Not understanding desire to map directory attributes to XML attri butes

[Kent Loving <KLoving@Paradigm4.com>]

Please excuse me if my tone seems blunt, but I'm just not getting something.
Please show me where I'm wrong.

Looking back at the recent threads "Draft Data Model" and "Samples for DSL
2.0", this group desires a data model which meets these goals:
1) It's XML
2) it represents directory data, accessed via LDAP, RDF or some wrapper over
relational data.
3) It's "easy" to access via XPATH.

In meeting those goals, it is OK to not remain backward compatible with DSML
1.0.

The proposed model (in the "Samples" thread) maps directory attributes to
XML attributes.  This is a significant departure from DSML 1.0.  I think the
reason for that mapping is to meet goal #3.  If that's true, exactly how
does this mapping make it easier to use XPATH?

I'm no expert in XPATH, but I think the following xpath finds my
telephoneNumber in a DSML 1.0 doc exported from my company's LDAP directory:
//entry[attr[@name="givenName"][value="Kent"]]/attr[@name="telephoneNumber"]
/value

If the model has directory attributes mapped to XML attributes, that XPATH
might become:
//entry[@givenName="Kent"]/@telephoneNumber

Not a fantastic improvement.

{I'm sure I have the syntax wrong in one or both of those examples, but I
think I'm close}

Besides not being backward compatible with DSML 1.0, the proposed model has
these shortcomings pointed out in the threads:

1) How will the new model represent entities that have more than one
directory attribute with the same name?
2) How will the new model represent entities that have "multiple
inheritance"?

BTW, I have no investment in DSML 1.0.  I just want to 'help' the adoption
of 2.0.  We are implementing an LDAP 'connector' which gets commands and
returns results in XML.  

Kent Loving
Sr. Software Engineer
kloving@Paradigm4.com