Well, yes, if you ignore SAML 2.0. Shibboleth 2 is a SAML 2 implementation
that happens to support older, legacy behavior on top of SAML 1.1.
> The only difference I notice is for the IdP Discovery. Is that the only
> addition I need to support as a conforming 2.0 SP? Is it optional as the
> shibboleth conforming spec is still the old one without mentioning this.
That conformance note is for the legacy extensions and profiles to SAML 1.1,
which are historical. The full set of behaviors that the software follows
are spread across all the specifications listed, assuming it's up to date.
What you support of SAML 2 probably depends on the options you feel are
needed in the context in which you're writing your own SP for some reason.
-- Scott
If your interest is in interoperating with IdPs in the InCommon
Federation, note that InCommon does not yet have a profile for SAML 2.
We'll have to do one, I suppose, as part of adding support for SAML 2 to
the federation, but it isn't there yet. It is my impression that most of
the other national higher ed federations are in more or less the same
situation.
As Scott implied, to interop with Shib 1.3 IdPs you continue to follow
those SAML 1.1 profiles regardless of whether you're using Shib 1.3 or
Shib 2.x.
- RL "Bob"
Winson Quock wrote on 2009-02-24:
> are identical links as those found in
> https://spaces.internet2.edu/display/SHIB/TechnicalSpecs for 1.3.
Well, yes, if you ignore SAML 2.0. Shibboleth 2 is a SAML 2 implementation
that happens to support older, legacy behavior on top of SAML 1.1.
> The only difference I notice is for the IdP Discovery. Is that the only
> addition I need to support as a conforming 2.0 SP? Is it optional as the
> shibboleth conforming spec is still the old one without mentioning this.
Migrating deployments takes time.
> It seems that SAML 2 is already optional in
> 1.3. Our SP implementation had it done anyway. I thought that it would
> become required in Shibboleth 2.
There's nothing to "require". Shibboleth 2 supports the final standard,
which people should use where possible, and supports backward compatibility
to help them get there.
1.3 never had any semblance of SAML 2 support. If you already supported it
in your original implementation, then I imagine you're done.
> But as I'm testing with testshibb now, if I remove my SAML1.1 profiles,
> testshibb-two cannot proceed and reports error (see statck trace below:)
So
> that implies one does not have to support SAML 2 to be shibb 2.0
conformant
> at all.
There is no such thing as "Shib conformant" anymore, and the behavior you're
talking about is because you're invoking the IdP with a legacy Shibboleth
request. Of course it's looking for SAML 1.1 support. If you send it a SAML
2 request, it will look for SAML support, obviously.
> Also from your reply, I take that IdP Discovery is optional as well
(besides
> it does not apply to our SP which already includes the WAYF page and need
no
> discoery service.)
It's optional, and like much of the specs, supporting it depends on what you
want to accomplish. My SP includes discovery as well, but not everybody
wants to do discovery the same way.
-- Scott