Okay, as mentioned previously, drools is 100% New Code, still.
Instead of starting with Natural Language functionality, though,
I'm opting for the simpler java syntax, interpreted at run-time.
Instead of using BeanShell or some other black-box expression
interpreter, I'm of course, rolling my o
Testing the jab.org mailarchive, again...
-bob
___
drools-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/drools-interest
Hokie dokie.
Alpha-3 is now out, and is the first announced version of the new Rete-OO
method of doing things.
Undocumented and virtually untested.
See test_data/Minimal.drl for an example rule. It's the one we've
seen in the mail-list here recently.
Many things don't work (initialization up
After making radical changes, I decided that the drools 1.x series was
purely for research, and Real Work begins with drools 2.x.
Anyhow, I just imported my most recent ramblings, which is a semantic-model
independent implementation of the Rete-OO algorithm. Will soon be providing
a semantic mo
Okay everyone. Go sync down the drools-2 CVS module.
./build.pl clean all test
Watch some rules get applied to XML documents.
There's org.drools.RoughDom4jTest which looks for two documents:
doc #1
...
doc #2
...
One must
Now that I'm playing with the dom4j example more, I'm starting to see
the possibilities.
...Build on top of JMS, and have an XML router that accepts documents,
and publishes them to different topics based upon their root elements,
and values of attributes and/or elements.
...Build an agent that
> Hello, my name is Marc Carrion,
Howdy Marc!
> I'm working in a project that needs some inference engine, by now
> Drools is the framework that fill more of our requests. We would like
> to colaborate with you, if possible, in the developing of Drools.
Absolutely. All are welcome.
Right n
Yeah, possibly. I need to execute a personal JSCP agreement with Sun,
because I don't think The Day Job has any interest in me doing this on
company time (unlike the JDOM JSR).
So far, I haven't been intrigued, due to a lack of information as to
what they are really doing.
Any clue?
-
Testing new archiver...
___
drools-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/drools-interest
Initial draft of the website. (copy/paste from Jaxen, so let
me know if I forgot to edit something correctly).
http://drools.org/
Also, released 1.0-alpha-1. Pretty much a useless release,
but had to have something to point the website/download.html
at.
-bob
___
!! warning: long and probably boring !!
Just wanted to commit some thoughts to archives before
I forget this.
Haven't seen this in print anywhere, but I'm not claiming
it's necessarily new. Just an epiphany on my part.
Normal RETE:
Most things are us
> Sounds cool. Other nice languages to be able to read for your rule base
> could be...
>
> Java code.
>
> person.getName().equals( "James" );
>
> XPath..
>
> //person[name='James']
>
> Or another thing to track could be RuleML which seems to be an XML markup
> language for specifying
Howdy John (thanks for the pointer deref, James)--
drools is still trying to find its way, at this point.
It started as an independant project, but now it's looking like
(hopefully) my Day Job will buy into it also.
While the current code-base is 'from scratch', I may actually
start over again
t; -Original Message-
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]] On Behalf Of bob
> mcwhirter
> Sent: October 13, 2001 10:30 AM
> To: James Strachan
> Cc: John P.Sheppard; [EMAIL PROTECTED]
> Subject: Re: [droolers] Re: A question pertaining to DROOL's
>
&g
> I finally had some time to look at this in detail. While I can't say I
> understand it completely yet I do understand enough to see that it looks
> pretty good so far.
>
> I was going to do the UML class diagram's for it; but you mentioned that
> you might abandon the current implementation for
> [ re: shipment rating, carriers, tariffs, etc ]
Gosh, you're in logistics too? May I ask for whom? I felt like I was
reading mail that I'd sent myself...
> > It started as an independant project, but now it's looking like
> > (hopefully) my Day Job will buy into it also.
>
> That sounds
Hokie dokie.
Day Job hasn't bought into the rules-engine idea Just Yet, so I'm going
to proceed down the drools path as 100% new development.
