jason-users Mailing List for jason

Hi,

you can launch saci with a parameter that connects saci with another  
host, something like

java saci.launcher.LauncherD -connect <the other host>

(LauncherD starts saci without the GUI)
of course you need to setup the casspath before.

Another solution using ant:
The Jason 1.3 creates a task "saci-client" in the bin/build.xml that  
uses -connect.  However the ant script is not prepared for the host as  
a paramter, so edit that task as follows:

    <property name="saci.remote.host" value="localhost"/>

     <target name="saci-client" >
        <echo message="remote host ${saci.remote.host}" />
        <java classname="${saci.main}" failonerror="true" fork="yes"  
dir="${basedir}">
           <classpath refid="project.classpath"/>
           <jvmarg value="-Djava.security.policy=jar:file:/Users/jomi/ 
bin/Jason-1.3/Jason.app/Contents/Resources/Java/lib/saci.jar!/policy"/>
           <arg line="-connect ${saci.remote.host}"/>
        </java>
     </target>

ant then run ant as follow:

        ant -f bin/build.xml -Dsaci.remote.host=thenameofremotehost  
saci-client

HTH,

Jomi


On Apr 28, 2009, at 7:48 PM, <fe...@al...>  
<fe...@al...> wrote:

> Hello,
> Already I achieved that it works saci with my example, thanks for the
> help. The problem that I have now is that I need to do many tests and
> to be in the habit of using a file in python that launches Jason with
> different configurations. When I was using it in centralized there was
> no problem, but now that is distributed, and I have to extract the
> window of saci (with " ant saci ") to choose the remote host, I cannot
> make it automatic. There is some file of properties that could be
> modified or something not to need the window of configuration of saci?
> Thank you.
>
> Regards
>
>
>
> ------------------------------------------------------------------------------
> Register Now & Save for Velocity, the Web Performance & Operations
> Conference from O'Reilly Media. Velocity features a full day of
> expert-led, hands-on workshops and two days of sessions from industry
> leaders in dedicated Performance & Operations tracks. Use code  
> vel09scf
> and Save an extra 15% before 5/3. http://p.sf.net/sfu/velocityconf
> _______________________________________________
> Jason-users mailing list
> Jas...@li...
> https://lists.sourceforge.net/lists/listinfo/jason-users

-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

Hello, 
Already I achieved that it works saci with my example, thanks for the 
help. The problem that I have now is that I need to do many tests and 
to be in the habit of using a file in python that launches Jason with 
different configurations. When I was using it in centralized there was 
no problem, but now that is distributed, and I have to extract the 
window of saci (with " ant saci ") to choose the remote host, I cannot 
make it automatic. There is some file of properties that could be 
modified or something not to need the window of configuration of saci?
Thank you.

Regards

Hello again,

attached you find a new version of the 'customised mind inspection'.

this example works for any kind of infrastructure (centralised, jade,  
saci, ...)

however, it does not inspect the agent from the environment class, but  
by means of a controller (a feature used also for the debugger)

the main idea is:
1. create system-cycles
2. each agent perform one reasoning-cycle each system-cycle
3. a method of the controller is called when the cycle finishes
4. this method is implemented in this example (class MyController)  
doing the following:
    - ask each agent its mind state
    - the result is a XML document with all the agent state (beliefs,  
intentions, options, suspended intentions, ...)
    - use a XSL style sheet to translate it to other thing (in the  
example it is translate to html and shown in the interface

if you prefer a text output and is not used to XSL style sheets, you  
may ask me for such an style.

note that the environment is not supposed to know the agents' mind, so  
this solution based on controller. in case you must do that in the  
environment class, we can discuss other solutions.

HTH

Jomi

Hello Ondřej,

I am sending attached an example of environment that gets the 'minds'
of the agents. It works only in centralised infrastructure. I'll send
a more general solution later.

