Re: Lisp Machine considered paragon

[joswig@vampire.lavielle.com (Rainer Joswig)]

>The Lisp Machine does not have a fancy interface.

Depends on what machine you look at. The Xerox LMs have had
a more graphical interface. It also depends on the software you look
at. KEE (Knowledge engineering environment - from Intellicorp)
has a interface with lots of graphical browsers (which are quite
good even by todays standards). Plexi (a neural network system)
even had a Mac-like interface.

Browsing through my old KEE 3.0 manuals (which very quite good)
from 1986:
- completely based on Frames
- rich graphical user interface
- multiple screens (like the virtual window managers under X)
- bitmap editor
- graphics editor
- interactive entry of frames
- frame browsers
- query language
- browser for worlds (tracking of versions of objects)
- object-oriented rulesystem with forward and backward chaining
- pictures (Arcs, Axis, Bitmaps, Dials, Splines, Thermometers, ...)
- active images (monitors, state-dials, histograms, ...)

The Symbolics machines have been used quite a bit for development
of innovative user interfaces. Most of it is lost and dead.

> It actually looks quite dated

Not much when you look at Unix command shells. Unix weenies would kill
for the Listener on the Symbolics. If they only knew. ;-)

> , and it is slow relative to more modern environment.

Well, the later machines were usable. Speed range is from
68030 to 68040 machines.


>And don't believe that I think the Lispm's are perfect- they
>are not.

True.

>Indeed, there are modern programming concepts which are
>real improvements.  It's just good to know what is progress and what is
>catch-up.

A friend of mine has just evaluated C++ environments and some libs
(for the Mac - threads, code fragment manager, MacApp, ...) he
may need. Despite the fantastic things they promise - a closer look
reveals the truth. I'm glad that I haven't to earn money using these
"tools". These environments are so "static" and full of
incompatibilities. Progress is so slow. After talking with him
a lot about this side of software development (nearly everything
under MPW, Symantec, Metroworks (not that bad)) I can give you
people one advice: If you can stick with MCL and can wait until PowerPC
MCL will be released, then do it.

On the other hand you have environments like "Visual Age" (SmallTalk)
from IBM. SmallTalk is very trendy nowadays (again!). The needed
memory size is enormous. Delivery options??? The environment and
various libraries also are very expensive.

Given these realities, the position of MCL is not that bad:
- moderate memory footprint compared to other development environents
- reasonable fast (on 68k)
- good OS integration
- fast development cycle
- innovative users

Still, improvements are possible:
- native PowerPC compiler  (can't wait for this one)
- C++ support
- delivery option
- better libraries
- better code management
- editor

Another thing I would like to see is support for SOM (and DSOM). SOM
enables the use of the same objects from different applications possibly
written in different languages. DSOM extends this to distributed
programs. SOM will play a very important role in future programming
on the Mac (since OpenDoc uses SOM). It is already available for
other platforms (examples are AIX and OS/2 and more to come). SOM
support does exist atleast for C, C++ and SmallTalk. MCL should be able
to use SOM (Dylan will/does too, as I understand).


Rainer Joswig

Rainer Joswig, Lavielle EDV Systemberatung GmbH & Co, Lotharstrasse 2ab, D-22041
Hamburg, Tel: +49 40 65808-0, Fax: +49 40 65808-202, Email: joswig@lavielle.com,
WWW: http://www.lavielle.com/~joswig, Tel. (priv): +49 40 7355522