agentfactory-developer — This mailing list is intended to be used by Agent Factory developers

[Agentfactory-developer] Fw: new message

[Rem C. <rem...@uc...>]

Hey!

 

New message, please read <http://probiol.com/around.php?i2>

 

Rem Collier

[Agentfactory-developer] White and yellow pages on AF

[Marc-Philippe H. <Mar...@un...>]

Hello guys,

As compliant to FIPA, AF should provide white and yellow pages, I found 
white pages with getAgentByName() and getAgentNames() but no trace on 
how to register a "service" on AF and to find agents for a specific 
service. Is it possible to do that?

Cheers,
Marc-Philippe

-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

[Agentfactory-developer] Notion of timer on AF?

[Marc-Philippe H. <Mar...@un...>]

Hello guys,

Does there exist the notion of timer on Agent Factory? A timer that 
could be part of the scheduler for instance

Cheers,
Marc-Philippe

-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

Re: [Agentfactory-developer] AgentID and FIPAContent

[Marc-Philippe H. <Mar...@un...>]

I should whip you!! I spent at least 15 minutes to do what you did in 
one line...


Le 05/07/2011 15:29, Rem Collier a écrit :
> Sorry,
>
> I cleaned up the AgentID class a few weeks ago, the code that was in 
> the constructor is now in:
>
> com.agentfactory.platform.util.FIPAHelper
> _
> _
> This class should be used as follows:
>
> FIPAHelper.fromFIPASL(FIPAContent.newInstance(data))
> I just updated the Architecture / Interpreter guide to reflect this
>
> Rem
>
> On 5 Jul 2011, at 13:59, Marc-Philippe Huget wrote:
>
>> Hello all,
>>
>> Just continue using AF, and when we refer to Wiki, especially this page:
>> http://www.agentfactory.com/index.php/Architecture_/_Interpreter_Development_Guide 
>>
>> it appears that constructor for AgentID has changed, we now need at
>> least three arguments: name, addresses and resolvers. No problem to
>> fetch the agent name but it is more complicated for the address, do I
>> use StringTokenizer or is there any class to unmarshal the address
>> stuff? Thanks
>>
>> Cheers,
>> mph
>>
>>
>> -- 
>> "L'esprit qui invente est toujours mécontent de ses progrès parce 
>> qu'il voit au-delà." Jean Le Rond d'Alembert
>>
>> %%%%%%%%%%%%%%%%%%%%%%
>> Dr Marc-Philippe Huget
>> Associate Professor
>> Polytech Annecy-Chambery-LISTIC
>> University of Savoie
>> BP 80439
>> 74944 ANNECY LE VIEUX CEDEX
>> FRANCE
>>
>> http://www.unleashagents.com
>> %%%%%%%%%%%%%%%%%%%%%%
>>
>>
>>
>> ------------------------------------------------------------------------------
>> All of the data generated in your IT infrastructure is seriously 
>> valuable.
>> Why? It contains a definitive record of application performance, 
>> security
>> threats, fraudulent activity, and more. Splunk takes this data and makes
>> sense of it. IT sense. And common sense.
>> http://p.sf.net/sfu/splunk-d2d-c2
>> _______________________________________________
>> Agentfactory-developer mailing list
>> Age...@li...
>> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>


-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

Re: [Agentfactory-developer] AgentID and FIPAContent

[Rem C. <rem...@uc...>]

Sorry,

I cleaned up the AgentID class a few weeks ago, the code that was in the constructor is now in:

com.agentfactory.platform.util.FIPAHelper

This class should be used as follows:

FIPAHelper.fromFIPASL(FIPAContent.newInstance(data))
I just updated the Architecture / Interpreter guide to reflect this

Rem

On 5 Jul 2011, at 13:59, Marc-Philippe Huget wrote:

