On Fri, Dec 9, 2011 at 2:43 PM, Sam Hartman
<[email protected]> wrote:
> I think it is fairly likely that the IDP and RP will have the software
> to do normal SAML things, but in some of the deployments we're looking
> at will not have the provisioning (keys, metadata etc) to do SAML over
> HTTP.

And I wouldn't want to encourage reliance on the HTTPS PKI.

However, what could be done is that an attribute with a URI could also
have a digest of the thing to be fetched via HTTP, and maybe a digest
of the server cert or an intermediate CA for it (or perhaps a key that
the actual attribute payload will be encrypted in, that way we can use
plain HTTP).  But this starts sounding hairy.

> Also, I actually think there will be intermediates that will want to
> rewrite attributes.

I imagine so.  I can see several reasons: 1) to rewrite attributes
understood by one side into attributes understood by the other, 2) to
apply privacy policies.  (2) might be common in a deployment with a
common, trusted trust broker, so to speak.

Nico
--
_______________________________________________
abfab mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/abfab

Reply via email to