xml-dev - RE: simple question on namespaces.

[Jonathan Borden <jborden@mediaone.net>]

Uche Ogbuji wrote:

> > >
> > > Who says "www.whatever.com/foo.bar" is not a URI reference?  Certainly
> not
> > > RFC 2396.
> > >
> >
> > Deep sigh.
> >
> > You really *are* trying to confuse everyone right? :-)
>
> _I'm_ trying to confuse everyone?  Well, it would be a neat effect if I
> could pull it off but no, I was genuinely puzzled.  I'll admit that I
> missed the exact context, or even who was quoted, but I assumed that the
> comment came in the context of Paul's query about W3C's using
>
> "www.w3.org/..."
>
> in their official namespaces rather than
>
> "http://www.w3.org/..."
>
> I thought someone was saying that this would be wrong because it was not a
> valid URI ref.  I didn't really make much connection to the talk about
> defaulting in user agents.

	That was me saying that 'www.w3.org/...' is was not a valid URI reference,
but I was strictly wrong. I was trying to make the point below:

>>
>> Though it is common practice to expand the string
>>'www.whatever.com' into
>> http://www.whatever.com , per RFC 2396 the (relative) URI reference

>> 'www.whatever.com' is not equivalent to the (absolute) URI reference
>> 'http://www.whatever.com'

> Of course.  Never said it was.

	Correct, and my apologies if I appeared too harsh (note the smiley -- it
was there for a reason).

>> <URI.esoterica>
[snip]
>> </URI.esoterica>
>>
>> Is that what you expect?

> I don't think it's really all that esoteric.

	Perhaps not for you but reading these threads there appears to be
widespread confusion regarding URIs. For example:

	//www.w3.org/foo *is* a relative URI which is shorthand for
	bttp://www.w3.org/foo

	/foo might be shorthand for:
	http://www.w3.org/foo (assuming the base URI originates at
http://www.w3.org)

	and

	'foo' alone (as a relative URI) cannot be interpreted without knowing a
base URI but it *never* means

	http://foo


> I have quite a few
> directories of my hard drive of the form "www.fourthought.com",
> "www.4suite.org", etc.  Several of the other software packages I
> install have similar directories.  There is no reason for one to think
> that such a name couldn't be part of a path.

	Exactly, but I would say that 99% of people who see www.w3.org assume that
this means the authority and not a directory on your hard drive (as I
assumed when writing my original numbered statements).

> But I think we're all on the same page now.  Sorry for the confusion.  In
> my defence, the sentence to which I responded *was* wrong.  No two ways
> about it.

	True. And let me end by making this point again: URIs are a foundation but
this specification doesn't alone specify everything we may wish to do on the
web. XML Namespace is also a specification layered on URIs but alone does
not add semantics to URIs, nor tell us how namespace URIs ought be
dereferenced. What we (desperately) need is a specification providing
guidance for how namespace URIs ought be dereferenced when deferencing is
appropriate. Only once such a specification is universally adopted will this
thread end.

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org