Due to .Net considerations and such, I *will* make the rule parser a
modular component. As long as something implements Parser, and builds
expression tr
Yah, that's about what I've gotten out of the proposal.
I see that it narrowly got approved, "with strong reservations" from
most of the voting members.
But still, what does it mean?
Is it a standardization of the assert(..) retract(..) and
modify(..) actions of working with facts within a w
Just curious, did someone update the jaxen-core.jar in the lib/ directory.
Two days ago, the build worked. Now it doesn't.
Anyhow, I've removed jaxen-core.jar, and we only have jaxen-full.jar
in the lib/ now.
-bob
___
drools-interest maili
Okie dokie...
Today, I think I'm going to werk on drools some, particularly
the Java Semantic Module.
An adjunct to that would be a java-centric 'rules file' format.
I'm thinking, though, that the action of a rule still should be
compiled code, at this point. Thus, I'm pondering a format
like
On Thu, 6 Dec 2001, Stuart Schmukler wrote:
> Bob,
>
> The format you propose is interesting. I believe that it is
> on flexible enough. Both the condition and the action should be
> expressions, otherwise the format is just some kind of dymanic
> preprocessor.
>
> I believe that one of your
Okay, I'm still going to need access to the Java expression AST. I don't
really want to write a Java statement evaluator all by myself, so I think
I'm going to use the ANTLR Java grammar, and instrument the AST-walker.
I know there's an issue with antlr.jar conflicting with an older version
tha
Okay...
It should now be possible to write rules with Java semantics. We
still don't have a convenient rules-file interface, but using the
API, you can create several types of Java FactExtractors.
There's a Darn Simple One, where you get an object, cast it, and do
whatever you want to with it.
Here's the plan:
Using the antlr-based java grammar for the basis of the java rules file.
But, once I've parsed and analyzed what I need, we'll actully pass all
execution over to beanshell (beanshell.org), so as to avoid having to
write an interpreter ourselves.
Following this method, I think t
> A moderately complex rule-based task that comes to mind is a
> calendar-scheduling sytem. It should respect holidays, find open times
> for meetings and resources, handle recurring events, schedule subsequent
> events based on current event type, etc...
Ayup, that'd be cool. Due to my Day Job
On Sun, 9 Dec 2001, bob mcwhirter wrote:
> > What will/does the java rules file syntax look like?
>
> Roughly:
>
> ruleset MyRuleSet
> {
> rule myRule(Foo rootFact1, Bar rootFact2)
> {
Nope, I lied. At least Right Now, it'll be an XML format.
On 9 Dec 2001, Nathan E Probst wrote:
> Awww...phooey! Why the sudden change of heart? I like the looks of the
> syntax below very much. XML, I suppose, is easier to parse, etc... Is
> that temporary arrangement?
Mostly due to ease of implementation to get something going.
Nothing will limi
The bsf integration is going pretty smoothly.
We now have a BeanShellAction that runs your script (a String) through
BeanShell, letting actions be a run-time thing.
Looking like I can dedicate most of this week to drools, so hopefully
we'll have something stable and useful by week's end.
I'm
> We wanted that the user will be able to define his owns rules, the facts
> will be represented in rdf (this will be our model, but is easy to
> transform to ruleml). Our main problem is the velocity, we need a very fast
> rule engine, so we will try with different engines, maybe we will develo
On Tue, 11 Dec 2001 [EMAIL PROTECTED] wrote:
>
> For me velocity is the priority. I don't care about the rest, my users
> will not use the engine directly, they will have an easier interface.
>
> Just another question. Is it posible to have a fact database and a rule
> database and at the e
On 11 Dec 2001, Nathan E Probst wrote:
> Well, I've always thought that Easy of Use trumps Speed. You can buy
> faster hardware, but programmer productivity is relatively fixed!
That's my take on it all, too.
> Since this is an OSS project, Extensibility and Modularity are definate
> virtues
Okie dokie--
The integration with BeanShell/ANTLR seems mostly complete.
In CVS, look at $ROOT/test_data/ruleset.xml, which I've handily
reproduced here.
