fork, task instance variable and process variable

[yxyang]

Hi

I have one fork node to fork two task-nodes. For each task node, there is a variable controller to populate the task instance variable from the process variable. So, in database, there are three records, one is for process variable, the other two for task instance variables.

After the first task node's task end(), this end will submit the task instance variable to database and override the process variable. That is to say, after this step, two of the three records are the same.

But the third which is the second forked task-node's task instance variable, has the original value of the process variable.

Is this normal? For me, if the second task node's task-instance variable can be synchronized automatically with the original process variable, it works better.

Is this a problem? Of course, i can load the process variable and overwrite the second task instance variable if i want by programing....

yang

[kukeltje]

Yep, each taskinstance gets a copy of the processvariable at the moment the task is created. The behaviour was different in 3.0

If you do not want tasklocal variables, just don't include them. The taskinstance has access to the process variables as well (at least in cvs-head, but afaik in 3.1 as well). The thing is that the forms do not know this, so if you want to use this in the jbpm forms, you have to do some work yourself. Or build your own webapp.

[yxyang]

i cannot see the value of using task local variable. It incurs lots of db IO.