HTH,

Jomi

2009/4/25 Rafael H Bordini <r.b...@ac...>:
> Hi all,
> Sorry I missed this auto-discarded message from 12/04.
> Ondrej: your message was discarded probably because you sent it from a
> different address from the one you registered in the list.
> Cheers,
> Rafael
>
> The attached message has been automatically discarded.
>
> ---------- Forwarded message ----------
> From: "Ondřej Malačka" <xma...@st...>
> To: jas...@li...
> Date: Sun, 12 Apr 2009 18:06:37 +0200
> Subject: Belief base inspect in Environment
> Hello, I need inspect belief base of agents in class which is extended from
> Environment class, because i had like to show some belefies of agents to
> user inside JPanel I made. But o don't know, if and how it is possible.
> Thanks a lot for help , Ondrej.
>
>
> ------------------------------------------------------------------------------
> Crystal Reports &#45; New Free Runtime and 30 Day Trial
> Check out the new simplified licensign option that enables unlimited
> royalty&#45;free distribution of the report engine for externally facing
> server and web deployment.
> http://p.sf.net/sfu/businessobjects
> _______________________________________________
> Jason-users mailing list
> Jas...@li...
> https://lists.sourceforge.net/lists/listinfo/jason-users
>
>



-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

Hi all,
Sorry I missed this auto-discarded message from 12/04.
Ondrej: your message was discarded probably because you sent it from a 
different address from the one you registered in the list.
Cheers,
Rafael

Dear users,

We have just released the version 1.3 of Jason!

This release fixes some bugs, improves the performance of the agents,  
and includes a new example. Details are included below in this email.

Download:
https://sourceforge.net/project/showfiles.php?group_id=98417&package_id=105352&release_id=677200

Regards,

-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

--
New features
- New kind of term: plan terms. It is any ASL plan enclosed
   by { and }. Internal actions that use plans and triggers
   as parameters use this new kind of terms now.
   E.g. we can use
                  .relevant_plans({ +!g(_) }, ListOfPlans)
       instead of .relevant_plans( "+!g(_)" , ListOfPlans)

- The performance is improved (+- 15%) by not cloning all the
   plan in the creation of intended means

New internal actions
- .puts: used for printing messages to the console (based on Ruby
   similar method). Developed by Felipe Meneguzzi.
- .all_names: get the name of all agentes in the system.
- .list_plans: print out the plans of agent's plan library
- .prefix, .suffix and .sublist: backtracks all prefixes, suffixes and
    sublists of a list

New example
- Blocks World

Bugs fixed
- arithmetic expressions were considered as Literal, causing
   failures in unification
- variable unified with atom atom couldn't be added in BB, as
   in X = p; +X;
- The third argument of .create_agent (list of options) caused
   failure
- Some concurrent execution of .wait and .drop_desire/intention
   didn't work (the intention isn't dropped)
- send askHow with 4th argument blocked the intention
- TimeSteppedEnvironment waited timeout even when not required

Hi Edd,

The VarTerm class have to be used only for variables, and not tuples  
like "(false, 0, 3)". If you are not very used to the Jason API, I'd  
suggest you to use the parser to create the Literals for perceptions.  
Something like"

String t =
    "terrain([" +
    "tile(", + tile1.isPath() +"," + tile.getCount() + .....
    ..... +
    "])"

Literal percept = Literal.parseLiteral(t);

...

Note that "(false, 0, 3)" is not a valid structure in Jason, so I  
added the functor tile: "tile(false, 0, 3)".


HTH

Jomi


> Hi,
>
> I am trying to add percepts to agents as a list, below is the Java
> code from the Environment Class;
>
> public void updatePercepts(String agName, Tile tile1, Tile tile2,  
> Tile tile3) {
>         clearPercepts(agName);
>
>         VarTerm t1 = new VarTerm("("+tile1.isPath()+",
> "+tile1.getCount()+", "+tile1.getTC()+")");
>         VarTerm t2 = new VarTerm("("+tile2.isPath()+",
> "+tile2.getCount()+", "+tile2.getTC()+")");
>         VarTerm t3 = new VarTerm("("+tile3.isPath()+",
> "+tile3.getCount()+", "+tile3.getTC()+")");
>
>         VarTerm[] t = {t1, t2, t3};
>
>         Structure percept = new Literal("terrain");
>         percept.addTerms(t);
>
>         //logger.info(percept.toString());
>
>         addPercept(agName, (Literal) percept);
>     }
>
> It adds a belief in the form;
>
> terrain((false, 0, 3), (true, 4, 1), (false, 2, 2))
>
> However, this does not come up in the agent belief base when I check
> the mind inspector.
>
> Firstly have I implemented the percept correctly and what is the best
> way to deal with this in the AgentSpeak,
>
> Regards,
>
> Edd
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by:
> High Quality Requirements in a Collaborative Environment.
> Download a free trial of Rational Requirements Composer Now!
> http://p.sf.net/sfu/www-ibm-com
> _______________________________________________
> Jason-users mailing list
> Jas...@li...
> https://lists.sourceforge.net/lists/listinfo/jason-users

-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

Hi all,

The message below was sent to the jason-users list by Edd Holliday
(again discarded as spam as the email in the from field is not the
same as registered in the list).

