Newsgroups: comp.lang.dylan
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!hudson.lm.com!netline-fddi.jpl.nasa.gov!elroy.jpl.nasa.gov!swrinde!howland.reston.ans.net!news.sprintlink.net!noc.netcom.net!ix.netcom.com!netcom.com!netcom4!haahr
From: haahr@netcom.com (Paul Haahr)
Subject: Re: Dylan in the News
In-Reply-To: lyman@cc.gatech.edu's message of 10 Jun 1995 21:41:50 -0400
To: lyman@cc.gatech.edu (Lyman S. Taylor)
Message-ID: <HAAHR.95Jun11104458@netcom4.netcom.com>
Sender: haahr@netcom4.netcom.com
Organization: NETCOM On-line services
References: <3rajcl$f3p@martha.utk.edu> <3rcjq1$jjr@newsbf02.news.aol.com>
	<3rdhku$495@pravda.cc.gatech.edu>
Date: Sun, 11 Jun 1995 09:44:58 GMT
Lines: 45

Lyman S. Taylor <lyman@cc.gatech.edu> wrote:

> Actually it's not "wool-gathering".  It is basically asking a implicit 
> question.  which is
> 		"Isn't most ( or significant fraction ) of the current
> 		 'proto-Dylan' implementation written in Dylan?"

I think Apple has been pretty public in saying that it was written in
MCL (Macintosh Common Lisp).

> If it is not then .. why weren't they writting it Dylan?

What development environment were they supposed to use?

It's the usual bootstrapping problem:  You can't build the first
implementation of X in X, because there's no X implementation to do it
with.

>   [ the expression  "make them eat their own dog food" comes to mind.
>     building a tool that isn't used by the tool builders themselves often
>     results in  "less than optimal" tools. ]

My guess is they'd rather like to work in Dylan.

I wasn't at the WWDC demo, and the videotape I saw didn't include the
Q&A session, so I'm not going to contradict someone who remembers a
someone saying that Apple would recode Apple Dylan in C or C++, but if I
had a substantive body of MCL code that I wanted to rewrite in a
language that made promises of being faster than CL, I'd pick Dylan
over C++ because Dylan's model is closer to CL's.

> [...]
> In short I don't think that you could or even should write the whole
> environment in Dylan ... down to the very last line.  In fact there should
> probably be some c/c++ and even still signficantly smaller portion of
> assembler ( or crufty C code that resembles assembler ). 

I'd hope that wouldn't be necessary.  Certainly no more often than a C++
development environment should drop into assembler.

Paul

(I guess I should mention that I work for Harlequin.  This note is my
personal opinion and speculation only, not a statement by my employer.
I couldn't make one of those if I wanted to.)
