Re: EndpointRefs-47

Henry Thompson asks:

> Why not use the following URI to identify that endpoint:
> 
>   http://example.com/fabrikam/acct?fabrikam:CustomerKey=412

I feel like I'm missing something.  This proposal seems to take the 
namespace prefix, which has almost no semantic significance in the EPR, 
and makes it the qualifier for the query string parameter in the URI.  As 
far as I know, the following two EPRs are semantically identical, yet your 
proposal suggests they would map to two different URIs:

EPR 1:

 <wsa:EndpointReference 
xmlns:fabrikam="http://example.com/FabrikamRefParms">
  <wsa:Address>
    http://example.com/fabrikam/acct
  </wsa:Address>
  . . .
  <wsa:ReferenceParameters>
    <fabrikam:CustomerKey>
      412
    </fabrikam:CustomerKey>
  </wsa:ReferenceParameters>
</wsa:EndpointReference>

and

EPR 2:

<wsa:EndpointReference xmlns:p="http://example.com/FabrikamRefParms">
  <wsa:Address>
    http://example.com/fabrikam/acct
  </wsa:Address>
  . . .
  <wsa:ReferenceParameters>
    <p:CustomerKey>
      412
    </p:CustomerKey>
  </wsa:ReferenceParameters>
</wsa:EndpointReference>

These seem to map to:

        http://example.com/fabrikam/acct?fabrikam:CustomerKey=412
        -and-
        http://example.com/fabrikam/acct?p:CustomerKey=412

I have at least a strong intuition that this ambiguous mapping is 
undesirable.  As we know, minting more than one URI for the same resource 
is to be avoided where possible;  a lot of software won't know that the 
two URIs above refer to the same thing.  I can easily imagine software 
that will fail to preserve prefixes when echoing an endpoint reference. 

I don't know whether the WSA folks are correct that their users need 
URI-qualified parameter names (which they implement as QNames), but I have 
to assume they know their customers.  Given that they do, I think you need 
to map the expanded name into the URI, not just the {prefix,localname} 
pair.

In fairness, WSA does claim not to offer rules for end point reference 
comparison [1].  Maybe that undercuts my argument, but I'm not convinced. 
Having the two EPRs above map to different URIs seems to me undesirable; 
it will tend to make the prefix names semantically significant in both the 
XML and URI forms of an EPR.

Am I missing something?  Thanks.

Noah

[1] http://www.w3.org/TR/ws-addr-core/#eprcomp

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------








ht@inf.ed.ac.uk (Henry S. Thompson)
Sent by: www-tag-request@w3.org
10/11/05 02:38 PM
 
        To:     "David Orchard" <dorchard@bea.com>
        cc:     <www-tag@w3.org>, (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        Re: EndpointRefs-47



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

So I see a better source for an example, in Hugo Haas's slides about
WS-A [1]:

 <wsa:EndpointReference . . .>
  <wsa:Address>
    http://example.com/fabrikam/acct
  </wsa:Address>
  . . .
  <wsa:ReferenceParameters>
    <fabrikam:CustomerKey>
      412
    </fabrikam:CustomerKey>
  </wsa:ReferenceParameters>
</wsa:EndpointReference>

Why not use the following URI to identify that endpoint:

  http://example.com/fabrikam/acct?fabrikam:CustomerKey=412

?  That's an endpoint I'd _like_ to have a URI for, if my customer key
was 412. . .

ht

[1] http://www.w3.org/2005/Talks/0511-hh-www2005/slide19-0.html
- -- 
 Henry S. Thompson, HCRC Language Technology Group, University of 
Edinburgh
                     Half-time member of W3C Team
    2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
            Fax: (44) 131 650-4587, e-mail: ht@inf.ed.ac.uk
                   URL: http://www.ltg.ed.ac.uk/~ht/
[mail really from me _always_ has this .sig -- mail without it is forged 
spam]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFDTAa4kjnJixAXWBoRAr1qAJ0cyqm8Y2LWdwoDAi4Pagj7u+lAIQCdGLHb
w/lIhGoboiyHd5Ix2jqImck=
=YEXI
-----END PGP SIGNATURE-----

Received on Wednesday, 19 October 2005 19:52:58 UTC