Ok, here a simple example. The problem occurs after a fact that matched
the accumulate statement was deleted. Accumulate recognizes the deleted
fact. Here the example:
(deftemplate Responsibilities
(slot id (type STRING))
(slot name (type STRING))
(slot description (type STRING))
(slot source (type SYMBOL))
)
(deftemplate P_CostOfChange
(slot value (type FLOAT))
(slot owner (type OBJECT)) ;Id of the owner of the property
(slot source (type SYMBOL)) ;Origin of the property value: ArchE, User
(slot status (type SYMBOL)) ;Indicators for the property:nil, conflict
)
(deftemplate AskQuestion
(slot questionId (type SYMBOL)) ;id of the question text to use
(slot questionGroup (type INTEGER)) ;the ID to enable grouping
(slot parent (type OBJECT)) ;The element that triggered the question
(multislot affectedFacts) ;List of facts where this question should
be shown
(multislot parameters)
(multislot options) ;objects among which the user can choose
(multislot default) ;the default answers
(multislot answer) ;the answers the user gave
(slot answerAvailable (type SYMBOL)) ;Describes if an answer from
the user is available
(slot log (type SYMBOL)) ;When set the Q/A fact will be logged and
then removed
)
(deftemplate ResponsibilityRefinementRelation
(slot id (type STRING)) ;Unique id for a relation
(slot source (type SYMBOL)) ;Documents who inserted this relation
(ArchE/User)
(slot parent (type OBJECT)) ;Id of the parent in this relation
(slot child (type OBJECT)) ;Id of the child in this relation
)
(reset)
(defrule RemoveCostOfChange
(declare (auto-focus TRUE))
?pc <- (P_CostOfChange (owner ?r))
(ResponsibilityRefinementRelation (parent ?r))
=>
(retract ?pc)
)
(defrule __AskcostOfChangeQuestion__
(declare (salience 500))
?rr <- (accumulate
(bind ?res (create$))
(bind ?res (insert$ ?res 1 ?_20_own))
?res
(P_CostOfChange (value 0.0) (owner ?_20_own) (source ArchE)))
(test (> (length$ ?rr) 0))
(not (AskQuestion (questionId costOfChange) (answerAvailable nil)))
=>
(printout t ?rr crlf)
(bind ?def (create$))
(foreach ?r ?rr (bind ?def (insert$ ?def 1 "0.0")))
(assert (AskQuestion (questionId costOfChange) (questionGroup nil)
(parent nil) (affectedFacts ?rr) (parameters ) (options ?rr) (default
?def) (answer ) (answerAvailable nil) (log nil))))
(bind ?r1 (assert (Responsibilities (name "EBB"))))
(bind ?r2 (assert (Responsibilities (name "EBA"))))
(bind ?r3 (assert (Responsibilities (name "EB"))))
(assert (P_CostOfChange (value 2.0) (owner ?r1) (source ArchE)))
(assert (P_CostOfChange (value 0.0) (owner ?r2) (source ArchE)))
(assert (P_CostOfChange (value 6.0) (owner ?r3) (source ArchE)))
-------------------------------------
now, if you follow the procedure below, the error will happen. Hope this
helps.