Cheers,

Rafael

---------- Forwarded message ----------
From: Edd Holliday <e.j...@du...>
To: <jas...@li...>
Date: Mon, 13 Apr 2009 14:29:47 +0100
Subject: Adding percepts as a list
Hi,

I am trying to add percepts to agents as a list, below is the Java
code from the Environment Class;

public void updatePercepts(String agName, Tile tile1, Tile tile2, Tile tile3) {
        clearPercepts(agName);

        VarTerm t1 = new VarTerm("("+tile1.isPath()+",
"+tile1.getCount()+", "+tile1.getTC()+")");
        VarTerm t2 = new VarTerm("("+tile2.isPath()+",
"+tile2.getCount()+", "+tile2.getTC()+")");
        VarTerm t3 = new VarTerm("("+tile3.isPath()+",
"+tile3.getCount()+", "+tile3.getTC()+")");

        VarTerm[] t = {t1, t2, t3};

        Structure percept = new Literal("terrain");
        percept.addTerms(t);

        //logger.info(percept.toString());

        addPercept(agName, (Literal) percept);
    }

It adds a belief in the form;

terrain((false, 0, 3), (true, 4, 1), (false, 2, 2))

However, this does not come up in the agent belief base when I check
the mind inspector.

Firstly have I implemented the percept correctly and what is the best
way to deal with this in the AgentSpeak,

Regards,

Edd

Hi,

I'll try to describe a more detailed procedure to launch a distributed  
MAS using either JADE or SACI.

Note that normally there are network issues like firewall, host's  
names resolution, ... The configuration I used to run/test the example  
was:
- no firewall
- local network (wireless)
- two MacOS X machines (named pandinha and jomilze)
- java 6 in all machnes
- shell command to start java programs (you may adapt it to your IDE)

Contents:

1. the MAS
2. running on JADE
3. running on SACI


1. the MAS

a very simple MAS is used.  The project:

MAS dist {
   infrastructure: Centralised
   agents:
       bob   at "jomilze";
       alice  at "pandinha";
}

Code of bob

!start.
+!start : true <- .wait(1500); .send(alice, tell, hello).
+hello[source(A)] <- .print("received hello from ",A).

Code of alice

+hello[source(A)] <- .print("received hello from  
",A); .send(A,tell,hello).

The output of running this MAS in centralised infrastructure is:

Running project dist
[alice] received hello from bob
[bob] received hello from alice


copy the MAS project to all hosts you will use


2. running on JADE

a) you need to setup the classpath in the command shell of every host.  
The following jars have to be included:
jason.jar
jade.jar
http.jar
jadeTools.jar
commons-codec-1.3.jar

in my hosts, I typed:

export JASON=/Users/jomi/Jason-1.2
export CLASSPATH=$JASON/lib/jason.jar:$JASON/lib/jade.jar:$JASON/lib/ 
http.jar:$JASON/lib/jadeTools.jar:$JASON/lib/commons-codec-1.3.jar

b) start Jade main container in host pandinha
	java jade.Boot -gui

   i.e. open a terminal/command shell/... in host pandinha, setup the  
classpath and run java jade.Boot

c) start alice in host pandinha
	cd <the directory of the jason project>
	java jade.Boot -container -host pandinha  
'alice:jason.infra.jade.JadeAgArch(alice.asl)'

d) start bob in host jomilze
	cd <the directory of the jason project>
	java jade.Boot -container -host pandinha  
'bob:jason.infra.jade.JadeAgArch(bob.asl)'

Note that the project file (.mas2j) is not being used here, all the  
startup is manually performed. The reason is that we still do not  
implement remote creation for jade.


3. running on SACI

a) edit the .mas2j project file changing the infrastructure to Saci

MAS dist {
   infrastructure: Saci
   agents:
       bob   at "jomilze";
       alice  at "pandinha";
}

b) using jEdit, run the system once. Two created files are important  
here: bin/build.xml and dist.xml. The former is the Ant script to  
launch saci and the latter a SACI script to launch agents.

c) there is a bug in the SACI infrastructure that do not include the  
hots in the dist.xml file. So edit it adding the host attribute, as  
follows:

