Hi there, some people complained that the Golfer-example from the JBoss-Drools distribution finds two identical solutions. I investigated a little and found the "problem": The rule is underspecified: It binds five Golfer objects, but only four of those are completely specified by their respective condition elements. The fifth, the "Freds right neighbour", lacks that completeness... What happens is the following: The "unknown golfer" can be bound to two different facts: I had the "unknown" golfer printed out - and see what happened: Fred 1 orange, Joe 2 blue, Bob 4 plaid, Tom 3 red unknown Tom 2 blue Fred 1 orange, Joe 2 blue, Bob 4 plaid, Tom 3 red unknown Joe 2 blue As the unknown golfer is NOT compared to the actual neighbor, Joe, there are two possible instantiations for our "unknown" golfer... The solution would be to constraint the unknown-golfer (= Fred's right neighbour) to the set of (Bob, Joe, Tom)... an eval sounds great - too bad the "in" constraint doesn't work... // freds right neighbour is either Tom, Bob or Joe! eval (($fn == $joe) || ($fn == $bob) || ($fn == $tom)) The simple (but un-elegant) solution is setting an agenda-group rule- attribute, prohibing the rule from firing twice. In his book on JESS, E. Friedemann-Hill shows JESS to deliver only a single solution ... I translated the Drools-version literally to his JESS version (I had to create GolferColor and GolferPosition classes, but needed only 32 facts in working memory, instead of 64 with the combined Golfer-drools version).... regards, Gernot Dr. Gernot Starke Doing IT Right --- Willi-Lauf Allee 43, D-50858 Köln gs(a)gernotstarke.de, +49 (0) 177 - 728 2570 http://www.gernotstarke.de Blog: http://it-and-more.blogspot.com **************************************** Das freie Portal für Software-Architekten: http://www.arc42.de _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

the language statement "this in ($joe, $bob, $tom)" sounds & looks great to me - more intuitive than the eval. imho itshortens down the overall LHS... Considering rule-performance, one should imho fail-as-early-as-possible... but for the example that surely does not matter... thanx for this improvement... will you commit this to the next release, so this sample works "as expected" from now on? regards from (sunny) Cologne, Gernot Am 11.08.2007 um 20:45 schrieb Edson Tirelli: Gernot, Thanks for taking the time to investigate and report your findings. I implemented a fix in a similar way to your suggestion, but instead of writing an eval() in the end, I reordered the patterns leaving the Fred's neighbor pattern as the last pattern: // The golfer to Fred's immediate right // is wearing blue pants $fn : Golfer( position == ( $fredsPosition + 1 ), color == "blue", this in ( $joe, $bob, $tom ) ) What do you think about this solution? It now correctly solves the problem, but I'm not sure it is as elegant solution as you was looking for. If you want to take a look, just make sure you use drools-core jar from trunk (not GA), since I found and fixed a minor bug while changing the golfer example. Edson 2007/8/10, Dr. Gernot Starke <gs(a)gernotstarke.de&gt;: > > Hi there, > > some people complained that the Golfer-example from the JBoss-Drools > distribution finds two identical solutions. > I investigated a little and found the "problem": > > The rule is underspecified: It binds five Golfer objects, but only > four of those are completely specified by their respective > condition elements. The fifth, the "Freds right neighbour", lacks > that completeness... > > What happens is the following: The "unknown golfer" can be bound to > two different facts: I had the "unknown" golfer > printed out - and see what happened: > > Fred 1 orange, Joe 2 blue, Bob 4 plaid, Tom 3 red > unknown Tom 2 blue > > Fred 1 orange, Joe 2 blue, Bob 4 plaid, Tom 3 red > unknown Joe 2 blue > > As the unknown golfer is NOT compared to the actual neighbor, Joe, > there are two possible instantiations > for our "unknown" golfer... > > The solution would be to constraint the unknown-golfer (= Fred's > right neighbour) to the set of (Bob, Joe, Tom)... > an eval sounds great - too bad the "in" constraint doesn't work... > > // freds right neighbour is either Tom, Bob or Joe! > eval (($fn == $joe) || ($fn == $bob) || ($fn == $tom)) > > > The simple (but un-elegant) solution is setting an agenda-group rule- > attribute, prohibing the rule from firing twice. > > In his book on JESS, E. Friedemann-Hill shows JESS to deliver only a > single solution ... I translated the Drools-version > literally to his JESS version (I had to create GolferColor and > GolferPosition classes, but needed only 32 facts in working memory, > instead of 64 with the combined Golfer-drools version).... > > regards, > Gernot > > > Dr. Gernot Starke > Doing IT Right > > --- > Willi-Lauf Allee 43, D-50858 Köln > gs(a)gernotstarke.de, > +49 (0) 177 - 728 2570 > http://www.gernotstarke.de > Blog: http://it-and-more.blogspot.com > **************************************** > Das freie Portal für Software-Architekten: > http://www.arc42.de > > > > _______________________________________________ > rules-users mailing list > rules-users(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/rules-users > -- Edson Tirelli Software Engineer - JBoss Rules Core Developer Office: +55 11 3529-6000 Mobile: +55 11 9287-5646 JBoss, a division of Red Hat @ www.jboss.com _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users Dr. Gernot Starke Doing IT Right --- Willi-Lauf Allee 43, D-50858 Köln gs(a)gernotstarke.de, +49 (0) 177 - 728 2570 http://www.gernotstarke.de Blog: http://it-and-more.blogspot.com **************************************** Das freie Portal für Software-Architekten: http://www.arc42.de _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users