Shibboleth Developers

["RL 'Bob' Morgan" <>]

On Thu, 12 Jun 2003, Scott Cantor wrote:

> > So we should just step up and realize we have a registry of
> > attr type URNs.  That registry can put them in:
> >
> >   urn:mace:dir:attribute-type:<well-known-short-name-of-attr-type>
>
> I really, really don't want to call this attribute-type. It's not a
> type. It's an attribute. The type is "string", or "integer", or "ASN.1
> blob in format x". Calling it type will really confuse things because
> SAML doesn't treat it as a type, and in fact we have to define a type
> when we send the attributes that isn't the same concept at all.
>
> Even LDAP has types, doesn't it? I thought the object class had
> attributes in it, and those attributes have a type, and a short name,
> and an OID.

This is just a terminology clash, then.  "Attribute type" is precisely the
phrase used in X.500 and LDAP for this entity.  X.520 is "Information
technology   Open Systems Interconnection   The Directory: Selected
attribute types".  It says, eg, "The Common Name attribute type specifies
an identifier of an object."  The bare word "attribute" refers to an
actual instance of the thing that holds the data in an actual directory.

I suppose my intent with "urn:mace:dir:attribute-type:" was to have a URN
that could refer to both the LDAP-context-defined and
SAML/XML-context-defined object with the same semantic.  I guess
"attribute-type" is too overloaded.  Just "attribute" is already
overloaded of course, since in XML it means something else.  I could live
with "attribute-def", but of course that means changing things ...

 - RL "Bob"

------------------------------------------------------mace-shib-design-+
For list utilities, archives, subscribe, unsubscribe, etc. please visit the
ListProc web interface at 

    http://archives.internet2.edu/

------------------------------------------------------mace-shib-design--