Re: RSA SAML Interop technical issues

From
Yuzo Koga <>
Date
2005-02-10T08:39:31+00:00
ID
Thread
Re: RSA SAML Interop technical issues
All,

Please find attached metadata files of NTT.
I have changed them so that our implementations (IDP and SP)
would use certificates that are provided during the dry-run event.
Endpoint has not been changed since the dry-run event.

Thanks and best regards,
Yuzo

Ciochon, Robert wrote:
> Hi,
> During the dry run we were able to assemble the metadata for all vendors 
> who are participating.   If you have had to change your metadata since 
> the dry run, please send it out to the list so everyone can maintain a 
> current version for their providers.  This will be one less step that 
> will need to be done at show setup.  Note that some vendors were running 
> without SSL at the dry run.  If you were one of them, SSL will be 
> required for the show and your metadata will need to change.
> 
> A few items came up during the dry run that were accomodated, but in the 
> interest of a smoother show setup everyone wanted standardized.  Please 
> respond if you disagree with the proposed requirements for the SAML Interop:
> 
>     * RelayState in an idP initiated SSO - This varied between vendors,
>       with some passing a valid URL, others sending an empty string,
>       others not sending it at all and still others using a special
>       string.  The concensus of those on the conference call today was
>       to specify that the RelayState is optional, but if sent, it MUST
>       be a valid URL.
>     * XML signature KeyInfo element - Some vendors were failing if an
>       XML sig was sent without having the key embedded in the KeyInfo
>       element.   The concensus on the call was to have it optional if
>       the KeyInfo is sent and not have it required by any vendor.
>     * Signing AuthnRequest - The Metadata standard provides for separate
>       settings for idP and SP on whether an AuthnRequest should be
>       signed, and they can conflict (the SP metadata specifying don't
>       sign it, the idP specifying it must be signed).  The concensus on
>       the call was to leave it up to the SP to specify if the
>       AuthnRequest was signed, and the idP would not have a preference. 
>       However, it appears from the spec this can't be set for the idP
>       (it requires either always or never signed), so instead, the
>       requirement is that all AuthnRequests will be signed.
> 
> Please respond as soon as possible to the above issues, as a decision 
> will be put in writing on Friday.
> Regards,
> Bob
> 
> Robert Ciochon
> eTrust Development Manager
> Computer Associates
> San Diego, California
> (858) 625-6866
> 
> 

-- 
Yuzo Koga <>
NTT Information Sharing Platform Labs.
tel: +81 422 59 3202, fax: +81 422 59 5659, aim: yzkoga