[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
RE: [hyades-dev] Notes from HCE protocol meeting 7/29/04

The check before killing could be useful.  The whole thing raises the
issue of what happens if at some later time the agent decides it is
ready to go away.  If the agent has to get a reference to itself, it
could release that reference.

One of the other situations I was thinking about that didn't make the
minutes (and which, therefore, I must not have expressed very clearly in
the meeting) is the case where an agent which may exist in multiple
instances wants to keep 'n' instances available (most likely with n=1).
So if, for example, three instances of the agent are running and then
all are released two would be shut down but one would be left running.
The HCE would have to be the one to keep track of this since we don't
want agents worrying about multiple instances of themselves.

Maybe that's just a hypothetical case that wouldn't come up in reality,
but I wanted to throw it out for discussion.

-Andy

-----Original Message-----
From: hyades-dev-admin@xxxxxxxxxxx [mailto:hyades-dev-admin@xxxxxxxxxxx]
On Behalf Of Barr, Dan
Sent: Thursday, July 29, 2004 11:37 AM
To: hyades-dev@xxxxxxxxxxx
Subject: RE: [hyades-dev] Notes from HCE protocol meeting 7/29/04

>>Someone brought up the reference counting of agents. They suggested
that 
some agents, once started, don't want to be stopped when the last client

disconnects....<<
Yes, I was the original troublemaker on this one -- sorry ;-)

Had a thought after our call: I'm not sure who originally suggested
putting
the "don't kill me" attribute in the agent metadata (probably Andy), but
it
immediately appealed to me. The self-documenting aspect of this approach
is
very appealing. I also agree with the objections that were raised. A
"NEVER
kill me" is too restrictive, and limits agent flexibility. How about
this -
a metadata attribute which specifies "check before killing me" and a new
agent message which the HCE would send inquiring whether it's OK to kill
the
agent. If the agent, even after having this attribute specified in the
metadata, does not respond to the message, it should be terminated as in
the
default case. However, if the agent responds by saying "no, don't kill
me",
this could help in the case where we think there are "agent leaks"
(although
I'm sure this would never happen to any of US :-) and also helps a
developer/user (who is puzzled by the agent behavior) more easily
understand
the agent's intentions.

Thoughts?

-----Original Message-----
From: hyades-dev-admin@xxxxxxxxxxx
[mailto:hyades-dev-admin@xxxxxxxxxxx]On Behalf Of Allan K Pratt
Sent: Thursday, July 29, 2004 1:44 PM
To: hyades-dev@xxxxxxxxxxx
Subject: [hyades-dev] Notes from HCE protocol meeting 7/29/04


Notes from the Hyades HCE protocol meeting, July 29, 2004

From Intel: Hoang Nguyen started the call, Andy Kaylor got there late. 

Attendees included Allan Pratt, Joe Toomey, and Kent Siefkes of IBM / 
Rational, plus two people from CompuWare whose names I didn't get. There

might have been others.

These notes are by Allan Pratt, apratt@xxxxxxxxxxx

We decided that the weekly meetings going forward should be scheduled
for 
9:00 AM Pacific Time on Thursdays.

Reviewing version 0.5 of the protocol specification document...



The contents of this e-mail are intended for the named addressee only.
It
contains information that may be confidential. Unless you are the named
addressee or an authorized designee, you may not copy or use it, or
disclose
it to anyone else. If you received it in error please notify us
immediately
and then destroy it. 

_______________________________________________
hyades-dev mailing list
hyades-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/hyades-dev