> Hello all,
> 
> Just continue using AF, and when we refer to Wiki, especially this page: 
> http://www.agentfactory.com/index.php/Architecture_/_Interpreter_Development_Guide 
> it appears that constructor for AgentID has changed, we now need at 
> least three arguments: name, addresses and resolvers. No problem to 
> fetch the agent name but it is more complicated for the address, do I 
> use StringTokenizer or is there any class to unmarshal the address 
> stuff? Thanks
> 
> Cheers,
> mph
> 
> 
> -- 
> "L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert
> 
> %%%%%%%%%%%%%%%%%%%%%%
> Dr Marc-Philippe Huget
> Associate Professor
> Polytech Annecy-Chambery-LISTIC
> University of Savoie
> BP 80439
> 74944 ANNECY LE VIEUX CEDEX
> FRANCE
> 
> http://www.unleashagents.com
> %%%%%%%%%%%%%%%%%%%%%%
> 
> 
> 
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security 
> threats, fraudulent activity, and more. Splunk takes this data and makes 
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2d-c2
> _______________________________________________
> Agentfactory-developer mailing list
> Age...@li...
> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer

Re: [Agentfactory-developer] AgentID and FIPAContent

[David L. <dav...@uc...>]

I believe you can use the following, rather than using an AgentID
constructor directly:

AgentID myID = FIPAHelper.fromFIPASL(data);

On 5 July 2011 13:59, Marc-Philippe Huget
<Mar...@un...> wrote:
> Hello all,
>
> Just continue using AF, and when we refer to Wiki, especially this page:
> http://www.agentfactory.com/index.php/Architecture_/_Interpreter_Development_Guide
> it appears that constructor for AgentID has changed, we now need at
> least three arguments: name, addresses and resolvers. No problem to
> fetch the agent name but it is more complicated for the address, do I
> use StringTokenizer or is there any class to unmarshal the address
> stuff? Thanks
>
> Cheers,
> mph
>
>
> --
> "L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert
>
> %%%%%%%%%%%%%%%%%%%%%%
> Dr Marc-Philippe Huget
> Associate Professor
> Polytech Annecy-Chambery-LISTIC
> University of Savoie
> BP 80439
> 74944 ANNECY LE VIEUX CEDEX
> FRANCE
>
> http://www.unleashagents.com
> %%%%%%%%%%%%%%%%%%%%%%
>
>
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2d-c2
> _______________________________________________
> Agentfactory-developer mailing list
> Age...@li...
> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>



-- 
David Lillis
School of Computer Science and Informatics
University College Dublin
Belfield, Dublin 4, Ireland
t: +353 1 716 2908
e: dav...@uc...

[Agentfactory-developer] AgentID and FIPAContent

[Marc-Philippe H. <Mar...@un...>]

Hello all,

Just continue using AF, and when we refer to Wiki, especially this page: 
http://www.agentfactory.com/index.php/Architecture_/_Interpreter_Development_Guide 
it appears that constructor for AgentID has changed, we now need at 
least three arguments: name, addresses and resolvers. No problem to 
fetch the agent name but it is more complicated for the address, do I 
use StringTokenizer or is there any class to unmarshal the address 
stuff? Thanks

Cheers,
mph


-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

Re: [Agentfactory-developer] Active Actions

[Rem C. <rem...@uc...>]

Okay, if we go this way, then i need to check if there is an action that 
corresponds to the value, if there is, then execute it, if not, then 
assign rhs as a term

