xml-dev - RE: Namespaces, XML in Browsers

[Andrew Layman <andrewl@microsoft.com>]

I've just read through your slides on XML Namespaces.  As you indicate in
several slides, the specification has been subject to a fair amount of
discussion.  In this context, I believe that extremely careful use of
terminology is important.  In several places, your rewording of aspects of
the specification alter somewhat the meaning of the actual specification or
its status.  

1.	A slide reads "URIs come with user understandings. Because URIs
include URLs, and most users associate URLs with retrievable resources,
there are constant questions regarding the relationship between the entity
body named by a URI and the namespace. (According to the spec, there isn't
any definite relation.)"  It would be more accurate to avoid saying that a
specific entity body is named by a namespace URI.  That is, avoid the
definite article and all it implies. The URI identifies a set of names.
What the specification stops short of saying is whether any concrete
resource MAY or MAY NOT be retrievable from that URI. This is not the same
thing.

2.	I do not think that discussions this summer regarding relative URIs
as namespace identifiers was primarily about their _meaning_, as one slide
states.  There was discussion regarding whether namespace names really were
URIs or really were strings, and there was also discussion regarding whether
relative (or more exactly, contextually-sensitive) URIs should be banned.

3.	It is not accurate to say that Namespaces in XML "specifies
character-by-character string comparison for namespace processing."  More
exactly, it defines the term "identical" to describe those namespace URIs
that are character-by-character the same.  I specifies certain validity
constraints related to identical names. It does not define the equivalence
or non-equivalence of URIs whose characters differ.  That would be governed
by the URI specification, IETF RFC2396.

4.	It is not certain that the status of relative URIs is a settled
issue.  (And, more exactly, the issue is not specific to relative URIs but
to URIs which are in any way contextually dependent, e.g.
"localhost://xyz".)

In addition, the following points were wrong or unclear in some text though
the correct interpretation was implied in other text.

5.	The primary motivation for scoped namespace declarations was modular
document construction by streaming processors.

6.	From the point of view of the XML Infoset, a namespace declaration
is not an attribute but a namespace declaration.  (See
http://www.w3.org/TR/xml-infoset#infoitem.namespace-decl)

7.	Namespace declarations are effective, not only inside the element in
which they are declared, but beginning with the GI of the element itself.

I hope that these observations are helpful.

Best wishes,
Andrew Layman

-----Original Message-----
From: Simon St.Laurent [mailto:simonstl@simonstl.com]
Sent: Thursday, November 16, 2000 6:53 PM
To: XML-Dev Mailing list
Subject: Namespaces, XML in Browsers


I just posted two of my XMLDevCon 2000 presentations to my site.  As per my
usual, there was more text on the slides than in my presentation, so
they're reasonably readable.

Namespaces in XML: Best Practices, Risky Business
http://www.simonstl.com/articles/namespaces/index.html

Cross-Browser XML
http://www.simonstl.com/articles/xbrowse/index.html

The presentations were only an hour long, so these are a bit shorter than
usual, but I hope folks find them useful.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
XHTML: Migrating Toward XML
http://www.simonstl.com - XML essays and books