[OGSA-AUTHZ] checkpointing the discussion on VO attributes

[Krzysztof Benedyczak]

Hello,

Valerio Venturi wrote:
> Hi,
> I'll try to checkpoint the discussion had so far.
> 
> As Krzysztof is planning to serve more than one VO with the same
> service, we cannot have a one to one relationship between entityIDs and
> VOs, this imply the need of having a VO attribute. Which was also more
> or less David's concern, an authority being able to assert whatever it
> wants. If we go wiht this, the VO attribute stays.
> We have two proposal so far. Tom suggested to use the MACE-Dir
> eduPersonScopedAffiliation attribute
> 
> <saml:Attribute
>   xmlns:xacmlprof="urn:oasis:names:tc:SAML:2.0:profiles:attribute:XACML"
>   xmlns:ldapprof="urn:oasis:names:tc:SAML:2.0:profiles:attribute:LDAP"
>   xacmlprof:DataType="http://www.w3.org/2001/XMLSchema#string"
>   ldapprof:Encoding="LDAP"
>   NameFormat="urn:oasis:names:tc:SAML:2.0:attrname-format:uri"
>   Name="urn:oid:1.3.6.1.4.1.5923.1.1.1.9"
>   FriendlyName="eduPersonScopedAffiliation">
>   <saml:AttributeValue        
>     xsi:type="xs:string">member at voName
>   </saml:AttributeValue>
> </saml:Attribute>
> 
> while in our first draft Krzysztof and I suggested the use of a specific
> 
> <saml:Attribute 
>   xmlns:xacmlprof="urn:oasis:names:tc:SAML:2.0:profiles:attribute:XACML"
>   NameFormat="urn:oasis:names:tc:SAML:2.0:attrname-format:uri"
>   Name="uri_to_define"
>   FriendlyName="vo"
>   xacmlprof:DataType="http://www.w3.org/2001/XMLSchema#string">
>   <saml:AttributeValue xsi:type="xsd:string">
>     voName 
>   </saml:AttributeValue>
> </saml:Attribute>
> 
> Let's try to agree on one.
I'd vote for the 2nd version, because of the arguments put by Valerio 
previously: eduPersonScopedAffiliation was generally designed to serve 
education purposes and it can be misleading here.

> There were concerns about Tom's proposal to use Grouper to express
> groups, specifically about the contents being an URN. Anyway, the
> specification doesn't mandate them to be URN, it recommends to use URIs
> is uniqueness is to eb achieved.

Please excuse me if I'll be totally wrong here. By any mean I'm not 
Grouper (or Signet) expert.
 From what I recall, in Grouper groups are expressed as 
[grp1]:[subgrp2]:..., and stems as it was proposed: stem1:stem2:...
Anyway Grouper doesn't publish this information directly by means of 
SAML but indirectly, e.g. through LDAP using ldappc and then via Shib IdP.

If I'm right here then the ':' instead of '/' as delimiter gives as 
little advantage and we can stick to quite popular and for me more 
intuitive VOMS syntax.
If I'm wrong then probably we should change to ':'.

In any case we must clearly define syntax of a group name (e.g. 
currently our service does allow for ':' in it) and comparison rules (as 
case sensitiveness).

Regards,
Krzysztof