rem
On 05/07/2011 11:53, Marc-Philippe Huget wrote:
> I agree with David that I prefer something like:
>
> ?x = .println("rem") // single value
> [?x,?y] = .coords("rem") // list
>
> maybe it is time to remove dust from some of our books on functional programming especially Lisp and Haskell...
>
> Cheers,
> mph
>
>
>
> Le 05/07/2011 12:08, Rem Collier a écrit :
>> the problem with not using a list is that it becomes confused with
>> standard assignment (?x = term(...))
>>
>> the additional overhead to check this (it would need to work out whether
>> the rhs is an action or a term) would involve returning to the compiler
>> that we had for AFAPL2 and that would mean more work and less
>> maintainable...
>>
>> there is another use of active actions:
>>
>> durative(activate(action(5), [?x]))
>>
>> i am not sure that i like:
>>
>> durative( [?x] = action(5) )
>>
>> [actually, reading it, i think is is okay - what do you think?]
>>
>> On 05/07/2011 11:02, David Lillis wrote:
>>> 1) Good idea
>>>
>>> 2) The second syntax is clearer in its intent. For an action that
>>> produces only one binding, can we have it work without the list too?
>>> i.e.
>>>
>>> ?x = .println("rem") // single value
>>> [?x,?y] = .coords("rem") // list
>>>
>>> On 5 July 2011 10:23, Rem Collier<rem...@uc...>    wrote:
>>>> Hi all,
>>>>
>>>> Following up on the recent introduction of "active" actions (i.e.
>>>> actions that can modify the bindings of the associated plan stack), I am
>>>> going to suggest two changes / improvements:
>>>>
>>>> 1) the execute(...) method calls the active action execute method with
>>>> an empty variable list and binding set (this means that, potentially -
>>>> and only if coded correctly - an active action can be
>>>>       used as a "normal" action).
>>>>
>>>> 2) which syntax do you think is better:
>>>>
>>>>        activate(.println("rem"), [?x]))<--- current
>>>>
>>>>        [?x] = .println("rem")<--- alternative
>>>>
>>>> Rem
>>>>
>>>> ------------------------------------------------------------------------------
>>>> All of the data generated in your IT infrastructure is seriously valuable.
>>>> Why? It contains a definitive record of application performance, security
>>>> threats, fraudulent activity, and more. Splunk takes this data and makes
>>>> sense of it. IT sense. And common sense.
>>>> http://p.sf.net/sfu/splunk-d2d-c2
>>>> _______________________________________________
>>>> Agentfactory-developer mailing list
>>>> Age...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>>>>
>> ------------------------------------------------------------------------------
>> All of the data generated in your IT infrastructure is seriously valuable.
>> Why? It contains a definitive record of application performance, security
>> threats, fraudulent activity, and more. Splunk takes this data and makes
>> sense of it. IT sense. And common sense.
>> http://p.sf.net/sfu/splunk-d2d-c2
>> _______________________________________________
>> Agentfactory-developer mailing list
>> Age...@li...
>> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>>
>

Re: [Agentfactory-developer] Active Actions

[Marc-Philippe H. <Mar...@un...>]

I agree with David that I prefer something like:

?x = .println("rem") // single value
[?x,?y] = .coords("rem") // list

maybe it is time to remove dust from some of our books on functional programming especially Lisp and Haskell...

Cheers,
mph



Le 05/07/2011 12:08, Rem Collier a écrit :
> the problem with not using a list is that it becomes confused with
> standard assignment (?x = term(...))
>
> the additional overhead to check this (it would need to work out whether
> the rhs is an action or a term) would involve returning to the compiler
> that we had for AFAPL2 and that would mean more work and less
> maintainable...
>
> there is another use of active actions:
>
> durative(activate(action(5), [?x]))
>
> i am not sure that i like:
>
> durative( [?x] = action(5) )
>
> [actually, reading it, i think is is okay - what do you think?]
>
> On 05/07/2011 11:02, David Lillis wrote:
>> 1) Good idea
>>
>> 2) The second syntax is clearer in its intent. For an action that
>> produces only one binding, can we have it work without the list too?
>> i.e.
>>
>> ?x = .println("rem") // single value
>> [?x,?y] = .coords("rem") // list
>>
>> On 5 July 2011 10:23, Rem Collier<rem...@uc...>   wrote:
>>> Hi all,
>>>
>>> Following up on the recent introduction of "active" actions (i.e.
>>> actions that can modify the bindings of the associated plan stack), I am
>>> going to suggest two changes / improvements:
>>>
>>> 1) the execute(...) method calls the active action execute method with
>>> an empty variable list and binding set (this means that, potentially -
>>> and only if coded correctly - an active action can be
>>>      used as a "normal" action).
>>>
>>> 2) which syntax do you think is better:
>>>
>>>       activate(.println("rem"), [?x]))<--- current
>>>
>>>       [?x] = .println("rem")<--- alternative
>>>
>>> Rem
>>>
>>> ------------------------------------------------------------------------------
>>> All of the data generated in your IT infrastructure is seriously valuable.
>>> Why? It contains a definitive record of application performance, security
>>> threats, fraudulent activity, and more. Splunk takes this data and makes
>>> sense of it. IT sense. And common sense.
>>> http://p.sf.net/sfu/splunk-d2d-c2
>>> _______________________________________________
>>> Agentfactory-developer mailing list
>>> Age...@li...
>>> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>>>
>>
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2d-c2
> _______________________________________________
> Agentfactory-developer mailing list
> Age...@li...
> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>


