xml-dev - Re: Begging the Question (the novel)

[Uche Ogbuji <uche.ogbuji@fourthought.com>]

> Uche Ogbuji wrote:
> > I've read XML Schemas only once.  I hope not to have to repeat the
> > experience any time soon.  But I don't see that it is relevant.  I don't
> > use XML Schemas, but I do use namespaces.
>
> lisa says:
>
> It's relevant because the XML Schema structures specification provides
> the syntactical "glue" to connect to a schema to a document for
> validation (what would probably be getting done if the URI was
> dereferenced, as is  perfectly legal, blah, blah blah).

In the general case, or only when XSchema is used?  IOW, does the spec
claim to establish this connection for RELAX, Schematron, etc.?

> If there were a single standardized way of referencing a schema from an
> XML document, and dereferencing a schema from a namespace URI violated
> it in some way, it would be a good reason to not do it.  But since it
> (the namespace URI value) is actually one of the "normal" ways to
> associate an XML document with its schema, it seems to suggest that this
> is a good idea, and that it "works" and might be useful for other
> applications too.

Hmm.  If this were "normal" in the strict sense, I'd have expected Rick
Jelliffe, who is on the Schemas WG, to have given it stronger endorsement.

> Or certainly, it wouldn't hurt anything :-)

We have indeed wandered into unfamiliar territory for me, but my guess is
that if the REC says so, and a document is given with an XSD doctype, let
it be as the Schemas REC says, including whatever namespace resolving
machinery they desire.

But at that point we're at that hypothetical layer above core namespaces
that I thought no one wanted to venture into until the discussion of namespaces
themselves had been hashed out.  RDF operates at that layer and so does
SOAP, and apparently XSchemas.  The problem is that for XML technologies
in general, it's Wild Wild West.

> And if you already knew everything I brought up in that last e-mail
> about XML Namespaces, then do you agree with me that these dereferencing
> issues are non-issues, because it doesn't look like we would be breaking
> anything in the process.

No, I don't.

> Because I'm thoroughly confused about how anyone who thoroughly
> understood XML Namespaces could argue about this whole "Tool X" taking
> over the world  with the evil defacto schema implementation bull.

Let me take you from hyperbole to the real world chamber of horrors.

MS-XSL
Intel AT RAM extensions
MS Kerberos extensions (not to mention PPP and DHCP)
Wonderwall
HTML 3.x
MS Java
SQL + Objects
favicon.ico (whimsical but not unconsidered entry)

If you think no one will try to embrace/extend/coopt XML, I'd suggest
you review the history of the computer industry.  My point is that a
sizable opening for this lies at the core.

Is this Megiddo?  Come off it.  I'm partaking in discussion, not
raising the home guard.


-- 
Uche Ogbuji                               Principal Consultant
uche.ogbuji@fourthought.com               +1 303 583 9900 x 101
Fourthought, Inc.                         http://Fourthought.com
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python