Call Activity 'End Error Event' parallel multi-instance

[brochet]

I defined a call activity with a parallel loopType.
My loop collection contains a list with 2 items. My call activity subprocess contains: one "start event" which goes to a "humentask" which goes to a "xor gateway" which goes on true to a normal "end event" and on false to the "end error event".
On my call activity i defined a boundary event to catch the subprocess "end error event" exception.
In my test i have 2 testcases.

1. one item leaves the subprocess over the "end error event" the second item is in the subprocess humentask
-> the catched boundary event path is executed but the second item instance ist still available, is this correct?
    I thought all instances that are still active should be destroyed.


2. both leave the suprocess over the "end error event"
-> the catched boundary event path is executed twice, is this correct?
    I thougth it should only execute one time.

Thanks

[zarezadeh]

I have this issue too. I created an unit test for this problem.

[frederikherema1]

The multi-instance behavior wraps the original behavior in a special activity, which executes the wrapped one multiple times. Looks like it doesn't take the error-event into account, and only destroying the single execution-scope rather that the FULL scope (containing all instances of the multi-instance).

Can you create a JIRA-issue for this, attaching the test case?

[zarezadeh]

I created an issue in jira for this bug. ACT-1295