Re: [wsbpel] RE: Questions (RE: [wsbpel] Proposed agenda for May28-29 WS BPEL TC face to face)

[Assaf Arkin <arkin@intalio.com>]

Satish Thatte wrote:

>I agree that humans make the need for concurrency more apparent.  The
>concurrency issue in any case only applies to default behavior since
>explicit invocation of compensation can be as concurrent as one likes.
>For default compensation the only concern about ordering is side effects
>on shared data external to the process since default compensation cannot
>use parameters for obvious reasons.  I need to think more about the
>reliability aspect.
>
>Satish
>
>  
>
Explicit invocation of compensation can indeed invoke compensation in 
parallel for scopes with different names. But it cannot invoke 
compensation in parallel for multiple scopes with the same name. In 
addition, the specification currently precludes the mixing of 
<compensate scope/> and <compensate> in the same compensation or fault 
handler. As a result, you either have to know the order for those 
compensations that depend on it, or resort to reverse order for all scopes.

arkin