I still need to create a friendly API for building RuleSets, and
such, but, we can now read .xml files that contain rules.
Filter condition
[[ cc:'d to the list ]]
> I'm having problems trying to access to cvs, we are using a proxy that
> almost never works :(
Hmmm... Then tarballs work best for you, right?
I'll see if I can't get nightly snapshots auto-building for ya. Might
be a few days, though.
James-- You have
Well kids, I'm unplugging for the night.
I think the Java Semantic Module is Done. I've included ease-of-use
features.
Tomorrow, I'll work on examples and end-user documentation (as opposed
to simply the nearly-complete JavaDocs).
Now's the time to start banging on it, and let me know what yo
> I am very interested in making a workflow engine (as I'm sure others on this
> list are interested in doing) and I wanted to know if anything has started
> in earnest based on drools?
Probably not, due to the ethereal nature of drools up until Today.
> In the project I work on (Tambora, tambor
> Is there an IRC room anywhere for drools?
Not as of right now.
Isn't there some open-source IRC network? If so, could you blurp me
some connection information, and we could pull up #drools.
-bob
___
drools-interest mailing list
[EMAIL
Okay, pushed out a zipfile and tarball of 2.0-beta-1.
Have fun. Some examples (and probably beta-2) will come tomorrow.
-bob
___
drools-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/drools-interest
Thanks to the suggestion by Jason, I've located an IRC network
I think we can use.
http://openprojects.nu/irc_servers.shtml
I'll probably be idling in the #drools channel.
-bob
___
drools-interest mailing list
[EMAIL PROTECTED]
http
Okay...
In CVS now is a Supply-and-Demand example, which attempts to match
up buyers and seller.
./build.pl examples
Anyhow, this has shown me that we might need some more functionality
in the core engine.
While it does successfully match a buyer who's willing to pay what a
seller wa
Howdy folks..
Just added RuleBase.createTransactionalWorkingMemory() which creates,
well, a TransactionalWorkingMemory.
The TWM extends normal WorkingMemory, but adds commit() and abort()
methods. It basically just caches assert/modify/retract calls
until commit() is called, and then performs
Okay, hopefully in the next 6-24 hours, according to sf.net, we'll have
a [EMAIL PROTECTED] list for watching commits to the codeline.
If this works, then I'll go ahead and modify my other projects to also
do this.
Still haven't figured out a handy way to do nightly builds.
-bob
tion.
-bob
>
> I hope this helps. I'll try to be clearer about the explanation
> when I figure out a better way to express it in terms of the DROOLS
> engine.
>
> Stuart
>
> BTW, ILOG has introduced 'collections' and collection rules to
> en
Okay, you can now subscribe to drools-commits, and see the changes fly by.
I guess this means I need to start writing better commit messages.
-bob
___
drools-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinf
For those of you doing side-by-side evaluations of rules engines, I've
added an example to the drools distribution the models the shopping-cart
in a fish-store example that ILOG uses in one of their whitepapers.
It's in CVS now, and will be available in a daily build tonight at
12:40am eastern t
New IRC server. OpenProjects IRC servers were weak, and Jon Stevens
(of jakarta-apache fame) has offered the use of his IRC server:
server: irc.whichever.com:6667
channel: #drools
-bob
___
drools-interest mailing list
[EMAIL PROTE
After talking to Stuart and Jason and Dave, along with reading the
various JRules papers, it seems there are a few concepts of time:
1) Time as a condition event:
when:
cinderalla.getLocation() == ROYAL_PALACE;
clock.strike
Okay, drools now handles temporal rules of one flavor (more flavors
to come lateron).
To a block, you can now add a sub-element
which becomes a durational condition.
ticket.getStatus() == Ticket.NEW
> To a block, you can now add a sub-element
> which becomes a durational condition.
>
>
>
Actually, there are several attributes you can use, in any combination:
-bob
___
drools-interest mailing list
[EMAIL
I'd like to start a quick discussion...
