Dear all,
I like the proposal of "reifying" the remove action
as a separate fact. I haven't thought about that.
But I would rather go for option #2 if possible, given
that Jess already implements the kind of "time machine"
required for that. It is called "Rete network" and can
be activated by the magic "not condition" switch. ;-)
Using the existing features of the Rete match algorithm seems
more clean to me, because I wouldn't have to introduce
new facts. But in the case at hand I'm not sure if there
is a suitable rule with a not condition, that can achieve
the desired effect. Any ideas on this would be very welcome.
Cheers,
Alex
Ernest Friedman-Hill schrieb:
<div class="moz-text-flowed" style="font-family: -moz-fixed">Lars,
If you think about it, there are only two ways to write a rule to
react to the absence of data:
1) Remember what the data was, so you can compare the present
situation to the previous situation;
2) Use a time machine, travel back in time, find out what the data
used to be, then come back to the present.
Of these, only #1 is practical to implement in software. :) This isn't
a problem with Rete; it's a problem with the fabric of reality.
Unfortunate, but it's something we've got to live with. Unless you can
come up with a way to build a time machine, you have to remember the
past if you're going to act on it.
But Martijn actually gave you two suggestions. The first one is my #1
above, which you've responded to as being inefficient.
The second one is not to react to changes, but to *participate* in the
changes, as they're happening. You need to make the changes explicit.
I don't think I'd do it quite the way that Martijn suggests, though.
But the basic idea is the same: it's too late to react to the change
after the change has been made; you have to react to the change *while
the change is being made.* For example, you could use explicit command
facts to remove items from the multislot, and one or more rules to
process the commands. So instead of saying
(defrule remove-something-from-c
?c <- (C (m ?a ?b ?c))
=
(modify ?c (m ?a ?c)))
You'd say something like
(defrule want-to-remove-something-from-c
?c <- (C (m ?a ?b ?c))
=
;; Instead of modifying ?c, issue a command to do so
(assert (remove ?c ?b)))
(defrule actually-remove-things-from-c
"Process commands to remove items from multislot m of C facts"
(declare (no-loop TRUE))
?c <- (C (m $?first ?x $?rest))
(remove ?c ?b)
=
(modify ?c (m $?first $?rest)))
(defrule do-something-when-something-gets-removed-from-c
"This is the rule you were asking about how to write"
?c <- (C (m $?first ?x $?last))
(remove ?c ?x)
=
(do whatever you want to do here))
(defrule clean-up-remove-commands
"A low salience rule to clean up remove commands when you're done
with them."
(declare (salience -1))
?r <- (remove ? ?)
=
(retract ?r))
On Feb 20, 2008, at 5:09 AM, Lars Braubach wrote:
Hi,
thanks for your reply Martijn. The solution should work
but is also rather inefficient because all facts would have
be stored twice. Maybe some Rete expert could argue, if such
a rule could be expressed direcly in a rete network (not considering
the specification language). If this is possible it might indicate
that just the CLIPS/JESS language is insufficient for expressing such
kind of behavior.
Kind regards,
Lars
Martijn Tromm schrieb:
Hi,
You must either keep the same data in another slot for reference like:
(C (m a b c) (m-old a b c d)) Now you can compare old and new and
detect changes between. (actually you store the mutation this way)
Or I think you have to store each object in a separate fact and keep
track of changes through activations (addition) or retractions and
logical dependencies (removal).
Regards,
Martijn
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of Lars Braubach
Sent: dinsdag 19 februari 2008 10:34
To: [email protected]
Subject: JESS: Rules for detecting changes of a multislot
Hi,
I am trying to design a rule that is able to detect
when a single fact of a multislot is added/removed.
Given that the contained multislot values are
backed with Java objects I would also like to
know which fact has been added/removed.
(deftemplate C (multislot m))
(deffacts facts
(C (m a b c))
)
For the addition of values I could figure out
a rule that does what I want, but its unclear
to me how this could be achieved for the removal
of values. (The addition rule simply uses a pattern
(? ?fact ?) in order to generate bindings for
each element of the multislot.)
Any ideas how this could be achieved?
Kind regards,
Lars
--
Psssst! Schon vom neuen GMX MultiMessenger gehört?
Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users
[EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify owner-jess-
[EMAIL PROTECTED]
--------------------------------------------------------------------
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.516 / Virus Database: 269.20.7/1286 - Release Date: 18-2-
2008 18:49
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.516 / Virus Database: 269.20.7/1286 - Release Date:
18-2-2008 18:49
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify
[EMAIL PROTECTED]
--------------------------------------------------------------------
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify
[EMAIL PROTECTED]
--------------------------------------------------------------------
---------------------------------------------------------
Ernest Friedman-Hill
Informatics & Decision Sciences Phone: (925) 294-2154
Sandia National Labs FAX: (925) 294-2234
PO Box 969, MS 9012 [EMAIL PROTECTED]
Livermore, CA 94550 http://www.jessrules.com
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify
[EMAIL PROTECTED]
--------------------------------------------------------------------
</div>
--------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the list
(use your own address!) List problems? Notify [EMAIL PROTECTED]
--------------------------------------------------------------------
