WSD Issue 272 is Closed

Dear Henry,
On behalf of WSD wg, I would like to thank you to give us detailed
Architectural comments and feedback for [1] in your email [2] which is
recorded as WSD Issue 272 [3]. We also thank you for joining us in the
f2f meeting [4] and clarifying the usage of NOTATION in XML Schema and
illustrating how we may be able to utilize it in our note.

As you probably know, the current approach taken has been the result of
evaluating several months of discussion within the WSD wg, XMLP wg and
consequently in the joint media types task force. We have evaluated
defining the full hierarchy of content type Schema Types, use of URNs,
use of XML Schema Annotations as well as NOTATIONS in several f2f
meetings and as part of the task force discussions. The current design
is the conclusion of this work and reflects the consensus of the wg. 

We think that using global attributes provide a flexible approach. This
approach is considered preferable for the following reasons: 

 -- The contentType and expectedContentTypes values are not fixed in our
note, hence they don't require us to define a fixed hierarchy and/or
require users utilizing an existing type hierarchy and extend it. The wg
does not want to define or maintain type hierarchies that are extensible
themselves and would like to retain the extensibility of the contentType
value domain. Instead, it is felt that referring to the definition of
Accept header values and Content type values would be sufficient as they
are already defined (with some restrictions) to provide the flexibility
and extensibility that is intended. In contrast to defining a type
hierarchy using schema types/notations which then would be required
either for us or for the users, the current approach allows the users to
use familiar contentType strings such as "image/jpg", instead of

-- We are asked to support both xs:base64Binary and xs:hexBinary types.
If we were to use type hierarchies, this will require two separate

 -- The current approach allows us to use accept-params as well as
parameters that we find it hard to define using other means, including

 -- The current approach allows us to use lists and wildcards within one
value and refer to the value easily.

-- Using a single entity for a reference, namely a QName, attribute, or
element is important for our use. The global attributes allow this kind
of use. 

-- It is desirable not to require dependency to XML schema or require
XML Schema references in the instance documents. 

In addition, some members of the wg believe that NOTATION is not widely
used and it is felt that retaining the current flexible and extensible
approach is more desirable.

Therefore, after a long discussion we have decided to close the issue
without changing the current architectural approach. We do hope that
this email clarifies the intent and the basis/motivation for this

We again thank you for participating in the wg meeting in Boston and
talking to us. The current updated draft can be found in [5]. 


--Umit Yalcinalp (On behalf of the WSD wg)





Received on Friday, 18 March 2005 02:31:56 UTC