Partially -- it was influenced quite a bit (as I remember) by the various
regcognizers in the PARC Smalltalk, which derived many of their main ideas
from Groner and Ledeen. In particular, though the Groner recog was
originally set up (1966) to be a multiple stroke recognizer, the RAND
people quickly realized that doing everything single stroke would be a big
win for speed and smoothness, and GRAIL (as far as I remember) did
everything single stroke. I used to show the GRAIL movie at PARC for
(Dynabook) inspiration. It was impressive, and in many ways Graffiti is a
pale shadow in comparison. The Smalltalk recognizers at PARC were a great
demo about how this stuff could be integrated into a comprehensive system.
Tablets were so expensive then that it remained a demo.
     The recog in Squeak uses my modifications on Gabe Groner's scheme --
which turns out very pretty in a language like Smalltalk. It was done
partly as a benchmark for the earliest Squeak, and partly in anticipation
of the cheap tablets and available PDAs with styli. It needs a UI for the
dictionary, and to be carefully trained. The dictionary supplied with the
standard release is simply a collection of example characters, and doesn't
include the entire set.

P.S. I don't think the Xerox patents will hold up, given the wealth of
prior art from the 60s ....

Cheers,

Alan


At 7:56 PM -0800 1/11/00, Russell Allen wrote:
>Hi
>
>I just read that Xerox is suing Palm based on a claim that Graffiti
>infringes a patent Xerox holds on a single stroke character recognition
>system developed at Xerox-Parc:
>
>http://dailynews.yahoo.com/h/nm/20000110/tc/tech_com_1.html
>
>http://www.patents.ibm.com/cgi-bin/viewpat.cmd/US05596656__
>
>Is this the same system used in Squeak?
>
>Russell
>
>----------------------------------------
>Russell Allen
>
>russell.allen at firebirdmedia.com
>
>----------------------------------------