Hi Mike,

I suspect you've had plenty of traffic on this already, but in writing up your
InsuranceClaim BPEL process in Chapter 10 (p.270) of 'Essential Business Process
Modeling,' you show a 'kill' operation tied to an <onMessage> activity of an
<eventHandlers> section. I'm sure you know the place, as you describe it as not
working, likely due to a bug (p. 284), and provide a specific workaround (p.277).

Did you ever resolve that? I've been digging away at it, and have discovered that,
indeed, it doesn't appear to work at process scope. But if you add a scope
immediately below the process level, and if you add the kill operation <onMessage> to
the <eventHandlers> for this scope instead of at the process level, it works just
fine, both correlated and uncorrelated.

Your workaround is a good example of what we all have to do to get things working
sometimes. But <eventHandlers> is too big and useful a topic to let slip this way.
This one in particular could potentially resolve a lot of the questions about
instantiation that we don't know how to ask.

Author's response:

I think Doug's comment is worth including under Errata. Considerations:

1. I developed this example on an early version of Oracle BPEL Process
Manager. If there was a bug in that version related to message handling
at process scope, perhaps it was caught and fixed in a later release. I
have not tracked this. Readers are invited to try this example (without
the workaround) on the latest release.
2. Readers are invited to try this example on other BPEL platforms too,
such as those from BEA, IBM, and Apache.
3. Doug's workaround of using a scope below process level is,
admittedly, less onerous than my workaround, but it's still a
workaround. If I were to revisit this example today, I would try it
without workarounds on the latest offerings of Oracle, BEA, IBM, Apache,
and so on.

Regards,
Mike Havey