I'm thinking of adding to conditions, to allow for a rule to
only match at a particular time:
bob.isIdle()
This rule would only be checked at 10am and 2pm.
Al
> I think that it is a mistake to fix the time unit to milliseconds.
> The having an external method of setting the time removes the limitations
> of the processor speed.
I'm not certain that I follow.
Plus, I've actually opted for only 1-second granularity. Good enough
for government work?
> T
> I hope that everyone had wonderful holiday and that work on Drools will
> continue...
drools is paused, since I've no customers asking for modifications
at the moment. As soon as someone needs work done on it, I'll gear
back up.
Meanwhile, I'm knee-deep in genetic algorithms, if anyone has a
Right.
drools currently does create 2 different nodes in the rete network;
one for each condition.
Some extra intelligence could be added to consolidate those conditions
which are simply different flavors of the same fact (ie, negated and
not negated).
-bob
On Mon, 14 Jan 2002, Stuart
To be honest, no clue.
JESS still seems non-OO, to me, in that you're still writing
cambridge-prefix types of rules. It's interpreted, as
is the Java Semantic Module of drools, so it may be similar.
I do think JESS has highly optimized their Rete network,
though, so they may beat us.
Someone
> This is a new area for me ( but exciting! ).
Glad to have you on board!
> I'm experimenting with using drools for filtering datasets.
>
> In my case so far, I have basically 4 rules with varying parameters.
>
> One of the rules looks something like this:
>
> loop over a collection of beans
> Whats up with the drools.org site i havent been able to connect to it for
> nearly a week.
Yikes! You're right. My nightly-build cron job must be futzing.
I'll try to get it back online in the next day or so.
-bob
___
drools-interest ma
> Maybe I have misunderstood things, but I think this is a bit odd.
>
> If I have a method getName() in a java object and use that as a
> condition, like this:
>
> someObject.getName().equals("");
>
>
> and then modify that object in the actions of another rule, like
> It looks like you can only define rules in XML - do you plan to support
> any other representation in the future ?
Howdy Robert--
There's support for any one to write any representation they want.
I did XML originally because:
1) dom4j (by your friend'n'mine, James Strachan) makes
fyi--
www.drools.org is back online.
-bob
___
drools-interest mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/drools-interest
> Just downloaded drools 2.0 beta 1, but having trouble
> understanding how to get started. I can tell by the mailing archives
> that there are people out there evaluating rules and using the tool, so
> I must be missing something. Is there a 'Cooks tour' or some sort of
> Hello World docu
> I'm really using it. It's part of a very large web application. On my list
> of things to do is add a non-programmer type language for descibing rules.
> Probably with some sort of web-based UI.
Woo-hoo!
> Some examples of using just the API would be useful.
When you say "using just the AP
> I was just reading the Drool mailing list and came
> across a posting on workflow. I was wondering if
> anyone had started an effort on a workflow engine.
I think Jason van Zyl was originally planning on doing it, and I was
hoping to assist. As far as I know, thus far, nothing has been done.
Pretty dumb order-in-which-loaded priorities for firing rules now exist
in CVS. Will Some Day make it more flexible so you can set your own
priorities, but for now, only the order the rules are loaded into the
rule-base count. This includes order of the rules-files, and order
of rules within th
Martin--
Thanks for sharing this with everyone. Having RealWorld(tm) examples
of drools is a very good thing.
I'll incorporate your changes so you don't have to work with a modified
code-base. In the future, if you need a feature, just ask, and I'll
try to hack it into the CVS mainline for yo
On Fri, 29 Mar 2002, Stuart Schmukler wrote:
> I have been trying to get up to speed with DROOLS now that I have the time.
> It seems that the cvs.drools.org does not exist. It is mentioned on the
> Drools.org pages.
Yah, I think sourceforge moved servers around, and broke that link.
I think
Does anyone, right off hand, know what might've spiked a sudden interest
in drools? Looking at the sourceforge.net statistics page, we went from
an average of 3-4 downloads a day to 126 downloads yesterday.
Any clue who's talking about us or linking to us?
