takes a great
deal of time and effort to get such contracts awarded. The award extended
much past our expected March start time. Thus, there was a long delay before
we could have had a means of subcontracting the 68000 work to you. You seem
to thing that contracting is much easier than it is. It is very tedious,
time consuming process, and frustrating. 

It was always our plan to subcontract the 68000 effort to you. During the
time we were waiting for the contract to be awarded, we experienced the
many frustrations with bank-switched TLC. Becuase of this we were reticent to
go ahead with the 68000 project. Given that the BBN people had had some
difficulty communicating with you about various bugs, some time ago I
suggested that they hire you as a consultant and get you down here to fix up
the bank-switched lisp. Your schedule did not seem to permit that. Now we get
a set of a notes in which you accuse us of attempting to "starve you". I, for
one, find such personal attacks offensive. 

Jim


-------

∂18-Aug-81  0622	JRA  	status   
To:   hollan at BBNC, stevens at BBNC
CC:   JRA at SU-AI 

jim,

this whole thing has been a comedy of errors that is not very funny.
we both seem to have "short fuses", and electronic mail seems to be a very
bad way to get things sorted out, but let me try anyway. 

the major probelem, it seems to me, is lack of communication. my notes lately
have become more and more frustrated as months roll by, equipment breaks down,
annd bills pile up. jim, i don't believe there is an effort to "starve me".
the probelm is that it is now august, with march long gone by. also long gone
by is my promise to ruth that is things didn't improve by jan 1, i'd quit this
business and get a "real" job.  i may be naiive about contracting with the 
government, but i have two-years experience and at least 2-dozen proposals to 
for-profit computer corporations, publishers, investors, and other assorted
types --sufficient to realize that getting people to say either "yes" or "no"
is a long expensive process. unfortunately all that crap has to be done 
single-handed. in that two year span i have watched good, solid citizens 
retract their word and "change their minds" about commitments to tlc.
it's an unfortunate result of missed communications in this case: my notes
got harsher because of lack of response. earlier, more rational messages
got no response; i have three kids now, no funds, and a serious
question as to whether i can attract customers for the products that i wish 
to build. lack of response imples lack of interest.

as i said at the beginning of this,
missed communicatioons are a REAL problem in this mess. your early responses
(november-jan) were glowing praise for tlc-lisp; somewhere in the midst of the
spring wasteland, something changed: by june, the project was hinged on the
multi-bank lisp --something that had not (as far as i recall) entered into the
earlier discussions. the m-b lisp was (and still is) an incomplete system.
it was never advertised as complete, only given to bbn as a "beta-test".

i must repeat (though at a reduced volume) that things HAVE to be spelled out!
names, dates and places have to be specified. e.g. is the 68k lisp "linked" to the
multi-bank lisp? if so, what constitutes acceptance of the m-b lisp?  what money
is involved in the 68k lisp effort, who pays for machines? and on, and on.
i'm indeed maive about business and contracting, but two years of blistered
fingers are beginning to sink in. the unfortunate situation is that the alternative
to trust is distrust and there is no way to  say "please put it in writing" that
doesnt sound obnoxious. its a vicious byproduct of business reality.

underlying this business is the unfortunate fact that tlc is the sole means of my
support, and i am currently to only person trying to keep things
together. as such i have  to budget my time on those things (indeed THAT thing)
that shows most promise in keeping me and tlc alive. between january and may
that appeared to be an desire on the part of santa clara university to establish a
substantial "computer literacy" program. if that had developed, it would have
given sufffient support to me and tlc to invest in other projects. given that
ti backed out very ungracefully (in case you thing this is exaggerated, talk
to mark miller or truman blocker at ti) the only "other project" was yours, jim.
if this business sounds like a juggling act, it is.  the santa clara business
fell though in the spring when they decided to reorganize the arts and sciences
school, putting and new programs off until a committee can study them --they'll
report back sometime before the second coming (they're jesuits, so they'll get
an inside tip).

indeed, education is a big thing, but you over state matters when you
said "What you really seem to want is someone to support you to do exactly
what you want to do: tell the world about the joys of lisp and the beauty
of computation" the content of the scu courses had nothing to do with your proposal.
i mentioned them to you in notes because you seemed to be interested too.
at no time was i asking you for funding for educational projects!
early on (~march) al reiterated the limited scope of this project, and i concurred.
i must admit, i was a bit suprised that the  reiteration had to be performed; i had,
and still have, no doubts about the scope --after the third or fourth time
there's much diminished "joy and beauty" in reimplementing lisp.

as to bug fixing, i have tried to get errors corrected and new disks back to
bbn. those fixes have been done at tlc cost, including the fed. express bills,
and though tlc is not solvent, i've refused payment for the multi-bank object code
(and the single-bank was gratis too). the multi-bank source has been delayed
because of  break-down of my disk drives. disk  drives in general seems to be
another whole area of contention here: who has what and who has which cpm.
the micro-market is a mess, as we've all discovered.

i don't know if this helps any. if you want to talk about it, call me;
i'll be at 408 353-2227 later today.