process resume

[galanfish]

in a parallel structure for example and-split -> and-join scene, more than one child-execution will be created,
if something goes wrong in alternative branch with a exception throw out,
in the exceptionhandler i persist the process instance,
is it possible to resume the process instance to the status before the exception happen, i mean just what it looks like as the moment the exception still haven't happen, i wanted to redo the logic
or what programmic should i do to resolve this problem

[kukeltje]

This is a topic for the user forum..

But imo the solution would be to not persist the process instance in the exceptionhandler. That way it would be the same as before the exception (a rollback)

[galanfish]

assue that i want to build a simple workflow with simple language definition on top of pvm,
and i want the workflow could have this feature:
when a activity execute failed, the process instance could persist,
and users could see the failed process instance and decide whther to resume,
if the user decide to resume, i except that the process could restart from the failed activity to redo it.

is the current pvm support this and developers could just easily to achieve the goals?
to meet this demand, how can i do it?

[kukeltje]

Please read the first line of my previous response...

In case it was not clear ;-) this is the developer forum

[galanfish]

i am sorry, i am new(just a college student)... but i......
i thought at first that some one develops a workflow-like thing with the pvm kernel would be called developer, and some one develops application-like with jbpm would called user...
i want to develop a simple workflow with my own simple language

[kukeltje]

well... yes and no... if you develop the pvm (or jbpm) itself you are a jpbm/pvm developer. If you develop with the pvm, like in your case your own language, or with jbpm you are a user.