Re: [XSCD] Last Call WD Comments

On Tue, 26 Apr 2005 15:00:38 -0700, <daniel.gabi@swissonline.ch> wrote:
> Hi,
> i read the Document  
> ( http://www.w3.org/TR/2005/WD-xmlschema-ref-20050329/ ) with big  
> interest and great confusion. here, some of my questions and comments:

Thank you for your comments, and sorry to have confused you.

> general:
> - XPath is a non-normative reference. this means, the steps defined in  
> the WD are not an "extension to XPath (1.0)"? but will be implemented  
> and resolved by XPath-processors, because you are using the same syntax?  
> XMLNS ist not mentioned in the references-list.

The important thing to understand, and this may be what is confusing
you: these paths do not navigate through an XML structure, but navigate
through the component model of the schema (which may, and likely was,
ultimately constructed from one or more XML sources).  (See the tableaux
in section 3 of http://www.w3.org/TR/xmlschema-1/structures.html, condensed
in the table of appendix A of the component designators draft.)
I wouldn't expect an XPath processor to implement this, as the information
model it is working with is not at all the same. This is why XPath is
not referenced normatively.  I would expect this to be less about
leveraging code and more about leveraging some conventions and
terminology, and the intuitions that go with them.

> - as i understood it: not (directly) accessible are
>     namespaces other than targetNS

In the component model, every component has a target namespace.

Suppose, for example, we have:
Schema document 1 (target NS uriA) that defines a top-level
element declaration A and imports namespace uriB from schema document
2. Schema document 2 has a top-level element declaration B in the
uriB namespace. If prefix a=uriA and prefix b=uriB, then /b:B is
a perfectly  acceptable reference to the element B in the combined
schema and /a:A is a perfectly acceptable reference to the element A
in that combined schema.

>     the baseURI of the schema-as-a-whole schema component (uh?)

There is no such property on that component, so, no, there is no
component reference for it.

>     particles (4.2.12)

Correct. Particles exist in a 1-1 relationship with terms and
we chose not to lengthen the paths to allow the particle and
term to be identified separately.

>     attribute use (4.2.14)

Also correct.  The case here is similar to particles: for every
attribute use there is either a global or local attribute declaration
that is manifest in the path instead.

>     abstract element declarations

If you mean element declarations with the attribute abstract=true
then, no, that is not correct: these can be referenced just as well
as any other element declaration.  Abstract or not, they still have
a component in the component model for the schema.

> - 4.1 / first list (or 2nd list):
> wildcards are / wildcard is not mentioned (see 4.2.13)

Wildcards only appear in the component model as terms and
are therefore accessible only via the {term} property.
That is why you don't see {wildcard} listed in the lists
in section 4.1 -- there is no such property in the component
model. On the other hand, since wildcards are accessible
components, section 4.2.13 tells about them.

Speaking only for myself,

I hope this helps.

//Mary

Dr. Mary Holstege
Lead Engineer
Mark Logic Corporation
2000 Alameda de las Pulgas
Suite 100
San Mateo, California  94403
+1 650 655 2336 Phone
mary.holstege@marklogic.com
www.marklogic.com

Received on Tuesday, 26 April 2005 23:30:28 UTC