OOP isomorphically relating ARPANET

John Hinsley jhinsley at telinco.co.uk
Sun Nov 18 04:29:26 UTC 2001


> Gary McGovern wrote:
> 
> On the 9th October 2001
> Alan Kay wrote:
> 
> >However, I'm not an exclusivist. I think people should be able to run
> 
> > anything they need side by side and interoperably on their computers
> 
> > regardless of where they came from. That was one of the main points
> > behind my particular conception of OOP (and how it would
> > isomorphically relate to the ARPAnet) back in the 60s.

Where does Alan say this? I can't find a message on the list for that
date which corresponds (which certainly doesn't mean it doesn't exist!).
Generally, I'd like more context before I give it a guess.


> Does anyone know if OOP being isomorphic to the ARPANET would mean
> that an object would be like a node in a network ?
> 
> If so would that mean ideally, objects should follow the distributed
> network model devised by Paul Baran and have redundancy in their
> interconnections? 

Well, let's remember that Paul Baran and Donald Davies developed the
idea of distributed networks independently (although Davies credits
Baran with getting there first) and with completely different ends in
view, and that no-one at ARPA seems to have so much as heard of Baran's
work until late 1967, two years after it had been abandoned, and then
only through one of Davies' colleagues!

Also that in the mid '70s, independent commentators who had probably
never heard of Baran, Davies, ARPA or RAND were talking of the
redundency inherent in the existing, circuit switched phone network in
the UK in the event of nuclear attack (Baran's main concern, but
something which Bob Taylor maintains -- and I'm inclined to believe him
-- was never anywhere on ARPA's list of priorities.)

Surely, there are two scenarios here:

1.	Objects which communicate on one computer.

2.	Objects which communicate over a network (which pretty much means
TCP/IP).

In case 1, redundency (of messages) is pretty well a waste of time as a
global event (cleaner pulls plug out, operator spills orange juice on
motherboard) is likely to take the whole thing out. Of course, we might
want to build in redundency in the form of RAID, hard wired power supply
and back ups, but this at a system level).

In case 2, redundency is already handled by TCP/IP given that we
implement the right network typology.

//snipped//

Cheers

John
-- 
Reputed to be the reason Windows 2000 was nearly a year late, (paid in
shares M$Ds needed the cash and kudos) Netproject's Eddie Bleasdale has
renewed his challenge to virus writers. The first person to infect his
Linux box wins 10,000 pounds.

http://www.silicon.com/bin/bladerunner?30REQEVENT=&REQAUTH=21046&14001REQSUB=REQINT1=48211




More information about the Squeak-dev mailing list