Re: [cgmo-webcgm] Re: namespace convention

[Mary McRae <marypmcrae@gmail.com>]

Hi Lofton,

  I'm telling you what's in the most recent version of the now-called
AIR (Artifact Identification Requirements). This document will be
submitted for member review within a week or two (really, I promise)
and you can state your case at that point in time for something other
than what I've quoted. The issue is that we need to have something
that works for all TCs. Some TCs *want* their namespaces to be
resolvable; others don't particularly care. As you state below, the
svg namespace does resolve to something, whether it's the actual
schema or not. We need to make sure that the location you use *could*
resolve to something, and to also make sure that it won't resolve to
something completely unexpected. http://docs.oasis-open.org/webcgm
will most likely resolve to something - a page that will list all of
the files/versions of webcgm.

The domain name to be used is either docs.oasis-open.org OR
cgmopen.org (since it's owned by OASIS).

Spec-id is the official short-name of the specification itself: webcgm

schema is required

schema-filename is the actual name of the schema file (whether it
resolves or not)

So, valid namespaces would be:
docs.oasis-open.org/webcgm/schema/XCFschema
cgmopen.org/webcgm/schema/mySchema
docs.oasis-open.org/webcgm/schema/WhateverNameYouLike

Regards,

Mary



On 6/10/05, Lofton Henderson <lofton@rockynet.com> wrote:
> I'm still a little confused.  SInce I'm trying to edit the spec now, can
> you clarify about the URL for the namespace declaration?
> 
> At 04:00 PM 6/10/2005 -0400, Mary McRae wrote:
> >[...]
> >On 6/10/05, Cruikshank, David W <david.w.cruikshank@boeing.com> wrote:
> > >[...]>
> > > 1)  url reference for the namespace declaration in our XML DTD - This
> > one is just a url defining the domain for the namespace.  It doesn't
> > necessarily even have to resolve, but we might want to put a generic page
> > in place to explain that it is the namespace domain for WebCGM.  I don't
> > think this one should be version dependent. It's just declaring the
> > domain space.
> > >
> >URLs or URNs (not both).
> >
> >URLs. It will be left to the TC's discretion if there will actually be
> >something at the other end of a URL namespace.
> >
> >The URL namespace will take the form of:
> >
> >{oasis domain}/{spec-id}/schema/{schema filename}
> 
>  From this I gather something like:
> 
> http://www.oasis-open.org/webcgm/schema/{schema-filename} or
> http://www.oasis-open.org/webcgm2005/schema/{schema-filename} or
> http://www.oasis-open.org/webcgmXCF/schema/{schema-filename} or ...
> 
> I put in the webcgm2005 to illustrate a possibility and a point.  The 2005
> is not a version, but is basically the year the namespace URL went into
> use.  It is a convention from my years in W3C.  It is not desired that it
> be updated for versions, and in fact it is detrimental (to implementors and
> interoperability) if it is updated for versions.
> 
> We can live without it (the "2005"), IMO (but I'm not claiming to speak for
> everyone.)
> 
> QUESTION.  What is implied by "schema filename"?  Are you saying, if the
> address resolves to anything, then must resolve to a schema?  We were
> thinking of doing something like SVG and some other W3C standards.  Its
> namespace URL is:
> 
> http://www.w3.org/2000/svg
> 
> and that resolves to a little HTML document, which contains the W3C logo
> and the text:
> 
> >This is an XML namespace defined in the Scalable Vector Graphics (SVG) 1.0
> >Specification. The latest version can be found http://www.w3.org/TR/SVG.
> >
> >For more information about SVG, please refer to the W3C SVG overview. For
> >more information about XML, please refer to The Extensible Markup Language
> >(XML) 1.0 specification. For more information about XML namespaces, please
> >refer to the Namespaces in XML specification.
> 
> While this little file mentions 1.0, in fact SVG 1.1 and 1.2 all use the
> same namespace URL, and each of them has a section saying that.
> 
> As I said, we were thinking -- if the namespace URL resolves at all -- that
> it should resolve to something like this, and not a schema or dtd.
> 
> So ... how to wrap this up so that we can have an operational namespace URL
> asap?
> 
> Thanks,
> -Lofton
> 
> 
>