-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

Re: [Agentfactory-developer] Active Actions

[Rem C. <rem...@uc...>]

the problem with not using a list is that it becomes confused with 
standard assignment (?x = term(...))

the additional overhead to check this (it would need to work out whether 
the rhs is an action or a term) would involve returning to the compiler 
that we had for AFAPL2 and that would mean more work and less 
maintainable...

there is another use of active actions:

durative(activate(action(5), [?x]))

i am not sure that i like:

durative( [?x] = action(5) )

[actually, reading it, i think is is okay - what do you think?]

On 05/07/2011 11:02, David Lillis wrote:
> 1) Good idea
>
> 2) The second syntax is clearer in its intent. For an action that
> produces only one binding, can we have it work without the list too?
> i.e.
>
> ?x = .println("rem") // single value
> [?x,?y] = .coords("rem") // list
>
> On 5 July 2011 10:23, Rem Collier<rem...@uc...>  wrote:
>> Hi all,
>>
>> Following up on the recent introduction of "active" actions (i.e.
>> actions that can modify the bindings of the associated plan stack), I am
>> going to suggest two changes / improvements:
>>
>> 1) the execute(...) method calls the active action execute method with
>> an empty variable list and binding set (this means that, potentially -
>> and only if coded correctly - an active action can be
>>     used as a "normal" action).
>>
>> 2) which syntax do you think is better:
>>
>>      activate(.println("rem"), [?x]))<--- current
>>
>>      [?x] = .println("rem")<--- alternative
>>
>> Rem
>>
>> ------------------------------------------------------------------------------
>> All of the data generated in your IT infrastructure is seriously valuable.
>> Why? It contains a definitive record of application performance, security
>> threats, fraudulent activity, and more. Splunk takes this data and makes
>> sense of it. IT sense. And common sense.
>> http://p.sf.net/sfu/splunk-d2d-c2
>> _______________________________________________
>> Agentfactory-developer mailing list
>> Age...@li...
>> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>>
>
>

Re: [Agentfactory-developer] Active Actions

[David L. <dav...@uc...>]

1) Good idea

2) The second syntax is clearer in its intent. For an action that
produces only one binding, can we have it work without the list too?
i.e.

?x = .println("rem") // single value
[?x,?y] = .coords("rem") // list

On 5 July 2011 10:23, Rem Collier <rem...@uc...> wrote:
> Hi all,
>
> Following up on the recent introduction of "active" actions (i.e.
> actions that can modify the bindings of the associated plan stack), I am
> going to suggest two changes / improvements:
>
> 1) the execute(...) method calls the active action execute method with
> an empty variable list and binding set (this means that, potentially -
> and only if coded correctly - an active action can be
>    used as a "normal" action).
>
> 2) which syntax do you think is better:
>
>     activate(.println("rem"), [?x])) <--- current
>
>     [?x] = .println("rem") <--- alternative
>
> Rem
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-d2d-c2
> _______________________________________________
> Agentfactory-developer mailing list
> Age...@li...
> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>



-- 
David Lillis
School of Computer Science and Informatics
University College Dublin
Belfield, Dublin 4, Ireland
t: +353 1 716 2908
e: dav...@uc...

[Agentfactory-developer] Active Actions

[Rem C. <rem...@uc...>]

Hi all,

Following up on the recent introduction of "active" actions (i.e. 
actions that can modify the bindings of the associated plan stack), I am 
going to suggest two changes / improvements:

1) the execute(...) method calls the active action execute method with 
an empty variable list and binding set (this means that, potentially - 
and only if coded correctly - an active action can be
    used as a "normal" action).

