Re: No Standard way to reference XML Schema? Was Re: (Many) XML Schema Questions

I wonder if there is any support for making a new spec for what is, in
effect, a better DOCTYPE declaration?   I think it is not quite within
the
ambit of XML Schemas, Packaging or Catalogs, so it is liable to fall
through
the cracks.  Even if it follows stylesheets with some useable mechanism
using PIs to allow immediate implementability, that would be OK.

 I think there is room for a spec to allow declarations that:
    * document type X is a certain collection of schemas with a list of
allowed roots.
    * document type X has a public identifier (a URN)
    * document type X is closed w.r.t to those schemas: no other
elements are allowed
    * a document of type X must be schema-processed before it can be
used, because the
schema contributes to the information set of the document.
    * the schemas can be located along a certain chain
    * some versions of the schemas may have different annotations
suitable for different purposes: documentation in different languages
(as distinct from datatypes in different languages?) for example
    * the document has been validated, and to give some signature to
authenticate this.

The simplest way is to have a PI on the instance:

<?xml-doctype system="...." public="...  ...   ..."  standalone="no"
key="....."?>

which references an explicit doctype language document:

<doctype urn="urn:mydoctype" root="x"  validation="closed"
certificate="...">
    <alt>
        <schema href="..."  title="schema"   />
        <schema  href="..." title="developers' version"   />
        <schema type="application/xml-dtd" href="..." title="schema
using DTD syntax" />
    </alt>
    <schema namespace="http://www.w3.org/1999/XHTML"
        href="..." title="strict html elements"  />
</doctype>

I think that there will always be a divide of opinion between those
who think that a document should be regarded as standalone by
default (i.e. what the WG implements??) and those who think that
a document with a schema should require the availability of the
schema by default.

Rick Jelliffe

Received on Thursday, 30 December 1999 10:41:28 UTC