[JBoss jBPM] - Problem with timers
by petia
Thanks Roland, here I am starting a new topic.
I have a problem with setting a timer on a Task Node.
I have defined the following process
| <?xml version="1.0" encoding="UTF-8"?>
| <process-definition xmlns="urn:jbpm.org:jpdl-3.2" name="timer-test">
| <swimlane name="Initiator">
| <assignment actor-id="admin"></assignment>
| </swimlane>
|
| <start-state name="start-state1">
| <task name="start task" swimlane="Initiator"></task>
| <transition to="TaskA"></transition>
| </start-state>
|
| <task-node name="TaskA">
| <task name="TaskA">
| <assignment actor-id="petia"></assignment>
| </task>
| <timer duedate="10 seconds" transition="skipB"></timer>
| <transition to="TaskB" name="execute B"></transition>
| <transition to="end-state1" name="skipB"></transition>
| </task-node>
|
| <task-node name="TaskB">
| <task name="TaskB">
| <assignment actor-id="petia"></assignment>
| </task>
| <transition to="end-state1"></transition>
| </task-node>
|
| <end-state name="end-state1"></end-state>
| </process-definition>
The idea is that if the timer in Task Node TaskA expires the process will be ended (and the Task Node TaskB skipped).
However, when running the process and the timer expires, the transition skipB is not executed (as specified in the code) and the process stalls at Task Node TaskA (which appears in the tasks list, but without an "End" button available for it).
I am running jBPM 3.2.3. From the discussion on http://www.jboss.com/index.html?module=bb&op=viewtopic&t=140634&postdays=... I can see that there have been some problems with timers on State Nodes which apparently have been fixed in 3.2.4. Does this apply to my problem with timers on task nodes? Have anyone else experienced a similar problem?
Note, the timer is important. Example of a business scenario: if a notification is not registered within a week, the process is ended otherwise it is continued. Therefore, the solution proposed by Alejandro (thanks Alejandro) suggesting that the timer is replaced by asynchronous continuation (see below), can not be applied in this case.
anonymous wrote : JBSEAM-1575 depends on JBPM-1135 which is already resolved. That said, I have two suggestions.
|
| The first is to replace the timer with an asynchronous continuation, unless you need the delay (which I understand you do not). Just set the node's async attribute to true. It does not need to be a separate emtpy state, you can decorate any of your existing nodes with that attribute.
|
| The second is to implement a ThreadLocal based solution which prevents recursive calls once the loop node is reached. I came to it for preventing the stack overflow in the BPEL while activity, but it can be implemented for jPDL as well. You can browse the code in the online repository. Look at Loop.execute(). It is much of a hack - you might want to stick to asynchronous continuations unless you need more performance.
| _________________
| Saludos/Regards,
| Alejandro
In summary: Does the Timer on Task Node work in jBPM 3.2.3? Does it work in jBPM 3.2.4? (Clarification: the timer seems indeed to expire, but the specified transition is not taken, which results in a stalling process instance)
Kind regards, Petia
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4171721#4171721
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4171721
18 years, 1 month
[JBoss jBPM] - Re: workaround for StackOverflow exception
by alex.guizar＠jboss.com
JBSEAM-1575 depends on JBPM-1135 which is already resolved. That said, I have two suggestions.
The first is to replace the timer with an asynchronous continuation, unless you need the delay (which I understand you do not). Just set the node's async attribute to true. It does not need to be a separate emtpy state, you can decorate any of your existing nodes with that attribute.
The second is to implement a ThreadLocal based solution which prevents recursive calls once the loop node is reached. I came to it for preventing the stack overflow in the BPEL while activity, but it can be implemented for jPDL as well. You can browse the code in the online repository. Look at Loop.execute(). It is much of a hack - you might want to stick to asynchronous continuations unless you need more performance.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4171652#4171652
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4171652
18 years, 1 month
[JBoss jBPM] - Inheriting from ProcessDefinition
by pjacobsen
To add some additional fields to my process definitions, I've subclassed the jBPM ProcessDefinition object with my own.
<subclass name="com.xxx.XXXProcessDefinition"
| extends="org.jbpm.graph.def.ProcessDefinition"
| discriminator-value="S">
| <property name="definitionName" type="string" column="SDS_NAME" />
| ....
However, I'm unsure of what would be the best way to create a my new process definition subclass. The archive parsers and process definiton itself all seem to use the static function on the ProcessDefintion: ProcessDefinition.createNewProcessDefinition(); which doesn't allow me to create my subclass while reusing the code to wire up the default module and such.
Has anyone done this in a manner that wouldn't be likely to break on new releases of jBPM (i.e. I don't think it's a good idea to make my own function to create a definition or attempt to clone my subclass from the ProcessDefinition) as it would seem somewhat brittle should modules go away... or something else be added.
Thanks!
Paul
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4171604#4171604
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4171604
18 years, 1 month