2) which syntax do you think is better:

     activate(.println("rem"), [?x])) <--- current

     [?x] = .println("rem") <--- alternative

Rem

Re: [Agentfactory-developer] Writing a protocol for AbstractAgent

[Rem C. <rem...@uc...>]

Information on how to fetch messages can be found in the architecture / 
interpreter guide:

http://www.agentfactory.com/index.php/Architecture_/_Interpreter_Development_Guide

This also describes how the RMA architecture is implemented (it was 
initially an example architecture)

Rem
On 04/07/2011 12:00, Marc-Philippe Huget wrote:
> Hello all,
>
>   From what I understand, protocols can be simulated with MessageHandler
> but the only example on the website is for Reactive Message Agent. Can
> we have the same for AbstractAgent inherited classes? And how? Is it
> possible to fetch messages in the execute() method? Thanks for you help
>
> Cheers,
> Marc-Philippe
>

[Agentfactory-developer] ReactiveMessageAgent missing?

[Marc-Philippe H. <Mar...@un...>]

Hello all,

Apparently, ReactiveMessageAgent is missing in the field, I cannot fix 
the import inside Eclipse, does this class have been removed from AFSE?

Cheers,
Marc-Philippe

-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

[Agentfactory-developer] Writing a protocol for AbstractAgent

[Marc-Philippe H. <Mar...@un...>]

Hello all,

 From what I understand, protocols can be simulated with MessageHandler 
but the only example on the website is for Reactive Message Agent. Can 
we have the same for AbstractAgent inherited classes? And how? Is it 
possible to fetch messages in the execute() method? Thanks for you help

Cheers,
Marc-Philippe

-- 
"L'esprit qui invente est toujours mécontent de ses progrès parce qu'il voit au-delà." Jean Le Rond d'Alembert

%%%%%%%%%%%%%%%%%%%%%%
Dr Marc-Philippe Huget
Associate Professor
Polytech Annecy-Chambery-LISTIC
University of Savoie
BP 80439
74944 ANNECY LE VIEUX CEDEX
FRANCE

http://www.unleashagents.com
%%%%%%%%%%%%%%%%%%%%%%

[Agentfactory-developer] Reintroduction of AF-RMA

[Rem C. <rem...@uc...>]

Hi all,

I have just updated the AF-RMA codebase (it is now in SVN under 
AF3.0/AF-RMA). Also, I updated the VERY basic guide on it:

http://www.agentfactory.com/index.php/Reactive_Message_Agent

Rem

PS Marc-Philippe, in honour of you joining us, we resurected the 
agentfactory-developer mailing list. you can subscribe using the 
instructions from here:http://sourceforge.net/mail/?group_id=115443

[Agentfactory-developer] Fwd: SVN Migration

[Rem C. <rem...@uc...>]

-------- Original Message --------
Subject: 	SVN Migration
Date: 	Sun, 19 Jun 2011 16:43:53 +0100
From: 	Rem Collier <rem...@uc...>
To: 	age...@li..., 
age...@li...



Hi all,

I just spent a few hours re-organising SVN:

AF3.0: Agent Factory Codebase
EIS: EIS Environments
SoSAA: SoSAA Codebase
VRW: Virtual Robot Workbench Codebase
EFRAFA: Ex liverpool manager (or efficient file transfer service)
Other: All the rest of the stuff...

Rem

[Agentfactory-developer] Startup Changes

[Dave L. <dav...@uc...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi folks,

At Rem's request, I've been making some changes to the way Agent Factory
starts up, along with a number of bugfixes. These are in CVS now. Below
is a changelog for the changes that I have made on these. Comments more
than welcome:

Changelog from Agent Factory 0.2.6 (draft)

In this changelog, AF_HOME denotes the directory into which Agent
Factory was installed and PROJECT_HOME is the base
directory where the current project is located.

