Yes, the Object idea was "very like Actors" that Hewitt later introduced.
As I mentioned in the "Early History of Smalltalk", a variety of influences --
biology, abstract algebras, the no-centers networking that ARPA was planning
for the ARPAnet, the process architectures starting to appear in time-sharing,
etc. -- gave rise to the desire for a software entities that acted like whole
computers, had very little overhead, could do loose coupling via
pattern-matching and message passing, etc.
Our main drive at PARC was to invent and make a version of "personal computing"
that had a kind of universality to it.
We experimented with more mechanisms than we generally used (for example with
highly parallel simulation control and scheduling subsystems) but in the end
the Alto did not have enough size and horsepower to deal with the additional
overheads needed to do things like this (and some of Fisher's other control
structure ideas). The original plan called for a next gen computer after the
Alto, but Xerox wasn't willing to fund it for quite a few years. This forced
everyone to do optimization for their next-gen software and pretty much removed
our invention hats.
However, the other abstraction mechanisms and the low overheads for Smalltalk
objects -- via Dan Ingalls' and others' brilliant touch -- were enough to allow
the big inventions to get done and built.
Cheers,
Alan
>________________________________
> From: Miles Fidelman <mfidel...@meetinghouse.net>
>To: Alan Kay <alan.n...@yahoo.com>
>Cc: Fundamentals of New Computing <fonc@vpri.org>
>Sent: Wednesday, April 11, 2012 12:27 PM
>Subject: Smalltalk & Actors?
>
>Hi Alan,
>
>Apropos some of the recent threads:
>
>As I recall, some of the very early Smalltalk versions had a more concurrent
>view of the world, and inspired Hewitt's work on Actors, as is now perhaps
>best embodied in Erlang.
>
>It's long occurred to me that I sure would love to have an environment that
>felt like Smalltalk (say Squeak) but that allowed objects to behave like
>actors - all the message passing is there, but message/event-driven flow of
>control isn't. It's always seemed to me that a merger of Smalltalk and an
>Erlang-like run-time environment would be an interesting direction - but I've
>yet to see any efforts toward concurrent or distributed smalltalk go very far
>(well, maybe Croquet qualifies).
>
>I wonder if you might have any comments to offer on why Smalltalk took the
>path it did re. flow-of-control, and/or future directions.
>
>Regards,
>
>Miles Fidelman
>
>-- In theory, there is no difference between theory and practice.
>In practice, there is. .... Yogi Berra
>
>
>
>
>
_______________________________________________
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc
