Not knowing the details, I can only quote "divide and conquer". If the
sets of facts (F) and rules (R) can be divided so that F = union(F1,
F2,... Fn) and R = union(R1, R2,...Rn) and these may be paired to run as
F1/R1,...Fn/Rn then you might be able to process even bigger Java programs.
Perhaps some of the investigative rules need only facts local to some
.java file or, perhaps, some package. I guess that only public/import
related issues would require a "global" view of things.
WL
Ernest Friedman-Hill wrote:
On Jun 27, 2007, at 3:22 PM, fan yang wrote:
I didn't take consideration of Q12 yet. I will look at it. But for
my case, I just tried to load facts and didn't try to fire the
rules. Do you have other advices for me? Thank you.
Pattern matching happens when working memory is changed -- i.e., all
the partial matches are formed when you assert the facts. When you
call "run", Jess already knows what rules to fire -- it's a fast and
efficient operation. So in fact this is precisely the solution to
your problem -- you just have to figure out how to properly implement
it.
On 6/27/07, Ernest Friedman-Hill <[EMAIL PROTECTED]> wrote: On Jun
27, 2007, at 1:47 PM, fan yang wrote:
> I have tried Jess Version 7.1a1 to load the same aforementioned
> facts file, I still got the following error without any luck.
> Ernest, you said that 7.1 parser would detect duplicate symbols and
> share them, how can I know this new function work? Need I give some
> commands to Jess and ask Jess parser detect duplicate and share them?
>
> If 7.1 parser will not be a solution for my problem, is there
> another way to fix or alleviate my problem? thank you.
Well, since there was a relevant new feature, that's the first thing
I mentioned; but there is a lot of old-school advice I can give, too.
The most important one is understanding and minimizing partial-match
count. For example, in the FAQ there is a question dedicated to this:
http://herzberg.ca.sandia.gov/jess/FAQ.shtml#Q12 . Have you taken
this into consideration yet? If not, that's where you go next.
>
> Fan
>
>
>
> Jess> Exception in thread "main" java.lang.OutOfMemoryError : Java
> heap space
> at jess.factory.FactoryImpl.newToken(FactoryImpl.java:27)
> at jess.Node2.tokenMatchesLeft(Node2.java:328)
> at jess.Node2.tokenMatchesRight(Node2.java :344)
> at jess.Node2.doRunTestsVaryLeft (Node2.java:381)
> at jess.Node2.runTestsVaryLeft(Node2.java:354)
> at jess.Node2.callNodeRight(Node2.java:234)
> at jess.Node1.passAlong (Node1.java:49)
> at jess.Node1TECT.callNodeRight (Node1TECT.java:40)
> at jess.NodeRoot.passAlong(NodeRoot.java:34)
> at jess.NodeRoot.callNodeRight(NodeRoot.java:14)
> at jess.FactList.processToken(FactList.java:31)
> at jess.FactList._assert (FactList.java:204)
> at jess.FactList.assertFact(FactList.java:178)
> at jess.Rete.assertFact(Rete.java:495)
> at jess.Jesp.loadFacts(Jesp.java:351)
> at jess.LoadFacts.call(IOFunctions.java :314)
> at jess.FunctionHolder.call(FunctionHolder.java:35)
> at jess.Funcall.execute (Funcall.java:325)
> at jess.Jesp.parseAndExecuteFuncall(Jesp.java:2190)
> at jess.Jesp.parseExpression (Jesp.java:453)
> at jess.Jesp.promptAndParseOneExpression(Jesp.java:303)
> at jess.Main.execute(Main.java:177)
> at jess.Main.main(Main.java:35)
>
>
> On 6/27/07, Ernest Friedman-Hill <[EMAIL PROTECTED]> wrote: It
> looks like your facts include a mostly symbols, with a lot of
> duplication. This is good news, because the parser in Jess 7.1
> (there's an alpha version available on the web site) will detect
> duplicate symbols and share them, which in your case could save a
> *lot* of memory. Try it out and see what happens!
>
> On Jun 27, 2007, at 9:34 AM, fan yang wrote:
>
> > Hi,
> >
> > I'm trying to use Jess to detect source code defects. The tool
> > first parse source code and generate facts which represent design
> > information such as classes, functions, attributes, inheritances
> > and invocations etc., then apply Jess rule to detect flaws. The
> > tools works on small size of source code. but It got OutOfMemory
> > error when it dealt with bigger projects such as JHotDraw
> > (SourceForge project).
> >
> > I have read Jess manual, mailing list and Jess in Action book, I
> > didn't get solution. Would you please tell me how to apply Jess on
> > large facts set. What is the best practice or procedure to make
> > Jess application use less memory. Or Is there a way to know how
> > many memory the different facts occupy?
> >
> > Here shows a little bit more details of my test case, can you
> > point me right way to avoid out of memory problem?
> >
> > JHotDraw project's facts file is about 4M bytes. My computer has 2G
> > RAM. I set maximum java heap size to 1G through JVM's -Xmx1024M, I
> > got OutOfMemory exception when I just tried to load facts using the
> > following code:
> >
> > (clear)
> > (batch "C:/output/project.clp")
> > (reset)
> > (load-facts "C:/output/project.facts")
> >
> >
> > Partial facts file:
> >
> > (package
> > (packageName org.jhotdraw.util)
> > (parentName nil))
> >
> > (classifier
> > (unqualifiedName StorableInput)
> > (package org.jhotdraw.util)
> > (classifierName org.jhotdraw.util.StorableInput)
> > (type class)(isAbstract concrete)
> > (isLeaf leaf)(isRoot root))
> >
> > (visibilityInPackage
> > (unqualifiedName StorableInput)
> > (package org.jhotdraw.util)
> > (classifier org.jhotdraw.util.StorableInput )
> > (visibility public))
> >
> > (access
> > (operation org.jhotdraw.util.StorableInput.retrieve)
> > (attribute org.jhotdraw.util.StorableInput.retrieve.ref)
> > (accessType nil))
> >
> > (operation
> > (unqualifiedName retrieve)
> > (classifier org.jhotdraw.util.StorableInput)
> > (operationName org.jhotdraw.util.StorableInput.retrieve )
> > (scope instance)(visibility private)
> > (polymorphism nil)(abstractOrConcrete concrete)
> > (constOrNonConst nonconst)(numberParameters 1))
> >
> > (parameter
> > (unqualifiedName nil)
> > (operationName org.jhotdraw.util.StorableInput.retrieve)
> > (parameterName nil)(order nil)
> > (direction return)
> > (parameterType org.jhotdraw.util.Storable))
> >
> >
> > Thank you.
> >
> > Fan
>
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