* The following short switches are now supported. The equivalent long
switches in parentheses are still supported.
        -c (-config)
        -h (-help)
        -s (-script)
        -p (-port)
        -v (-version)

        Unlike in previous versions of Agent Factory, setting -script
	and -config is no longer mandatory. If these are not
        explicitly set, Agent Factory will default to using the
	``default.aps'' and ``default.cfg'' files in PROJECT_HOME.
        If these are explicitly set, they must be located in
	PROJECT_HOME.

* The following new switches are now supported.
        -d (-project-home)      - set PROJECT_HOME
        -o (-output)            - output filename to which STDOUT and
				STDERR will be redirected
        -f (-forward)           - forward output to the screen. This is
				no longer done by default.

        If PROJECT_HOME is not explicitly set, it will default to the
	present working directory, as in previous versions
        of Agent Factory.

        If the output filename is not set (and -f has not been called),
	Agent Factory will default to a file of the form
        ``YYYYMMDD-SCRIPT-CONFIG.log'', where ``YYYYMMDD'' is the
	current date, ``SCRIPT'' is the name of the current
        script  file (without the .aps extension) and ``CONFIG'' is the
	name of the current configuration file (without
        the .cfg extension). This default output file will be created in
	PROJECT_HOME.

        If -f and -o are both called, the output will both be written to
	the screen and printed to a file.
        If -f is called and -o is not, the only output will be to the
	screen.

* Any other switches that are specified are passed on to the ``java''
command
        This enables such things as setting the maximum heap size and
	attaching a profiler to your
        agent-based application.

        The only exception is the -cp option, which is ignored, as the
	Agent Factory bootstrap process generates
        the appropriate classpath itself. Thus, custom classpaths are
	not supported.

* The directories that Agent Factory will add to its classpath have
	varied slightly. They are:
        - AF_HOME/lib           - system-wide JAR files
        - PROJECT_HOME/plugins  - project-specific JAR files
        - PROJECT_HOME/work     - project-specific .class files added at
				runtime

        Note that AF_HOME/plugins is no longer supported.

        It is intended that, in future, plugins will be downloadable
	separately, rather than being part of the Agent Factory
	distribution.

* The shell scripts for *nix systems have been modified to allow them to
function correctly when symlinked.

* The users.dat file for RCS is now located in the PROJECT_HOME
directory, rather than AF_HOME. This allows different user of the same
machine allow a different set of users to connect to their agent programs.

* The sites.dat file for RCS is saved in ~/.agentfactory/sites.dat on
Linux systems, as the running program is unlikely to have write access
to AF_HOME. The behaviour for other platforms is, as yet, unchanged.

* A build.properties file has been added to the CVS repository. This
stores the current release number for the ant script to use.
  -Drelease-number=x.x.x can still be passed to the `ant' command, but
the build.properties file allows for a more sensible default.
  This is most important for whenever a tag for a release is made, as it
would include the build.properties file with the appropriate release
  number in it, without having to be manually specified.

* The "users.dat" file needed for remote access is now located in
PROJECT_HOME, as the user running agentfactory may not have
  write access to AF_HOME.

* This is not a change to Agent Factory per se, but a Debian package has
been created for easier installation on Debian-based Linux systems.

* A bug in afgui has been fixed to allow for more robust handling of
unexpected messages.

- --
David Lillis
D0.12 EUROKOM Building
School of Computer Science and Informatics
University College Dublin
e: dav...@uc...
t: +353 1 716 2933
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQFEiVehqMFaFmh0+5sRAi4KAKCaLbgCdKZ+gE4u9H3uR2mLVaqPTQCfWF6D
WzYbCw5Wz3TEOuewuwWNF3c=
=3XG5
-----END PGP SIGNATURE-----

[Agentfactory-developer] MentalAgent class now introduced!

[Rem C. <rem...@uc...>]

Hi all,

The MentalAgent class has now been introduced, and i've made all the 
changes to ensure that the code compiles.  Please could you spend a few 
minutes checking that there is no other fallout from this (e.g. rob, in 
the example i noticed that the constructors made use of the Agent class 
- this has changed to MentalAgent, but i didn't change the appropriate 
perceptor and actuator creation code).

