Re: (Fwd) XML Schema and public identifiers

[dcpleland@ftnetwork.com]

Hi Lauren;

I thank Michael for his response, and express my regret that the Schema WG did not agree with our direction.  I think that the end result will be a missed opportunity for added functionality and flexibility in the implementation of XML Catalogs.

I note that it seems that the central point of Michael's position seems to be:

>the provision of
>a keyword to allow lookup of a schema document (using a
>system identifier) using a namespace name would provide
>(it seems to me, at least) the same practical advantages
>that catalog lookup of public identifiers has always done
>for the use of DTDs. 

I've not tested it yet, but it appears that this position discounts the utility of having a centrally maintained 'universal' catalog, and leaves us all to maintain our own system versions.  With the likely result of different behavior and different results in each implementation, unless each system is very carefully maintained with reference (anyway!) to a centrally maintained 'universal' catalog before processing doc instances/data files.

If that's the case, then this issue will raise on the priority ladder, and will be revisited later, after it has been flagged as a problem.

Regards,
David Leland
************************************************************
lauren@softquad.com wrote on 4/13/01 10:02:39 PM
************************************************************
The formal comments.

Lauren

------- Forwarded message follows -------
Date sent:      	Fri, 13 Apr 2001 13:38:59 -0600
To:             	Lauren Wood <lauren@sqwest.bc.ca>
From:           	"C. M. Sperberg-McQueen" <cmsmcq@acm.org>
Subject:        	XML Schema and public identifiers
Copies to:      	Paul Grosso <pgrosso@arbortext.com>,
       	Norman Walsh <Norman.Walsh@East.Sun.COM>,
       	W3C XML Schema WG <w3c-xml-schema-wg@w3.org>,
       	W3C XML Coordination Group <w3c-xml-cg@w3.org>

Lauren:

I write you in your capacity as head of the Oasis
Entity Resolution TC.

In a recent phone call, you and I discussed the request
from the Oasis Entity Resolution technical committee for
a change to the W3C XML Schema specification, to make
the XML Schema spec include support for (formal) public
identifiers.  I believe that Paul Grosso, who brought the
request to the XML Coordination Group, has already
reported back to you on the eventual outcome, but since
we agreed on the phone that it might be useful to have a
formal response, I am writing you now in my dual capacity
as co-chair of the XML CG and as co-chair of the XML Schema
WG.

The XML Coordination Group discussed the request from
Oasis at its meeting of 30 January, and agreed after
discussion to forward the request to the XML Schema WG
for consideration.  (I note that the CG was divided over
whether to request that the WG consider it for XML
Schema 1.0 or to request only that they log it as a
request for enhancement and take it up in the context
of work on XML Schema 1.1.  The comment period for the
XML Schema candidate recommendation ended in December.)

The XML Schema WG considered the question at its meetings
of 1 and 8 February, and discussed it at some length.
I won't attempt to summarize the discussion beyond saying
that many WG members felt that the functionality provided
by public identifiers can be provided without any change
to XML Schema, although a number of WG members (notably
including those who are also members of the Oasis
Entity Resolution TC) continued to argue that this was
not the case and that in addition to the namespace name
and zero or more schema-location hints for each namespace,
the XML Schema spec should provide a slot to allow users
to specify a formal public identifier for the schema
document(s) as well.  In the end, the WG decided against
making the suggested change.

In addition to the request formally made through the XML
CG, Norm Walsh proposed that the XML Schema WG provide
formal public identifiers for the XML Schema DTD and
for the schema for schemas.  The WG did not take any
action on this proposal.

Speaking both for the WG and as an individual, I would
urge the Oasis Entity Resolution TC in the strongest possible
terms to recognize and exploit the capabilities offered by
catalogs for the interoperable identification of schema
documents (and DTDs, for that matter) -- the provision of
a keyword to allow lookup of a schema document (using a
system identifier) using a namespace name would provide
(it seems to me, at least) the same practical advantages
that catalog lookup of public identifiers has always done
for the use of DTDs.  (My memories of SGML usage before
the wide adoption of SGML Open catalogs have left me with a
far lower estimate of the intrinsic utility of public
identifiers than some of our colleagues appear to have.
The utility lies quite clearly, I think, in the idea of
an interoperable catalog.  Without it, formal public
identifiers did me no particular good at all.)

I think a keyword specifically for schema lookup, using
the namespace name as the lookup key, would be useful,
though I am not sure it's essential.

An explicit reminder to users and implementors that system
identifiers can also be redirected using catalogs
would be equally useful.  And a statement that, for purposes
of schema lookup, any schema location hints should be
treated as system identifiers would be helpful. (There is
a view that says they should be treated as schema or
namespace identifiers, but I think personally that that
is not a good idea.)

I hope this response is useful to you and helps complete
the record.

best regards, and best wishes for the successful completion
of the TC's work,

Michael Sperberg-McQueen
------- End of forwarded message -------

------------------------------------------------------------------
To unsubscribe from this elist send a message with the single word
"unsubscribe" in the body to: entity-resolution-request@lists.oasis-open.org


___________________________________________________________________________________________________
Get your free e-mail account with *unlimited* storage at  http://www.ftnetwork.com

Visit the web site of the Financial Times at  http://www.ft.com