Tim Rowledge wrote:

The original ARM development machines (the "silver fox" desktops) were ARM1 and then many were hand upgraded to ARM3 when they started rolling off the lines a year or so later. I had Little Smalltalk running in a couple of days on the ARM1, but you can't call it any major triumph of engineering since it was so utterly trivial to do.

Thanks for the details! I had thought you were talking about your Active Book or RiscPC projects, but I suppose this was a bit earlier.

A modern equivalent, and much more useful, would be Spoon.

Indeed, though for some applications it might be interesting to add a text interface like Little Smalltalk and GNU Smalltalk have. Actually, the one from Self was a little nicer (we only got a GUI in Self 4.0).

Another Smalltalk that was created for this kind of system was a very stripped down Squeak called Fly Smalltalk. I saw a poster about it at OOPSLA 2003 but I never heard anything more about it.

http://dl.acm.org/citation.cfm?id=949418

Fly had 29 classes and 170 methods, while Little Smalltalk 3 has 53 classes and 559 methods. Little Smalltalk 4 has 50 classes and 468 methods in the basic image plus a bunch of .st files that add 21 more classes and 140 additional methods. But LST4 does in Smalltalk stuff that LST3 does in the VM, so the very similar sizes is somewhat misleading.

Recently I found the 1973 paper by L Peter Deutsch which was the starting point for the InterLisp virtual machine, "A LISP Machine with very compact programs"

http://dl.acm.org/citation.cfm?id=1624860 http://ijcai.org/Past%20Proceedings/IJCAI-73/PDF/076.pdf http://www.softwarepreservation.org/projects/LISP/interlisp-d/Deutsch-3I JCAI.pdf

This would have been extremely helpful to me when I was doing a children's computer in 1983. There is always some more history to learn and make things more interesting in the present.

-- Jecel