-bob
To everyone addressed in one way or another by this message...
I'm back in The States now from Amsterdam, much relaxed.
Once I readjust my internal clock by 6 hours, I'll be functional
and working on the various projects again.
-bob
___
Re: Forgy's paper.
I was unable to locate a digital copy anywhere. You *may* be able to
contact Carnegie-Mellon Univerisyt (cmu.edu), and try to get a copy of
his thesis, or similar.
-bob
___
Don't miss the 2002 Sprint PCS
Ayup.
-bob
On Mon, 13 May 2002, James Strachan wrote:
> Just checking; is the drools-2 the correct CVS image to work from these
> days?
>
> James
>
>
> _
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.
Howdy folks--
Catching up on email, and I've had lots of requests for the examples,
etc. They didn't make it into the tarball, but I figured everyone
was using CVS. Apparently not.
So, this week, I'm going to jump back into drools, and try to press
out another beta tarball, complete with exam
Howdy--
Please just send plaintext mail to the list, as HTML can be difficult
for some folks (ie, me) to deal with.
fwiw, drools tarballs are out-of-date, especially with regards to the
examples. I plan to correct that soon, but until I do, you may desire
to simply use drools from the CVS repo
Not certain what's up with all the messages and recalls, but...
The attachableNodes is a set of all nodes that still need to be
attached to the RETE network. Basic premise is to iterate over
the attachableNodes set until it's empty, which indicates that
all nodes have been attached into the act
Oh, to further explain...
Sometimes, we'll take 2 nodes out of the attachableNodes set,
create a JoinNode with the 2 nodes as children, and place the
join back into the set, as it hasn't yet been glued to the tree,
I think.
ie:
## Start with 2 nodes...
an = { node-A, node-B }
## Remove them
AssignmentNodes are *sometimes* the column-generator nodes.
name = person.getName()
That takes a table with only a 'person' column, and it generates a
matching 'name' column, populated with the name of the person.
AssignmentNodes are *sometimes* the join-nodes.
name = otherNam
In old-style (ie, traditional lisp/cambridge-prefix) rules engine,
that defines a rule that finds cousins.
The "(defrule cousin" starts the rule, and names it.
It's followed by 3 patterns that define what matches the rule.
The first two patterns are asserting a parent/child relationship:
Actually, AssignmentNodes are never JoinNodes. But, things that
look like assignment statements may result in either an AssignmentNode
or a JoinNode.
Just to clarify.
-bob
On Mon, 27 May 2002, bob mcwhirter wrote:
>
> AssignmentNodes are *sometimes* the column-generator
> I am interested in the use of Drools, and I have read the license in
> CVS and it seems to be compatible for the use together with an LGPL API.
> Am I right or have I not well interpreted the Drools' license?
I believe that the ASF/BSD license is compatible with the LGPL.
You may want to chec
A duration is a way to say how long a rule must hold true before
firing.
ie, you may have a rule that only 100 people can fit in this room.
But, you don't call the cops until more than 100 people have been
in the room for over 5 minutes.
With the escalation example, the duration basically is us
The way RETE works, is you start with a single root node in
your tree. The tree gets wider as you add various column-generator
nodes to it, and starts to get narrowed back down as your
use join-nodes to combine 2 nodes into 1 in the next level of
the tree.
Eventually, the last level will contai
On 2 Jun 2002, Jason van Zyl wrote:
> Hi Bob,
>
> If you can give me the versions of the dependencies that drools has I
> will whip up a Maven descriptor.
antlr-2.7.1
bsh-1.2b3
dom4j-core (1.1 or whatever is latest)
jaxen-1.0-FCS
saxpath-1.0-FCS
For those of you who aren't in the know...
Maven is Jason van Zyl's new Jakarta project that makes building projects
a lot easier and nicer.
Maven will end up making projects a lot more consistent in their build
and release methodologies, which is a Good Thing.