Jess> (run)
Jess> FIRE 1 MAIN::__AskcostOfChangeQuestion__ f-0, f--2,,
(<Fact-2>)
==> f-7 (MAIN::AskQuestion (questionId costOfChange) (questionGroup
nil) (parent nil) (affectedFacts <Fact-2>) (parameters ) (options
<Fact-2>) (default "0.0") (answer ) (answerAvailable nil) (log nil))
1
Jess>
Jess> (assert (ResponsibilityRefinementRelation (parent (fact-id 1))
(child (fact-id 3))))
Jess> ==> f-8 (MAIN::ResponsibilityRefinementRelation (id nil) (source
nil) (parent <Fact-1>) (child <Fact-3>))
==> Activation: MAIN::RemoveCostOfChange : f-4, f-8
<Fact-8>
Jess>
Jess> (run)
Jess> FIRE 1 MAIN::RemoveCostOfChange f-4, f-8
<== f-4 (MAIN::P_CostOfChange (value 2.0) (owner <Fact-1>) (source
ArchE) (status nil))
1
Jess>
Jess> (retract 7)
Jess> <== f-7 (MAIN::AskQuestion (questionId costOfChange)
(questionGroup nil) (parent nil) (affectedFacts <Fact-2>) (parameters )
(options <Fact-2>) (default "0.0") (answer ) (answerAvailable nil) (log
nil))
==> Activation: MAIN::__AskcostOfChangeQuestion__ : f-0,,,
TRUE
Jess>
Jess> (run)
Jess> FIRE 1 MAIN::__AskcostOfChangeQuestion__ f-0, f--2,,
(<Fact-2> <Fact-1>)
==> f-9 (MAIN::AskQuestion (questionId costOfChange) (questionGroup
nil) (parent nil) (affectedFacts <Fact-2> <Fact-1>) (parameters )
(options <Fact-2> <Fact-1>) (default "0.0" "0.0") (answer )
(answerAvailable nil) (log nil))
1
Jess>
Jess> (facts *)
Jess> f-0 (MAIN::initial-fact)
f-1 (MAIN::Responsibilities (id nil) (name "EBB") (description nil)
(source nil))
f-2 (MAIN::Responsibilities (id nil) (name "EBA") (description nil)
(source nil))
f-3 (MAIN::Responsibilities (id nil) (name "EB") (description nil)
(source nil))
f-5 (MAIN::P_CostOfChange (value 0.0) (owner <Fact-2>) (source ArchE)
(status nil))
f-6 (MAIN::P_CostOfChange (value 6.0) (owner <Fact-3>) (source ArchE)
(status nil))
f-8 (MAIN::ResponsibilityRefinementRelation (id nil) (source nil)
(parent <Fact-1>) (child <Fact-3>))
f-9 (MAIN::AskQuestion (questionId costOfChange) (questionGroup nil)
(parent nil) (affectedFacts <Fact-2> <Fact-1>) (parameters ) (options
<Fact-2> <Fact-1>) (default "0.0" "0.0") (answer ) (answerAvailable nil)
(log nil))
For a total of 8 facts in all modules.
Jess>
Felix Bachmann wrote:
I did a little bit more research to try to find out what is going on.
I printed the fact id of all the facts that match the accumulate CE,
and there fact id's show up from facts that are already deleted. Here
is the rule that actually does the deletion:
(defrule RemoveCostOfChange
(declare (auto-focus TRUE))
?pc <- (ModifiabilityReasoningFrameworks::P_CostOfChange (owner ?r))
(MAIN::ResponsibilityRefinementRelation (parent ?r))
=>
(retract ?pc)
)
Looks like something is going terribly wrong.
Facts in the program are either inserted by calling assertFact from
java or the usual assert/modify/retract commands from Jess.
Not sure that this helps
Felix
[EMAIL PROTECTED] wrote:
I can't reproduce the problem with a simple program, so it's more
complex than just what we can see here. There have been a few patches
to "accumulate" over the last few releases, but none of them dealing
with anything like this. What this looks like to me is working memory
corruption due to out-of-band data changes -- for example, setting
slots value in a Fact directly rather than using "modify", or using
instances of a class with a mutable hashCode as slot data without
using set-nonvalue-class.
In the next release, by the way, the "value class" default is going to
be reversed, so that the slow, safe mode is the default, and the fast
but breakable mode is an option you must turn on explicitly.
I think Felix Bachmann wrote:
[Charset iso-8859-1 unsupported, filtering to ASCII...]
I have a little problem with the accumulate CE. I try to collect a
slot value of a set of facts in a list with the following accumulate
statement (produced with (ppdefrule). The interesting slot is the
"owner" slot. I provided a list of facts (Fact-246 - Fact-259) that
match the CE of the accumulate. As a result of this rule a fact is
inserted (MAIN::AskQuestion), which should contain the collected
list in slot (affectedFacts ?rr). The resulting fact Fact-466
contains duplicates and values that are not even in the matching facts.
Is this a bug, or what am I doing wrong here?
