Re: [SpecGL-impl] xml:id Last Call

At 04:49 PM 1/4/2005 -0500, Karl Dubost wrote:
>Lofton,
>
>I have read again the xml:id specification and again I find everything,
>but maybe I have my own way of reading things.

Clearly, we disagree about it.


>I would like to know if in xml:id specification you want something
>like, which will be an abstract of the prose of the specification?

I want something as simple as a clear list of CoP, each of which is well 
defined, and something about conformance.  I don't want it to consist of 
some vague references that are obscurely buried in prose, in such a way 
that it has to be read creatively to figure out what are the CoP.  The 
latter is the case with xml:id -- whatever you can glean from the Intro is 
mostly the sort of imprecise generic blither that one finds in almost every 
specification.

I don't want xml:id to be considered as a good example of how to satisfy 
the CoP requirement.  I would like to see something approximating our 
techniques for this requirement, to quote:
"Give the classes of products in the specification:
1. Think about all the types of products or services that will implement 
this technology, group those that are similar and/or
basically achieve the same purpose, and determine the generic name for the 
group.  This would be the class of product
2. List these classes of products in the specification
3. Describe them as part of the scope"



>Suggestion:
>[[[
>* Classes of Products
>         here is the list of classes of products which should implement this
>specification.
>         * XML Processors
>         * DTD and XML Schemas (non required but encouraged)
>         * Author of XML documents (non required but encouraged)
>]]]

s/which should implement this specification/for which this specification 
defines conformance requirements/.

With that change of wording, your suggested text would be a vast 
improvement over what's there now.

>[...]
>Reminder: the techniques are one way to achieve things but not the only
>way.

IMO, the xml:id Intro is not a suitable way.


>Techniques of 2.2.A - SpecGL , 22 Nov 2004
>[[[
>         •       Give the classes of products in the specification:
>         1.      Think about all the types of products or services that will
>implement this technology, group those that are similar and/or
>basically achieve the same purpose, and determine the generic name for
>the group.  This would be the class of product.
>         2.      List these classes of products in the specification.
>         3.      Describe them as part of the scope.
>]]] -

Considering where this Requirement came from, i.e., how concisely it was 
specified as a Checkpoint for CoP, I think these Techniques are to be 
strongly preferred.

In what follows, you present careful analysis of the text and argumentation 
about how it actually satisfies the Requirement.  In my view, it should not 
be that obscure, and should not require so much work to find a precise CoP 
list!  It should be as simple as what you wrote above (okay, recognizing 
that no one uses the terminology "classes of product" yet).

So basically ... we disagree.  You think it should be construed as 
satisfying the CoP requirement.  I don't.

-Lofton.

>http://www.w3.org/TR/2004/WD-xml-id-20041109/#intro
>
>[[[A mechanism allowing unique element identifiers to be recognized by
>all conformant XML processors, whether they validate or not, is
>desirable in making XML sub-resource linking robust.]]]
>
>1st class of product: XML Processors
>
>[[[An additional problem is that DTD-based and XML Schema-based
>identifiers are exposed through different conceptual mechanisms - the
>[attribute type] infoset property, and the [type definition] family of
>properties respectively. A uniform mechanism for recognizing
>identifiers is desirable.
>
>This specification provides such a mechanism: it describes the
>semantics of xml:id attributes.]]]
>
>2nd class of product: DTD and XML Schemas
>
>[[[Implementors are encouraged to support xml:id processing and to make
>ID assignment the default behavior of their processors.]]]
>
>Who: Implementers of processors.
>
>
>http://www.w3.org/TR/2004/WD-xml-id-20041109/#syntax
>
>[[[Authors of XML documents are encouraged to name their ID attributes
>"xml:id" to increase the interoperability of these identifiers on the
>Web.]]]
>
>Who: Author of XML documents. Though it's not mandatory just a
>suggestion.
>
>
>
>
>--
>Karl Dubost - http://www.w3.org/People/karl/
>W3C Conformance Manager
>*** Be Strict To Be Cool ***
>
>