On 13-Oct-07, at 12:17 PM, Brad Fuller wrote:


> can you tell us more about the Active Book you worked on 20 years ago?

Oh my, most of the old-timers will know that I can go on about that  
for hours. Days!

OK, try to keep it short tim -

The Active Book was a neat idea (especially for 1988) to make a  
portable tablet like computer with a digitiser screen, lots of power  
and a decent user interface. In short, someone's idea of a dynabook.  
The UI concept was to emulate a book but extend the idiom in the ways  
that a dynamic system could allow. So, documents were to be on pages  
and flipping pages would move between docs and the systems running  
them. No real use for overlapping windows - they were a foreign  
concept to almost everyone outside a limited world back then. The  
idea of the book was really to provide a concept of a contained world  
with logical separation between parts; just like a book has  
everything between the front and back pages and separated into  
chapters etc. I suspect it wouldn't really seem so neat these days  
now that we're all so used to Windows and OSX etc.

The hardware was about the size of a MacBook, a bit thicker since the  
practical batteries back then were rather large cylinders. It had a  
whopping great big screen of amazing 640 by 400 resolution,  
monchrome. There was  stylus based digitiser and a panel to one side  
(can't remember which) that was a finger activated digitiser,  
intended for 'other input'. I'm pretty sure it was Bill Buxton that  
convinced them of the usefulness of that idea.

It had an ARM2aS cpu running at 4MHz (maybe 6?) and 1Mb ram for  
everything - screen memory, fax filing, OS, the lot. Oh and 1Mb ROM  
too. The ARM was custom designed in-house to extend the basic ARM  
architecture to allow implementation as a static logic part, meaning  
we could simply turn off the power and it would sit there until it  
was time to go again. For cpu history buffs, the ARM 2 series had no  
cache; none at all. Not even an instruction prefetch queue. It was  
the first broadly used RISC machine (yes I know about the Sparc) and  
required  whopping great 25,000 transistors worth of silicon. In fact  
we used to joke about '25,000 transistors and no Gates'. Even the  
latest ARM cores need barely 50,00 transistors I'm told. Despite the  
above, the Acorn desktop machines that were on sale back then and  
using the base ARM2 cpus were several times faster than the hottest  
intel based boxes of the time.

The OS was a version of Helios (eek, I still have a copy of the  
Helios manual in my bookshelf!) which I think started out life as  
TripOS at Cambridge (the real university, not the US) as a multi- 
processor system using messaging, or something like that. It was  
wrapped in a unix-like layer to please the unix-weenies.

On top of that was Smalltalk. The VM was derived from EliotMiranda's  
famous BrouHaHa with a lot of extension for multiple non-contiguous  
memory spaces, ROMability, survivability, extra prims, blah blah  
blah. One of the acceptance tests for releases was to use up enough  
memory to get down to a couple of hundred bytes left and see if it  
could open a notifier, let the user kill the process and recover to  
full usefulness. Try doing that in most modern systems.. The image  
was licensed from ParcPlace, probably through Smalltalk Express  
though who knows what went on in that particular liaison? A small  
number of smart people worked on the image to make it do what was  
needed and some really interesting stuff was written. What really  
seems amazing to me now is that even with such a puny cpu and tiny  
amount of memory the UI was very snappy. You could for example draw/ 
scribble on the screen and produce a drawing (not a bitmp type  
scribble) that kept up with the scribble. Then choose the eraser and  
scribble with that to erase *sections* of the lines. I bet they were  
even the correct sections! Remember, this is a cpu with less power  
than the one in your keyboard, probably with less memory than in your  
keyboard, running a fairly simple interpreted Smalltalk VM. You can  
see why I've always laughed hysterically at people that tried to  
claim Smalltalk is slow.

Anyway, we worked on this for a couple of years and the prototypes  
were getting good reviews from early testers. Apple was peripherally  
involved with the ARM since it had decide the Hobbit/CRISP that had  
been intended for what became the Newton was no good; around then  
Acorn, Apple and VTI formed ARM  Ltd. Interestingly enough a little  
company in silicon valley decided to use the Hobbit for their attempt  
at a tablet; eventually they realised that the Active Book was much  
better and bought out the owners. And just junked everything. Oh,  
yes, that was ATT, who had created that dreadful thing called  
PenPoint and the tablet that ran it. It never went anywhere. I had  
one for a while. Dreadful; slow as... well about as slow as a lot of  
modern machines when they haven't enough memory, despite their  
adverts claiming how it was so fast because of the carefully hand  
optimised code.

Oh, we should note that in the same time period there was another  
tablet machine around that happened to be based on Digitalk's  
Smalltalk; the Momenta.

tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Useful random insult:- IQ = dx / (1 + dx), where x = age.