Matthieu,
Looking at how to model the timeout condition in Intalio so that it executes as I expect ... the attached project provides for
1. a sequence (subprocess) that will 'alarm' if the allowed duration of intermediate timer event attached to the border is exceeded.
2. if the alarm is raised then the timeout subprocess will be executed including the explicit fault end event which halts the normal processing within the subprocess (T1-T2)
3. the fault event is trapped in error handler attached to the outer subprocess
4. In all cases the reply message is dispatched.

This works (no doubt as some developer intended) but I do not think that it is a correct implementation of BPMN! Is thee a better way??

Note that the remedy suggested by Antoine
You can create the timer event on the border, connect it to nothing. When the timer is triggered, the scope will exit by its normal output. does not have any effect on flow as the intermediate timer does not break out of the subprocess.

David,

if a BPMN construct is missing in the modeler, you are very welcome to enter a bug here:
http://bugs.eclipse.org
Choose the STP project, the SOA product, and the BPMN component.

Note that the BPMN modeler is only about BPMN. No BPEL transformation bugs will be accepted there.

There is a fine line between modelling and then transforming the process for execution, and have a direct approach where you build an executable process. We have chosen the best compromise between modelling and executability and we will try to improve on that going forward.

Thanks,

Antoine

Although it is of interest how the Intalio product is built up from a number of sources, it is actually the final product that needs to have the compliance with standards and documentation. I am sure that you can communicate the issues across your organisation.
It is of interest to Intalio to get the BPMN compliance fixed, rather less so for me ... I just need to judge whether it is compliant/likely to be compliant/has known issues/etc.