Xref: utzoo comp.lang.lisp:3302 comp.lang.smalltalk:2009 comp.lang.scheme:1354 comp.sys.ibm.pc.programmer:1953 comp.unix.i386:5857 comp.lang.forth:3097
Path: utzoo!attcan!uunet!munnari.oz.au!uhccux!dunn
From: dunn@uhccux.uhcc.Hawaii.Edu (John Dunn)
Newsgroups: comp.lang.lisp,comp.lang.smalltalk,comp.lang.scheme,comp.sys.ibm.pc.programmer,comp.unix.i386,comp.lang.forth
Subject: Re: interface/graphics development environments for 386/RS6000
Summary: You might consider Protected Mode Forth from LMI
Message-ID: <8167@uhccux.uhcc.Hawaii.Edu>
Date: 13 Jun 90 19:31:08 GMT
References: <835@anaxagoras.ils.nwu.edu>
Reply-To: dunn@uhccux.UUCP (John Dunn)
Followup-To: comp.lang.lisp
Organization: University of Hawaii
Lines: 46

In article <835@anaxagoras.ils.nwu.edu> cearl@aristotle.ils.nwu.edu (Charles Earl) writes:
>We will be developing multimedia (e.g. video, animation, digitized images,etc.)
>based educational software using ibm RS/6000s and 386's as development 
>platforms -- assume UNIX environment. Initial development will most likely
> occur on the RISC machines but
>the 386's are intended target (we will actually be using 386s for some months
>until RISCs arrive).
>
>We have been a MacII shop, doing most development in Mac Common Lisp.
>I am trying to decide on an appropriate development environment and
>would like feedback from the experts. We would like to use
>Lisp/Scheme, but Smalltalk, C++, C, etc. would be ok if no other
>options for graphics environments exist.
>
If you are most comfortable working in an interactive Lisp environment,
but you need to produce real-world working code, you might consider
Laboratory Microsystem's 386/UR Forth.  It runs in protected mode,
has a first rate virtual memory facility, and is faster then C by
a fair degree.  Several times I have gone to assembley language for
loop-critical components, only to find I had a 2* savings - the speed
of this product is difficult to believe.  Compile speed is likewise
considerably faster than you are likely to see with C or Pascal - 
especially when you include link time.

I believe LMI has a compatible version that works under UNIX.  They
definitely have one that works under OS/2, and have made noises about
bringing out a Windows 3 version.

My own experience was similar to yours - I had worked in a Lisp
developemt mode until there was enough groundwork to take the project
to a real-world releasable version.  After close to 6 months of false
starts, I tried out 386 UR/Forth.  The only thing that wasn't available
was some means of treating data as objects - I had become use to 
Flavors in Lisp.  Eventually, after an exaustive search of the available
Forth code (there is a mountain of it, but it is mostly unusable IMHO),
I gave up, bit the bullet and wrote a object-type memory manager
with very fast automatic garbage collection.  This whole system has 
worked so well for me that now, even if there were to be a Lisp
package around that included a reasonable means of delivering the
finished software to the end user, I would stick with 386 Forth.

Oh yes, the Memory Object Package is in the Public Domain - You will
find it on the UR/Forth BBS system that you will have access to when
you buy any of the UR/Forths.
 
-John Dunn