On 6/20/07, Fred Dushin <[EMAIL PROTECTED]> wrote:


I think we need to think long and hard about how WS-SecurityPolicy
fits into the current WS-Security interceptor.

As a start, I'd recommend we remove any WSS4J-isms in the exposed
interfaces (programmatic and configuration), and instead focus on
exposing these through JAX-WS features.


We can then focus on how WS-SecurityPolicy objects clash/complement
features defined statically in config.


Yeah. The current ws-security stuff is very basic and is in desperate need
of an overall as we do the WS-SX stuff.  We need to make sure the
WSSecurityFeature API, SecurityPolicy and WS-SX stuff all align well, which
is no trivial task. Or maybe it is - I honestly am not enough of a SecPol/SX
expert to comment. Maybe we can start a wikipage and start capturing
requirements/ideas.

Before we do any more feature work, though, I think we need to get a
definition of jaxws:features at Bus scope, though I think there may
be a slight semantic disconnect there.  The beauty of defining things
at Bus scope is that it can serve as a fallback/default mechanism for
configuration that is not defined on a specific client or endpoint.



https://issues.apache.org/jira/browse/CXF-742 - I targeted this for 2.0.1


But the way the AbstractFeature stuff is defined, setting "features"
on an InterceptorProvider, when the InterceptorProvider is a Bus, is /
additive/ on the ultimate list of interceptors -- it's not, in
particular, a default fallback mechanism.  (Please correct me if I've
got that wrong)  So I think a little more design work needs to go
into that.



Are you thinking about a case where two features might add the same
interceptors to an InterceptorProvider? Can you explain a little bit more
the case where multiple features are adding the same interceptors to a
Bus/Client/Server?

Thanks,
- Dan

--
Dan Diephouse
Envoi Solutions
http://envoisolutions.com | http://netzooid.com/blog

Reply via email to