<startAgent
		name="bob"
		society.name="dist"
		class="jason.infra.saci.SaciAgArch"
               host="pandinha"
		args="jason.architecture.AgArch jason.asSemantics.Agent  
'jason.bb.DefaultBeliefBase'  '/Users/jomi/dist/bob.asl' options  
verbose=1" />
<startAgent
		name="alice"
		society.name="dist"
		class="jason.infra.saci.SaciAgArch"
               host="jomilze"
		args="jason.architecture.AgArch jason.asSemantics.Agent  
'jason.bb.DefaultBeliefBase'  '/Users/jomi/dist/alice.asl' options  
verbose=1" />

(this bug will be fixed for the next release of Jason, Jason > 1.2)

d) copy bin/build.xml to bin/c-build.xml

e) edit bin/c-build.xml adding a new task (change the parameter  
accordingly your configuration -- path to saci.jar and master host):

  <target name="saci-client">
     <java classname="saci.launcher.LauncherD" failonerror="true"  
fork="yes" dir="${basedir}">
        <classpath refid="project.classpath"/>
        <jvmarg value="-Djava.security.policy=jar:file:/Users/jomi/ 
Jason-1.2/lib/saci.jar!/policy"/>
        <arg line="-connect jomilze "/> <!-- jomilze is the master  
hosts here -->
     </java>
  </target>

f) copy the project to all machines you will use

g) start the SACI network.

g1) start master host on jomilze
	ant -f bin/c-build.xml saci

the output is
   [java] Universidade de Sao Paulo
   [java] Departamento de Engenharia de Computacao e Sistemas Digitais
   [java] Laboratorio de Tecnicas Inteligentes
   [java]
   [java] Simple Agent Communication Infrastructure
   [java]            http://www.lti.pcs.usp.br/saci
   [java] -----------------------------------------
   [java] (saci 1.0.12   build on 2005/04/18 15:29)
   [java]
   [java] Copyright (C) 2001-2004 Jomi Fred Hubner
   [java] This is free software, and you are welcome
   [java] to redistribute it under GNU GPL conditions
   [java]
   [java] starting rmiregistry...ok

the error that follows may be ignored.

g2) start SACI on other machines. On pandinha:
	ant -f bin/c-build.xml saci-client

h) start the MAS. On some SACI machine:
	ant -f bin/c-build.xml
	

-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

if the Jade sniffer is not appropriate for you, it is quite easy it  
code yours. The example/sniffer shows how it can be done. However, as  
jade sniffer, it is more related to exchange of messages. Maybe you  
can filter  'achieve' messages to select exchange of goals...

You may also consider writing you own mindinspector, the code is in  
src/jason/control/ExecutionControlGUI.

HTH,

Jomi


On Apr 8, 2009, at 4:48 AM, Rafael H Bordini wrote:

> Hi all,
>
> The attached message was discarded, I think Christian is not in the  
> list.
>
> Christian, I don't think we'll be able to help much. If it's the  
> JADE tools you're interested in, it's the JADE documentation/support  
> you should search for. But, in case this helps... One of the JADE  
> tools I know of and find quite useful, and seems close to what you  
> need, is the JADE "sniffer" which shows the sequence of message  
> exchange among agents. But as I said I think you'll need to look  
> into the JADE documentation on how to run it.
>
> Cheers,
>
> Rafael
>
>
> From: Christian Guttmann <chr...@gm...>
> Date: April 7, 2009 4:15:20 AM CEDT
> To: jas...@li...
> Subject: Visualise agent behaviour in Jason
>
>
> Hello,
>
> I have been using Jason for a while now, and would like to visualise  
> the exchange of goals and beliefs between agents (and represent each  
> agent showing its goal and belief state, and ideally how messages  
> "move" from one agent to another).
> The debugging window offers great features, but is not very  
> demonstrator friendly and you can only inspect one agent at a time.
> Are there existing tools that would visualise the exchange between  
> several agents (and their corresponding belief and goal base)?
>
> I think Jomi Hubner mentioned tools that were used for JADE in a  
> previous thread. Apparently, they could be used for Jason?
> Would it be possible to me more specific about these tools and where  
> they are located, please?
>
> Thank you,
>   Christian
>
> -- 
> Christian Guttmann, PhD
> http://www.csse.monash.edu.au/~xtg/
> +61 (0) 417375679
>
>
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by:
> High Quality Requirements in a Collaborative Environment.
> Download a free trial of Rational Requirements Composer Now!
> http://p.sf.net/sfu/www-ibm-com_______________________________________________
> Jason-users mailing list
> Jas...@li...
> https://lists.sourceforge.net/lists/listinfo/jason-users

-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner



-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

Hi all,

The attached message was discarded, I think Christian is not in the list.