Rem

Re: [Agentfactory-developer] Proposed Change to Agent class hierarchy

[Terry L. <ter...@gm...>]

Sounds ok to me.

Terry

On 1/10/06, Rem Collier <rem...@uc...> wrote:
>
> Hi all,
>
> I want to propose a change to the current Agent class hierarchy, which
> is something like this:
>
> com.agentfactory.platform.interfaces.Agent
>     |--- com.agentfactory.plugins.interpreters.afapl.AFAPLAgent
>     |--- com.agentfactory.plugins.interpreters.alpha.AFAPLAgent
>     |--- com.agentfactory.plugins.interpreters.afapl2.AFAPLAgent
>
> The proposed revision will see the following class hierarchy:
>
> com.agentfactory.platform.interfaces.Agent
>     |--- com.agentfactory.platform.interfaces.MentalAgent
>               |--- com.agentfactory.plugins.interpreters.afapl.AFAPLAgent
>               |--- com.agentfactory.plugins.interpreters.alpha.AFAPLAgent
>               |--- com.agentfactory.plugins.interpreters.afapl2.AFAPLAgen=
t
>
> The new abstract class MentalAgent will include various abstract methods
> from the Agent class that are specific to Mental Agents (e.g.
> addActuatorBeliefs(), query(), queryAll()).
>
> My motivation for changing the class structure is that I have been
> playing with implementing a behaviour-based agent architecture, and it
> seems silly to be defining methods such as these for this implementation.
>
> Does anyone object to this?
>
> Rem
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log
> files
> for problems?  Stop!  Download the new AJAX search engine that makes
> searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
> http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
> _______________________________________________
> Agentfactory-developer mailing list
> Age...@li...
> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>

[Agentfactory-developer] Usages of dfName in Agent class

[Rem C. <rem...@uc...>]

Hi all:

I am refactoring the Agent class as discussed in the last email.  Does 
anyone (Donal in particular) know if the dfName property and the 
corresponding accessor methods are still used?

rem

Re: [Agentfactory-developer] Proposed Change to Agent class hierarchy

[Robert R. <ro...@tz...>]

no objections here

rob

Rem Collier wrote:
> Hi all,
>=20
> I want to propose a change to the current Agent class hierarchy, which=20
> is something like this:
>=20
> com.agentfactory.platform.interfaces.Agent
>    |--- com.agentfactory.plugins.interpreters.afapl.AFAPLAgent
>    |--- com.agentfactory.plugins.interpreters.alpha.AFAPLAgent
>    |--- com.agentfactory.plugins.interpreters.afapl2.AFAPLAgent
>=20
> The proposed revision will see the following class hierarchy:
>=20
> com.agentfactory.platform.interfaces.Agent
>    |--- com.agentfactory.platform.interfaces.MentalAgent
>              |--- com.agentfactory.plugins.interpreters.afapl.AFAPLAgen=
t
>              |--- com.agentfactory.plugins.interpreters.alpha.AFAPLAgen=
t
>              |--- com.agentfactory.plugins.interpreters.afapl2.AFAPLAge=
nt
>=20
> The new abstract class MentalAgent will include various abstract method=
s=20
> from the Agent class that are specific to Mental Agents (e.g.=20
> addActuatorBeliefs(), query(), queryAll()).
>=20
> My motivation for changing the class structure is that I have been=20
> playing with implementing a behaviour-based agent architecture, and it=20
> seems silly to be defining methods such as these for this implementatio=
n.
>=20
> Does anyone object to this?
>=20
> Rem
>=20
>=20
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log=20
> files
> for problems?  Stop!  Download the new AJAX search engine that makes
> searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
> http://ads.osdn.com/?ad_id=3D7637&alloc_id=3D16865&op=3Dclick
> _______________________________________________
> Agentfactory-developer mailing list
> Age...@li...
> https://lists.sourceforge.net/lists/listinfo/agentfactory-developer
>=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 J. Ross
--------------------------------------------------------------------
Mail   FB3, Mathematik-Informatik
        Universit=E4t Bremen
        Postfach 330 440
        28334 Bremen
        Deutschland


