This has indeed been discovered recently, for certain scenarios where activations were queued in the wrong order after fact modification.

The fix is in ./drools-core/src/main/java/org/drools/core/util/BinaryHeapQueue.java:

--- a/drools-core/src/main/java/org/drools/core/util/BinaryHeapQueue.java
+++ b/drools-core/src/main/java/org/drools/core/util/BinaryHeapQueue.java
@@ -185,7 +185,7 @@ public class BinaryHeapQueue
                 compareToParent = compare( this.elements[index],
                                            this.elements[index / 2] );
             }
-            if ( index > 1 && compareToParent < 0 ) {
+            if ( index > 1 && compareToParent > 0 ) {
                 percolateUpMaxHeap( index );
             } else {
                 percolateDownMaxHeap( index );


-W

2011/4/27 Tihomir Surdilovic <tsurdilo@redhat.com>

Hi Chris,
since you mention to already have a support license for JBoss Enterprise BRMS, the best place to ask these types of questions is at the excellent JBoss Customer Support Portal (https://access.redhat.com/home) where your question will be handled under SLAs ensuring timely response and continuous quality assurance monitoring of the same.

Just FYI, I believe the issue you are mentioning has been fixed in trunk (see Jira https://issues.jboss.org/browse/JBRULES-2942) which should make the fix available in the next BRMS 5.2 release. Please confirm with the Red Hat support engineers through a support case as they are much more knowledgeable on product releases, and direct any further questions regarding the supported BRMS bits to them.

If there are any further issues that you would like to see prioritized for the supported BRMS bits you are using, we will be glad to work with you through the JBoss Customer Support Portal.

Thanks.
Tihomir

On 4/27/11 2:31 PM, Chris Selwyn wrote:

I am finding that the "salience" feature is acting very erratically.

Some of my rules modify the working memory. So I would like them to execute before the others that simply read the memory after modification and report on certain data conditions that are left after all modifications have happened.

The "modifying" rules have a salience of 5. The "reading" rules have a salience of 0.

Using the rules logging I can see activations of my modifying rules being created and activations of the reading rules being created.
And I can see "reading" rules (with salience 0) being executed before "modifying" rules (with salience 5) even though no other activations are being created in between them.

I am not using agenda groups or anything "fancy" like that.

Debugging through the code I can see the "MAIN" agenda group is a queue organised as heap.
However, the order in which things happen is very non-deterministic (presumably due to hashing or something like that) and I am finding it very difficult to actually pin down an actual 100% reproducible case.

Is there any known problem with the salience mechanism?

I am using JBoss Rules 5.1.0 (with a support licence).

Chris Selwyn

_______________________________________________
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users

_______________________________________________
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users

_______________________________________________
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users

No virus found in this message.
Checked by AVG - www.avg.com
Version: 10.0.1325 / Virus Database: 1500/3601 - Release Date: 04/27/11