So, drools CVS source code may
> Here's the first crack at a POM. The repository will have to be
> structured a bit and the POM updated slight. If you give me access I
> would be glad to do a little reorg. Good prep for the workflow endeavor
You've now got developer Karma for drools-2 CVS on sourceforge.
Feel free to go ahead
Howdy folks--
drools has now been almost completely mavenized.
So, if you want to build drools these days, you'll need maven:
http://jakarta.apache.org/turbine/maven/
You'll also need ant installed globally on your machine, along
with junit in ${ant.home}/lib. No more build.pl script
> Is there any attempt being made to maintain interface compatibility with
> JSR 94 ?
I haven't seen the final spec for JSR 94, but from what I had learned,
a while ago, it really doesn't address many aspects of rules engines.
> In other words, if I wrote my application to use Drools, and I wa
> I'm going to have a crack at a semantic module for drools to allow it to
> process RDF data. I'm going to have to do a fair bit of experimentation, but
> my starting points are going to be :
>
> aiming towards input/output from drools in RDF, and manipulation of drools'
> wm based on RDF declar
On Wed, 3 Jul 2002, Danny Ayers wrote:
> Hi folks,
> Just been trying to get Maven going, wasted a lot of time with the download
> packages so now trying to get the cvs. Unfortunately it keeps timing out on
> the login. Does this look ok? :
Apache seems to be having CVS issues, at the moment.
W
Due to various reasons, I'm currently in the process of moving my projects
away from SourceForge.
drools will get a new home at
http://drools.werken.com/ sometime soon.
Meanwhile, I've already discontinued using sf.net's CVS and this will be
the last message that I send/read to the dr
> Hope all is well.
Yah, finally got drools compiling against the latest Maven.
> Quick question - where's the package fr.dyade.jdring from?
www.dyade.fr, somewhere, I'd guess. ;)
It's a scheduler. And I think we're not using it anymore, since
JDK 1.3 has a lightweight scheduler built in. O
On Thu, 18 Jul 2002, Danny Ayers wrote:
> Hope all is well.
btw, let's move this discussion over to the new home at:
[EMAIL PROTECTED]
---
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Great thoughts. If you don't mind, head to http://jira.werken.com/
and file them there.
> 2. Abstract Interpreter
> If the scripting bit in org.drools.semantic.java was abstracted out some
> more (i.e. having a drools Interpreter interface etc.) then it should be
> straightforward to add other
t; mahi
> >
>
>
>
>
> ---
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> ___
>
t; > Thanx
> > mahi
> >
>
>
>
>
> ---
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______
> drools-inter
Howdy--
Since you're still using the 'old' sourceforge list, I'm not convinced
you're using the latest code.
If you don't mind, please subscribe to the new list (info available
through http://drools.org/) and repost your questions/bugs once you've
determined you're using the latest code.
Als
_
> drools-interest mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/drools-interest
>
--
Bob McWhirter[EMAIL PROTECTED]
The Werken Company http://werken.com/
---
This sf.net emai
gt; >
> >
> >Hi
> >Can any body please tell me how to control the order of
> >execution of the
> >rules in a rule set
> >thanx
> >mahi
> >
> >
> >
> >---
> >This sf.net email is sponsored by:ThinkGeek
> >Welcome to geek heaven.
> &
Both the types?
>
> Thanx
> mahi
>
>
>
> ---
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> ___
> drools-interest mailing list
> [EMAIL PROTECTED]
> https://lists.source
On Tue, 6 Aug 2002, John Pritchard wrote:
> hi Maheedhar,
>
> there is no control over the ordering of rules' evaluation.
No, drools does follow a rule's priority, which currently is defined
only as the order in which rules are added to a RuleBase.
More flexibility will be added soon.
> the e
Just forwarding this to the old list for folks who haven't migrated to
the new list.
-bob
-- Forwarded message --
To: [EMAIL PROTECTED]
From: bob mcwhirter <[EMAIL PROTECTED]>
Date: Sat, 10 Aug 2002 03:01:15 -0400 (EDT)
Subject: Update
Howdy folks--
97 matches
Mail list logo
