Re: Monthly FAQ Posting

[Tom Gordon <thomas@hera.gmd.de>]

> From: marykuca@sol.UVic.CA (Brent Marykuca)
> Subject: Re: Monthly FAQ Posting
> Content-Length: 2215
> 
> >This is a surprise and a disappointed.  Am I mistaken, or didn't Scully
> >announce Dylan at the same time as Newton, and specifically claim that 
> >Dylan would be the programming language for Newton?
> 
> I am certain that this is not the case.  I'm sure I read nearly every news
> release and promotional blurb about Newton and never saw any mention of
> Dylan.
> 

I may have been mistaken about Byte, but the German language MACup magazine's
report on Newton was very clear (MACup, September, 1992, page 58):

"Dylan -- Newton's Programming Language

Newton's operating system and applications are specified in a new object-
oriented dynamic language called Dylan ..."

> 
> >From things that I have read around the net and elsewhere, Newton will be
> released soon, i.e. within the next three or four months.  From what I read
> here, Dylan, while largely specified, is still brand-new.  It seems
> unlikely that Apple would base a "revolutionary" technology like Newton on
> (sorry, everyone) an untested and unproven language.
> 

I too had wondered how it is possible that Newton would be released before 
Dylan is finished, but we do know that Apple has several implementations
of Dylan and that they are trying to assure that changes to the language
are minimal and upwards-compatible.  Also, perhaps the current Newton software
really is only "specified" in Dylan, but implemented in, perhaps,
assembly language.  Finally, I do not agree that Dylan is "untested and 
unproven", as it is a conservative consolidation of well-tested design
concepts from Scheme, Smalltalk and CLOS.  There is nothing revolutionary
about Dylan, except Apple's intention to really commit to a Scheme-like
language for a product like Newton (if this would only be the case).

> 
> I don't want to get into a discussion of Newton in this forum, except to
> say that it would probably go nowhere if you could only program it in
> Dylan.
> 

Just as there are a multitude of programming environments for the Mac, 
even though Pascal was originally chosen as its native language, so
to would I expect there to be other alternatives for Newton, at least with
time.   

--------------

Aside from the non-authoritative report in MACup, there is considerable
circumstantial evidence supporting the claim that Newton and Dylan *are*
very well related:

1) Larry Tesler, an Apple Vice-President, wrote the forward to the Dylan
book, and just also happens to be one of the leaders of the Newton
group (also according to MACup).   Larry was also one of the prominent
members of the Smalltalk group at Xerox before going to Apple, and was, if
I remember my history correctly, the father of the Dynabook idea, which
lead to Newton.   Smalltalk was a central part of the Dynabook idea, so
it would be natural for Dylan to play the same role for Newton.

2) Apple's choice of the ARM610 processor for the Newton was motivated by,
among other factors, the need for on-chip support for garbage collection.
Why, if programmer's are going to be responsible for memory management,
using languages like C++?

To sum up, I do not want to continue to spread rumours, which I didn't
start, about Dylan's role for Newton.   I have no idea why Apple is 
hedging its bet here.  Perhaps there are some internal wars at Apple
between Dylan and C++ factions.  Who knows.  But I am concerned and
somewhat worried.   It would be comforting to hear some announcement soon
about just what role Apple intends Dylan to play in future Apple
products.

Tom Gordon
GMD