smalltalk and lisp (was: strangest systems I've sent email from)

[Jecel Assumpcao Jr.]

Sean Conner wrote:on Wed, 27 Apr 2016 16:13:07 -0400
>   The 6908 *is* better than either the Z80 or the 6502 (yes, I'm one of
> *those* 8-)

To be fair, the Z80 and 6502 had to compete against the 8080 while the
6809 came out after the 8086 and 68000.

>   Citation needed.  C derivatives?  The only one I'm aware of is C++ and
> that's a far way from C nowadays (and no, using curly braces does not make
> something a C derivative).

Objective-C was the only other C derivative to have a significant
impact. CTalk and a bunch of others were mere historical curiosities.

>   While the Canon Cat was innovative, perhaps it was too early.  We were
> still in the era of general purpose computers and the idea of an
> "information appliance" was still in its infancy and perhaps, not an idea
> people were willing to deal with at the time.  Also, how easy was it to get
> data *out* of the Canon Cat?  (now that I think about it---it came with a
> disk drive, so in theory, possible)  You could word process, do some
> calculations, simple programming ... but no Solitare.

The Cat could execute any Forth code embeded in your text document, so
it could have had a Solitaire. It was extremely well received by its
intended users: secretaries. It was killed off to stop a dispute between
the chiefs of the computer and typewriter divisions of Canon and not due
to any commercial reasons.

>   To be fair, *everybody* missed the essence of what they did at PARC; even
> Alan Kay wasn't of much help ("I meant message passing, *not* objects!"
> "Then why didn't you say so earlier?" "And Smalltalk was *never* meant to be
> standardized!  It was meant to be replaced with something else every six
> months!" "Uh, Alan, that's *not* how industry works.").

You are paraphrasing the "burn the disk packs" discussion at a Learning
Research Group retreat. Alan Kay wanted to throw everything away and
create a new end user programming system. Dan Ingalls wanted to clean up
Smalltalk-74 even if it became harder for children to use it and then
implement a system for children on top of that (see Scratch and Etoys).
Alan let Dan do it his way (Smalltalk-76), and even supported the
attempt go from research to development in the Notetaker project
(Smalltalk-78) even though that is not his thing.

When an executive from Xerox headquarters flew in just to kill the
Notetaker ("Xerox doesn't do microcomputers", he said) Alan left for
Atari and the rest of the group decided to create Smalltalk-80 as a
software-only product to run on other company's machines. I would say
that at the time Smalltalk-80 was the most standard language out there
along with UCSD Pascal and with its set of books it was the best
documented one.

>   The problem with the Lisp machines (from what I can see) is not that they
> were bad, but they were too expensive and the cheap Unix workstations
> overtook them in performance.  Had Symbolics and LMI moved their software to
> commodity hardware they might have survived (in this, Bill Gates was
> right---it was software, not hardware, where the money was).

Symbolics moved all its stuff to the Alpha processor, but that was the
wrong choice. Note that there were software-only Lisps for workstations
like Franz Lisp.

Some people, when they hear that I develop Smalltalk computers, like to
inform me that there used to be Lisp machines but they failed (for some
reason they seem to think I wouldn't know that). Their conclusion is
that language specific computers are a bad idea. My reply is that all
the generic computers except two (PC and 360) have also died, and one of
the survivors is such a tiny niche.

>   Smalltalk has other issues.  In the 80s, there were not many machines
> capable of running Smalltalk (I'm not aware of any implementation on micros,
> serious or not) and by the time workstations where becoming cheap enough,
> the Smalltalk vendors were charging rediculous amounts of money for
> Smalltalk.

As others have pointed out, Apple sold Smalltalk-80 for the Mac.
Unfortunately you had to be a registered developer to be able to buy it.
A reduced version could run on a Mac 512 but for the full thing you
needed at least 1MB. As Smalltalk-80 accessed the hardware directly, it
couldn't deal with ADB keyboards (but the mouse worked for some reason).

Digitalk came out with Methods for the PC (with at least 512KB) in 1985
and Smalltalk V in 1986. It soon had a Smalltalk V/Mac (which is why
Apple stopped selling their own Smalltalk and didn't updated it for
newer Macs), Smalltalk V/286 (for a long while the only language to be
able to used more than 640KB on a PC) and eventually Smalltalk V/Win.

ParcPlace was spun off from Xerox in 1988 and focused on workstations
with extremely expensive products (which continues to this day - see
Cincom). By the early 1990s there were some more nice Smalltalks for PCs
(Object Studio, Dolphin, Smalltalk MT) and Macs (SmalltalkAgents).

>  The other issue is the image---how do you "distribute" software
> written in Smalltalk?  Smalltalk is the application is Smalltalk, so you
> have to buy into the whole ecosystem (perhaps in the 90s---today, probably
> less so).

This certainly wasn't the case for Smalltalk V/Win, but other Smalltalks
had solutions for "stripping" and image.

Alan Kay makes an interesting comment about Lisp on page 524 (page 15 of
the PDF) of "An Early History of Smalltalk":

http://www.vpri.org/pdf/hc_smalltalk_history.pdf

He had learned Lisp at Utah, but nobody there used it except for
teaching so he only learned about CAR, CDR, CONS etc. Only when he spent
some time at SAIL (Stanford Artificial Intelligence Laboratory) did he
finally get the idea of Lisp and what EVAL and APPLY were all about. I
would not be surprised if many students who had Lisp in school and hated
it never really got it like Alan hadn't.

-- Jecel