Re: What if:

[Charles Reitzel <creitzel@mediaone.net>]

True enough.  But returning something has no bearing on defining an XML namespace.  You could just as easily argue that the lack of an associated resource and/or entity (depending on whose terminology you like) makes mailto: a good NS URI scheme.   I'm not selling the idea.  Nor am I rejecting it.   No, I don't use mailto: myself.

But one size does not fit all.  As I say, give the archives a scan.  I think you'll see thorough treatment of these issues.  More importantly, you'll see the diversity of interpretations and viewpoints.  There's no point in being rigid.  XML Names is what it is.  

Based on how NS aware XML software will actually use the data, the NS URI is, in effect, just an attribute value.  If you like, you can just put in "ooga booga" and it will work just as well.  If you don't believe me, ask your favorite XML tools.

take it easy,
Charles Reitzel

P.S. If you'd like to continue this discussion, let's take it off-list.

At 03:02 PM 4/26/01 -0500, you wrote:
>Charles Reitzel <creitzel@mediaone.net> wrote:
>
>> You are absolutely right.  But, of course, http: refers to web pages (of all
>> things). 
>
>Actually, the HTTP spec makes it relatively clear that HTTP URIs are
>designed to represent _resources_ -- not any specific kind of resource (such
>as a web page). That is why we can get music, images, 3D models, RDDL files,
>HTML, etc. all over HTTP. It also says that HTTP does not deliver the
>resource itself, but merely a related network entity.
>
>So HTTP is perfectly usable for an abstract entity like a namespace. It will
>simply return a representation -- a RDDL document, or schema, or something
>-- of the namespace.
>
>
>-- 
>[ Aaron Swartz | me@aaronsw.com | http://www.aaronsw.com ]


take it easy,
Charles Reitzel