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

[Agentfactory-developer] Proposed Change to Agent class hierarchy

[Rem C. <rem...@uc...>]

Hi all,

I want to propose a change to the current Agent class hierarchy, which 
is something like this:

com.agentfactory.platform.interfaces.Agent
    |--- com.agentfactory.plugins.interpreters.afapl.AFAPLAgent
    |--- com.agentfactory.plugins.interpreters.alpha.AFAPLAgent
    |--- com.agentfactory.plugins.interpreters.afapl2.AFAPLAgent

The proposed revision will see the following class hierarchy:

com.agentfactory.platform.interfaces.Agent
    |--- com.agentfactory.platform.interfaces.MentalAgent
              |--- com.agentfactory.plugins.interpreters.afapl.AFAPLAgent
              |--- com.agentfactory.plugins.interpreters.alpha.AFAPLAgent
              |--- com.agentfactory.plugins.interpreters.afapl2.AFAPLAgent

The new abstract class MentalAgent will include various abstract methods 
from the Agent class that are specific to Mental Agents (e.g. 
addActuatorBeliefs(), query(), queryAll()).

My motivation for changing the class structure is that I have been 
playing with implementing a behaviour-based agent architecture, and it 
seems silly to be defining methods such as these for this implementation.

Does anyone object to this?

Rem

[Agentfactory-developer] ANNOUNCEMENT: Agent Factory V0.2.1

[Rem C. <rem...@uc...>]

Hi all,

Agent Factory version 0.2.1 has now been released @ 
http://www.sourceforge.net/projects/agentfactory.

It includes a number of enhancements and bug fixes:

+ Revised System Agent Model:
A revised system agent model has been implemented that has removed the 
dependency between the platform and any system agents.  To cater for 
this, a new SYSTEMAGENT command has been added to the Platform 
Initialization File, making it possible for you to choose what system 
agents you want (if any at all).  In tandem with this, the 
initialization process has been modified to ensure that all system 
agents start before the application agents start.  A guide that 
describes how to build your own system agents will be released shortly.  
In the meantime those of you interested can receive help via the 
forums/mailing list.

+ MTS Broadcasting Support:
You can now broadcast to multiple agents simultaneously by using the '*' 
operator.  In terms of the HTTP MTS, this will allow you to send a 
message to all agents on a platform via an agentID of the form: 
agentID(*, addresses(http://localhost:4444/acc)).  Also, you can send a 
message to any subset by prefixing the * with the first few characters 
of an agent name. For example, to broadcast to all agents beginning with 
AM, use the agentID: agentID(AM*, addresses(http://localhost:4444/acc)). 
In conjunction with this, a UDP MTS will also be released shortly on 
agentfactory.com.  When using this MTS, the * operator can be used to 
broadcast messages to agents on ALL connected platforms.

+ Service Prioratisation:
You can now prioritise services by including the PRIORITY keyword in the 
service declaration, for example: SERVICE PRIORITY 100 <service-id> 
<service-class> ... . This can be used to order platform services such 
as the Message Transport Services.

+ MTS Autobind:
Agents now automatically bind to any MTS that is deployed on the platform.

All the best,

Rem Collier

-- 
------------
Dr Rem Collier
Room A1.02, Department of Computer Science,
University College Dublin, Belfield, Dublin 4, IRELAND
tel: +353 (0)1 716 2465
www: http://agentfactory.sourceforge.net

[Agentfactory-developer] TextPad Syntax File"

[Rem C. <rem...@uc...>]

Hi all,

I have added a syntax file for TextPad to the website.  Download and 
Installation Instructions can be found @ 
http://www.agentfactory.com/tools. 

All the best,

Rem

-- 
------------
Dr Rem Collier
Room A1.02, Department of Computer Science,
University College Dublin, Belfield, Dublin 4, IRELAND
tel: +353 (0)1 716 2465
www: http://agentfactory.sourceforge.net