[Agentfactory-developer] Re: language guide

[rem c. <rem...@uc...>]

hey rob - the list now seems to be working - but there is a BIG delay=
=20
in the delivery of messages...


Robert Ross wrote:


>Hi Rem,
>I've cc-ed the developer list in case it starts coming through.
>
>First, let me see if I understand the intended difference between AP=
S=20
>and RCS? You see the APS as something passive in nature, where a scr=
ipt=20
>is loaded, listed commands are executed, and results are channeled i=
nto=20
>stdout. While, on the other hand, the RCS would be a less passive=
=20
>protocol intended for hooking up to a remote GUI.
>
>If I my interpretation is correct, then I see a point to the=20
difference,=20
>but it is clear that the APS would be a subset of RCS commands, and=
=20
>perhaps the implementation could reflect this.

yeah - i agree.  the only problem is with the link is that the APS=
=20
should only employ the agent management subset of the RCS commands.

>Also, do you really feel that it is worthwhile to let the current GU=
I=20
>design die? While deployment to computationally limited devices, and=
=20
>eventual deployment to enterprise style servers might favor your=
=20
>approach -- an awful lot of development (both developer development =
and=20
>user development) will most likely take place with a platform runnin=
g=20
>locally.

good point - so, we can keep the local gui as a testing and debugging=
=20
environment - maybe we should create a "debugging mode" for agent=
=20
factory?

my plan for the RCS part of a bigger plan to move to the concept of a=
n=20
"agent cluster" - namely that a set of machines can have af installed=
=20
on them, be remotely managed (i.e. we can use linux run level 3 so we=
=20
don't get the hit from the use of the gui).  users should then be abl=
e=20
to "inject" agents into the cluster (code etc is automatically=20
deployed) and a built in autonomic management system will ensure that=
=20
the set of agents running of the machine is optimally configured for=
=20
robustness and efficiency... finally, my expectation is that the=20
cluster will be able to manage multiple concurrent agent applications=
=20
and there will be a number of views of the cluster (e.g an applicatio=
n=20
view will show the current distribution of the agents)

>That said, then maybe we can agree that the run-time environment mig=
ht=20
>have an RCS GUI (bandwidth permitting), but that the IDE (once=20
>redeveloped) would not be based on this design.
>
>Better go... I've got a meeting to go to.
>
>cheers
>
>rob
>
>
>
>
>rem collier wrote:
>> hey rob!
>>=20
>> i've spent the last 8 hours trying to send the following message t=
o=20
the=20
>> developer's list, but it hasn't come through yet, so i'm sending i=
t=20
>> direct to you (i'll cc all future ones).  the mail is basically ab=
out=20
>> the RCS service...
>>=20
>> were we go:
>>=20
>> Hi guys,
>>=20
>> As my first post, i want to have a quick discussion about the remo=
te=20
>> command service (RCS) for agent factory.  this service should allo=
w=20
>> users to remotely administer agentfactory platforms.
>>=20
>> currently, af provides a facility to remotely execute commands fro=
m=20
the=20
>> APS scripting language.  this facility is an option that can be us=
ed=20
>> INSTEAD of specifying a startup script.  while this is useful, it =
is=20
>> not the same as the RCS.  the RCS service should be able to run in=
=20
>> tandem with a startup script.  in addition, there are a number of=
=20
>> operations that an RCS should support which should not be supporte=
d by=20
>> the startup script (e.g. generating user friendly lists of active=
=20
>> agents).  my goal for the RCS is to be much more than this.  it sh=
ould=20
>> also allow users to manage the platform services that are availabl=
e. =20
>> in particular, i would like to be able to remotely start a monitor=
ing=20
>> service that can be used to present debugging information to the u=
ser.=20
=20
>> i could then user the RCS to shutdown this service when it is no=
=20
longer=20
>> needed.  by including this facility, i am extending the remit of t=
he=20
>> RCS beyond the context of the scripting language (IMO: the scripti=
ng=20
>> language should NOT support configuration of platform services - t=
his=20
>> is the job of the platform configuration files).
>>=20
>> anyway, this is my opinion based upon a quick review of the existi=
ng=20
>> infrastructure.  if there are no real problems, i intend to contin=
ue=20
>> with my work on this service (which will itself be deployed as a=
=20
>> plugin) and will show you a reasonable first version of it in the =
near=20
>> future.
>>=20
>> rem
>>=20
>> ps here is a list of commands that i would expect the RCS to suppo=
rt:
>> connect XX
>> close
>> create agent
>> create service
>> terminate agent
>> terminate service
>> resume agent
>> suspend agent
>> show agents (nicely formated output)
>> show services (nicely formated output)
>> list agents ('|' and CR delimited output)
>> list services ('|' and CR delimited output)
>> shutdown
>> restart
>>=20
>>=20
>>=20
>> ----------------------------------------
>> Dr Rem Collier<BR>Lecturer<BR>Department
>>  of Computer Science,<BR>University Coll
>> ege Dublin,<BR>IRELAND.<BR>www: http://w
>> ww.agentfactory.com<BR>tel: +353 1 716 2
>> 465
>>=20
>
>--=20
>=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
>Robert Ross
>--------------------------------------------------------------------
>Mail   FB3, Mathematik-Informatik      Deptartment of Mathematics
>        Universit=E4t Bremen                   & Computer Science
>        Postfach 330 440                University of Bremen
>        28334 Bremen                    P.O. Box 330 440
>        Deutschland                     28334 Bremen
>                                        Germany
>
>Phone  +49 421 218 7129
>Fax    +49 421 218-3054
>Web    http://www.informatik.uni-bremen.de/~robertr
>Email  ro...@in...
>=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
>
>

----------------------------------------
Dr=A0Rem=A0Collier<BR>Lecturer<BR>Department
=A0of=A0Computer=A0Science,<BR>University=A0Coll
ege=A0Dublin,<BR>IRELAND.<BR>www:=A0http://w
ww.agentfactory.com<BR>tel:=A0+353=A01=A0716=A02
465