RE: now://example.org/car (was lack of consensus on httpRange-14)

Let's take it from the top, comparing the two different scenarios.

1.
Somebody creates an XML namespace and issues a not-on-the-web identifier.
Everybody that uses the namespace gets by without representations from the
identifier.  No confusion about whether to deref or not.

Then at some point later, somebody decides to make representations on the
web.  One example might be to refer to a document with RDF assertions in it.
The author will want all the clients to "break" if they use the name as an
identifier as the author wants the users to do something with the
representation.  The author will want all the software that used the
namespace to change as it has to "know" that the namespace name changed from
being not dereferencable to being dereferencable.  Thus the software will
have to change.  The namespace name has to change from one scheme to
another.  And clients will break, probably automatically.   Automatic
detection of a change in the intent of the identifier.

2.
Somebody creates an XML namespace and issues a web identifier but there's no
representation.  Everybody that uses the namespace gets by without
representations from the identifier.  Some confusion about whether to deref
or not.

Then at some point later, somebody decides to make representations on the
web.  One example might be to refer to a document with RDF assertions in it.
The author will want all the clients to "break" if they use the name as an
identifier as the author wants the users to do something with the
representation.  The author will want all the software that used the
namespace to change as it has to "know" that the namespace name changed from
being not dereferencable to being dereferencable.  Thus the software will
have to change.  The namespace name may or may not have to change.   But how
does the client software "know" to break, especially if the identifier stays
the same?  The Author doesn't have any ability to force a change with out
changing the namespace name.  Seems much more likely that the software
creator would change the namespace name to force a break.  Different
Resource and all that.  And if the author does change the namespace name,
it's exactly like case #1 except it's more confusing.

It seems to me that adding representations on the web means that the author
wants users to do something different with the namespace name, and clearly
distinguishing between non-representation oriented and representation
oriented identifiers means that the author gives clearer intent.  And it
gives software a better chance of distinguishing what to do with
identifiers.

Cheers,
Dave



> -----Original Message-----
> From: www-tag-request@w3.org
> [mailto:www-tag-request@w3.org]On Behalf Of
> Bill de hÓra
> Sent: Wednesday, October 09, 2002 2:00 AM
> To: 'Julian Reschke'; 'Champion, Mike'; www-tag@w3.org
> Subject: RE: now://example.org/car (was lack of consensus on
> httpRange-14)
>
>
>
>
>
>
> > From: www-tag-request@w3.org [mailto:www-tag-request@w3.org]
> > On Behalf Of Julian Reschke
> >
> > So what do you do when you have chosen a
> > not-on-the-web-identifier for your XML namespace and later
> > find out that actually you *do* want representations for it
> > to be on the web?
>
>   GET /car HTTP/1.1
>   Host: example.org
>
> The scheme doesn't matter - to be precise the http: scheme is used to
> locate resources, not to dereference them. What matters is the ability
> to plonk a hostname-port pair in the namespace (at least for
> the web as
> it stands today). You'll want to do that anyway to ensure you have a
> unique name.
>
> Bill de hÓra
> --
> Propylon
> www.propylon.com
>
>
>
>
>

Received on Thursday, 10 October 2002 15:39:02 UTC