Experiments with command-line and web flow adaptor

[Cantor, Scott]

On 4/8/13 3:49 PM, "Tom Zeller" <tzeller at dragonacea.biz> wrote:

>Yes, I think so. But I also think we should re-define the word
>"profile" in a non-SAML specific way, e.g. Marvin wondered to me
>"should I be operating on profiles?" What would that definition be ?

It's been used all along in a non-specific way, but with a definition that
obviously adheres to SAML, one of which is, "a unit of behavior observed
externally as a standard interaction".

In SAML, there's always a profile, even if the protocol has no behavior
that has to be nailed down to make it something more than the protocol
started life as. The original Shibboleth protocol is thus a profile
inherently, as is CAS.

The other place this manifests is that it's the definition used when you
want to plug something into SAML metadata. What you express in an endpoint
element is a profile that the endpoint supports.

>The experience I had with the Action boilerplate code was : look for
>@Nonnull, check code to make sure it was implemented, decide which to
>fix ;-)

Yes. I'm in favor of making that explicit in code using method names and
types if we can define appropriate behavior when null checks are violated
that leaves us in control of error handling.

-- Scott