Kind regards
Felix Bachmann
--------------------------------- rule -----------------------
(defrule ImpactAnalysis::__AskcostOfChangeQuestion__
(declare (salience 500))
(MAIN::initial-fact)
?rr <- (accumulate
(bind ?res (create$))
(bind ?res (insert$ ?res 1 ?_20_own))
?res
(ModifiabilityReasoningFrameworks::P_CostOfChange (value 0.0)
(owner ?_20_own) (source ArchE)))
(test (> (length$ ?rr) 0))
(not (MAIN::AskQuestion (questionId costOfChange) (answerAvailable
nil)))
=>
(bind ?def (create$))
(foreach ?r ?rr (bind ?def (insert$ ?def 1 \"0.0\")))
(assert (MAIN::AskQuestion (questionId costOfChange)
(questionGroup nil) (parent nil) (affectedFacts ?rr) (parameters )
(options ?rr) (default ?def) (answer ) (answerAvailable nil) (log
nil))))
------------------ rule fires ---------------------
Jess> FIRE 1 ImpactAnalysis::__AskcostOfChangeQuestion__ f-0, f--2,,
------------------ facts immediately after this rule -----------------
f-246 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-106>) (source ArchE) (status nil))
f-247 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-105>) (source ArchE) (status nil))
f-249 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-103>) (source ArchE) (status nil))
f-251 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-101>) (source ArchE) (status nil))
f-252 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-100>) (source ArchE) (status nil))
f-253 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-99>) (source ArchE) (status nil))
f-254 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-98>) (source ArchE) (status nil))
f-255 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-97>) (source ArchE) (status nil))
f-256 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-96>) (source ArchE) (status nil))
f-257 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-95>) (source ArchE) (status nil))
f-259 (ModifiabilityReasoningFrameworks::P_CostOfChange (value
0.0) (owner <Fact-93>) (source ArchE) (status nil))
....
f-466 (MAIN::AskQuestion (questionId costOfChange) (questionGroup
nil) (parent nil) (affectedFacts <Fact-93> <Fact-106> <Fact-94>
<Fact-94> <Fact-95> <Fact-96> <Fact-97> <Fact-98> <Fact-99>
<Fact-100> <Fact-101> <Fact-102> <Fact-102> <Fact-103> <Fact-104>
<Fact-91> <Fact-91> <Fact-104> <Fact-92> <Fact-92> <Fact-105>)
(parameters ) (options <Fact-93> <Fact-106> <Fact-94> <Fact-94>
<Fact-95> <Fact-96> <Fact-97> <Fact-98> <Fact-99> <Fact-100>
<Fact-101> <Fact-102> <Fact-102> <Fact-103> <Fact-104> <Fact-91>
<Fact-91> <Fact-104> <Fact-92> <Fact-92> <Fact-105>) (default "0.0"
"0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0"
"0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0" "0.0") (answer )
(answerAvailable nil) (log nil))
[Attachment, skipping...]
---------------------------------------------------------
Ernest Friedman-Hill Advanced Software Research Phone:
(925) 294-2154
Sandia National Labs FAX: (925) 294-2234
PO Box 969, MS 9012 [EMAIL PROTECTED]
Livermore, CA 94550 http://herzberg.ca.sandia.gov
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
------------------------------------------------------------------------
begin:vcard
fn:Felix Bachmann
n:Bachmann;Felix
org:Software Engineering Institute - Carnegie Mellon University;SEI-PLS
adr;dom:;;4500 5th Avenue;Pittsburgh;PA;15213
email;internet:[EMAIL PROTECTED]
tel;work:+1 (412) 268 6194
tel;fax:+1 (412) 268 5758
url:http://www.sei.cmu.edu
version:2.1
end:vcard
begin:vcard
fn:Felix Bachmann
n:Bachmann;Felix
org:Software Engineering Institute - Carnegie Mellon University;SEI-PLS
adr;dom:;;4500 5th Avenue;Pittsburgh;PA;15213
email;internet:[EMAIL PROTECTED]
tel;work:+1 (412) 268 6194
tel;fax:+1 (412) 268 5758
url:http://www.sei.cmu.edu
version:2.1
end:vcard
