|
One of the facets of the conversation that has come
up (again) is the following question:
Should a namespace URI reference an actual resource
(such as a schema definition) or not?
In the end, it seems to me that there is only one
reasonable answer: It can be if it wants to be.
Requiring explicit use or non-use of URIs as URLs
defeats the purpose of flexibility that XML and related technologies are meant
to provide. According to the namespace
specification, the only explicit purpose of the URI is to group elements and
attributes together that conform to a given standard. A standard may
or may not have a defined schema and the schema may or may not be found at the
given namespace URI.
Let's suppose we are processing a document that
uses namespaces. Here are the following possible ways we can view the XML
document:
1) We are expecting a specific type
of document and/or namespace, for which we internally understand how to
process. In this case, it doesn't matter whether the URI points to a
schema definition or not. We aren't looking at that external document
anyhow.
2) We are expecting any document and/or namespace
that requires external validation. In this case, we attempt to access the
external document using the URI. If it cannot be found, then the document
is rejected (this would be true even if the URI was a URL and the host server
wasn't responding). Yes, we used the URI as a URL here, since our
application required it. If a document was sent that did not use a URL as
a URI, then the application wouldn't know what to do with it
anyhow.
3) We are expecting a specific type
of document and/or namespace that requires external validation. In
this case, we first attempt to see if the namespace URI is one that we should
accept and perform further processing on. If so, then we continue as if
folling the second case above. If not, then the document would be
rejected.
4) We are expecting any document and/or namespace
that we don't need to know anything about. As with the first case, it
doesn't matter whether the URI points to a schema definition or not. It
may be accessed if available, but doesn't need to be.
As a result, I would expect an application to
process namespaces something allong the following lines:
SELECT CASE
CASE (namespace is simply a URI that does not
indicate any association to a schema definition)
process document
accordingly
CASE (namespace is recognized
internally)
process
document according to internal schema
CASE (specific
external schema is required or any valid external schema will do)
IF (namespace URI references an accessible and understandable
schema)
process
document according to external schema
ELSE
cannot
validate document. handle accordingly.
ENDIF
ENDSELECT
In the end, how a namespace URI is interpreted
beyond the explicit definition(s) in the namespace standard is up to the
application processing the document. How we should choose to assign
additional meaning to the URI should not be forced upon us by a
specification.
| 