It seems suitable to bring this discussion here...

To bring other ppl up to speed, Matt and I have been talking about
potentially picking up the Agent Perl project again.  In general, we've
been talking about architectures, and technologies available.  In
particular, there's a lot of resources available at:

	http://www.fipa.org/
	   - ACL, architecture, example systems

	http://www.w3.org/2001/sw/WebOnt/
	   - OWL, DAML+OIL derivative


On Thu, 15 Aug 2002, Matthew Lawrence wrote:

> > [snip!]
> >Just briefly, I think an Agent Perl project could be
> >split up into 2 broad areas already - server infrastructure, and
> >communications tools (& examples).  But I'll leave that 'till later.
>
> Do you see that like a two level thing, with the server infrastructure 
> providing an API to the communication tools? Or are they side by side?

I see the server not really giving a rats ass about anything higher
than the transport layer.  So it wouldn't know about the
ACL/OWL/perl/english/etc. messages that it let agents send.  That way
seperate API's could be written for ACL/OWL and release on CPAN...

Of course, to go with the spirit of ACL & collaborating agents, you'd have
to write some standard agents that know about ACL/OWL, and act accordingly
(like getting an 'ask-all' ACL message, and actually asking all its
contacts.  or getting a 'host-me-pls' PERL kinda message, and hosting a
mobile agent).

So maybe writing a set of example agents would be the third task that
glues it all together.

+--
 Steve Purkis <spurkis@epn.nu>