Changes to web flow action proposal

[Cantor, Scott]

On 4/19/13 2:43 PM, "Brent Putman" <putmanb at georgetown.edu> wrote:

>Aside from the subtleties of the handling choices:  Don't forget about
>java.lang Error, the other
>type of unchecked, undeclared thing that can be thrown.  So perhaps some
>or all of the catching
>should be for java.lang.Throwable, which handles both RuntimeException
>and Error cases.

K, will review.

>I didn't get a chance to say on the call, but whatever it winds up
>looking like, I agreed with
>Scott's comment that the same doPre/do/doPost pattern ought to be used in
>MessageHandlers as well,
>at least for ones that have any kind of state, etc.  I guess for both
>handlers and actions, ones
>that are simple and stateless don't necessarily need that pattern.  But
>also wouldn't hurt to just
>make them all consistent.

The non-stateful ones would just implement doExecute as usual and it
shouldn't matter much.

Was looking at the auto-wiring stuff a little. There's no way to do
something non-invasive to every bean definition because the autowrire
attribute defaults off. Also seems to be a little invasive in that it will
try to autowire all your properties, maybe it's more constraining to just
do "constructor" and limit it to constructor argument injection.

At a minimum, I think I like switching the OpenSAML action interface to
assume injection of the context by property, and even if that isn't done
up top, the IdP code can do the injection for any non-Spring aware object
below. That can be promoted to having Spring to do it later if that turns
out to be viable.

-- Scott