Christian, I don't think we'll be able to help much. If it's the JADE 
tools you're interested in, it's the JADE documentation/support you 
should search for. But, in case this helps... One of the JADE tools I 
know of and find quite useful, and seems close to what you need, is the 
JADE "sniffer" which shows the sequence of message exchange among 
agents. But as I said I think you'll need to look into the JADE 
documentation on how to run it.

Cheers,

Rafael

---------- Forwarded message ----------
From:  <jas...@li...>
Date: 2009/4/6
Subject: Auto-discard notification
To: jas...@li...


The attached message has been automatically discarded.

---------- Forwarded message ----------
From: Christian Guttmann <chr...@gm...>
To: jas...@li...
Date: Tue, 07 Apr 2009 12:15:20 +1000
Subject: Visualise agent behaviour in Jason
Hello,

I have been using Jason for a while now, and would like to visualise
the exchange of goals and beliefs between agents (and represent each
agent showing its goal and belief state, and ideally how messages
"move" from one agent to another).
The debugging window offers great features, but is not very
demonstrator friendly and you can only inspect one agent at a time.
Are there existing tools that would visualise the exchange between
several agents (and their corresponding belief and goal base)?

I think Jomi Hubner mentioned tools that were used for JADE in a
previous thread. Apparently, they could be used for Jason?
Would it be possible to me more specific about these tools and where
they are located, please?

Thank you,
  Christian

--
Christian Guttmann, PhD
http://www.csse.monash.edu.au/~xtg/
+61 (0) 417375679

Hi Stephen,
Sorry about that, the documentation is wrong, we'll fix it. The Jason 
syntax is actually as in the version that works for you, that is, if 
there is a second brackets with a plan-body inside, that's the "else" 
part. So "if ... then ..." is
if (...) {...}
in Jason, and "if ... then ... else ..." is
if (...) {...} {...}
Perhaps we should change the second to a "ifelse" token or something 
like that.
Cheers,
Rafael

Stephen Cranefield wrote:
> The documentation for the class if_then_else has a comment stating that the syntax for this construct is:
>   if ( logical formula ) {
>      plan_body1
>   [ } else { plan_body2 ]
>   };
> 
> However, including the "else" results in a syntax error. If it is omitted, everything seems to work OK. For example, the following plan cannot be parsed if the "else" is included:
> 
> +game_started <- 
>     if (.random(X) & X<0.5)
>     { .print("I'm feeling nice today"); 
>       cooperate
>     }
>     { defect }.
> 
> Regards,
> Stephen
> 
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> Jason-users mailing list
> Jas...@li...
> https://lists.sourceforge.net/lists/listinfo/jason-users

Dear Stephen, the syntax of your code is correct. I tested it with  
Jason 1.2 without syntax error. Which version of Jason are you using?

Regards,

Jomi


On Apr 2, 2009, at 3:25 AM, Stephen Cranefield wrote:

> The documentation for the class if_then_else has a comment stating  
> that the syntax for this construct is:
>  if ( logical formula ) {
>     plan_body1
>  [ } else { plan_body2 ]
>  };
>
> However, including the "else" results in a syntax error. If it is  
> omitted, everything seems to work OK. For example, the following  
> plan cannot be parsed if the "else" is included:
>
> +game_started <-
>    if (.random(X) & X<0.5)
>    { .print("I'm feeling nice today");
>      cooperate
>    }
>    { defect }.
>
> Regards,
> Stephen
>
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Jason-users mailing list
> Jas...@li...
> https://lists.sourceforge.net/lists/listinfo/jason-users

-- 
Jomi Fred Hubner
ENS Mines Saint-Etienne
158 Cours Fauriel
42023 Saint-Etienne Cedex 02
France
http://www.emse.fr/~hubner

The documentation for the class if_then_else has a comment stating that the syntax for this construct is:
  if ( logical formula ) {
     plan_body1
  [ } else { plan_body2 ]
  };

However, including the "else" results in a syntax error. If it is omitted, everything seems to work OK. For example, the following plan cannot be parsed if the "else" is included:

+game_started <- 
    if (.random(X) & X<0.5)
    { .print("I'm feeling nice today"); 
      cooperate
    }
    { defect }.

Regards,
Stephen

Hello again,

I continue trying that works my example with Saci and do not obtain it, 
apparently it had a problem of compatibility of versions of java, and 
after the same version putting them to both machines it continues 
without going. Please, might you spend to me some easy example that 
works? Thank you.

Regards.