URI for XML schema and namespace wasRe: urn was Re: Model draft

["tom.petch" <[e-mail]>]

I said I would raise the question of what to use to name a namespace on the Apps
Discuss list and I did, under the Subject above.  The discussion was most
interesting with perhaps the most significant view being that of Chris Newman,
as follows
============================================
Date: Fri, 04 Jan 2008 15:46:02 -0800
From: Chris Newman <[e-mail]>

As Applications Area Director, I'm not aware of anyone ever asking for that.

I'm holding a discuss position on:
   draft-narten-iana-considerations-rfc2434bis-08
because the present text forbids use of IANA URLs and I consider that an
unacceptable new restriction for IANA considerations.

Current IETF practice discourages use of IANA URLs in IETF specifications.
However, we have one case, RFC 4790, where stable iana.org http URLs are
provided to registry elements and that was done with IANA's permission.  In that
case, the registry was created to be useful to both the IETF and the W3C. The
stable http URLs make the W3C happier so it was worth doing that way.

I would support a similar approach for iana.org http XML namespaces if someone
spent the time to write up the rules and get IANA's consent.

Using an ietf.org http URIs for XML namespaces is a bad idea.  The IETF has
deliberately kept the registry function for our standards separate and I
consider that a feature.  Also, ietf.org is operated by the Secretariat
function so using that domain for registrations would require additional (and
more expensive) coordination than iana.org.  It _might_ be feasible to set up a
redirect from ietf.org to iana.org for XML namespaces, but I worry that's just
one additional level of complexity where things could break (especially if we
transition the secretariat function between vendors periodically as we're in the
process of doing now).

To summarize the options:
1. URNs for XML namespaces -- present IETF common practice
2. non-IETF http URLs for XML namespaces -- acceptable
3. iana.org http URLs for XML namespaces -- nobody has tried this, seems
feasible to me.
4. ietf.org http URLs for XML namespaces -- nobody has tried this.  I think it's
a bad idea.

                - Chris

David Harrington wrote on 1/4/08 11:47 -0500:

> Has anybody fashioned an explicit proposal for the establishment of
> permanent IETF URIs?
>

============================================

Points to emerge were that whatever is used should be persistent and stable.  I
also note this exchange between Scott Hollenbeck and Julian Reschke.

> -----Original Message-----
> From: Julian Reschke [mailto:[e-mail]]
> Sent: Friday, January 04, 2008 10:54 AM
> To: Scott Hollenbeck
> Cc: 'tom.petch'; 'Apps Discuss'
> Subject: Re: URI for XML schema and namespace
>
> Scott Hollenbeck wrote:
> >
> > Those folks looking at 3470 might have missed the third paragraph in
> > section 4.9:
> >
> > "In the case of namespaces in IETF standards-track documents, it would
> > be useful if there were some permanent part of the IETF's own web
> > space that could be used for this purpose.  In lieu of such, other
> > permanent URIs can be used, e.g., URNs in the IETF URN namespace"
> > ...
>
> Unfortunately, the IETF so far hasn't followed the BCP's
> advice to actually make http-based namespace URIs available.

As one of the authors of 3470 I can say very clearly that 3470 doesn't
provide advice on this point.  It states fact:

"Typically (and recommended practice in W3C) is to assign namespace names using
persistent http URIs."

Notice that there's no "may", "should", "must", etc.

> Thus, if it's considered A Good Thing to provide a
> description of the namespace *at* the namespace URI - and
> when would it not? - I'd suggest to still use an http based
> namespace URL.
>
> There are some IETF specs that for that very reason use URI
> assigned by the W3C (Atom, Atompub), and lots of other IETF
> specs use URIs on purl.org.

That's certainly a valid option.  I'm not saying that one is better than the
other. What I am saying is that people shouldn't think that 3470 recommends one
over the other.

-Scott-
==============================================

So RFC3470 does not make a recommendation:-)  And, unless and until there is a
proposal of the kind that Chris Newman alludes to, I think that the urn: scheme
is the best choice.

Tom Petch

----- Original Message -----
From: "tom.petch" <[e-mail]>
Sent: Friday, January 04, 2008 2:20 PM
Subject: urn was Re: Model draft


> ----- Original Message -----
> From: "Joel M. Halpern" <[e-mail]>
> To: "tom.petch" <[e-mail]>
> Sent: Wednesday, January 02, 2008 3:09 PM
> Subject: Re: Model draft
>
>
> > There appear to be two issues you are raising with the name space
> > declaration in this document.  One issue is whether the version goes
> > before or after the model scoping.  Because the schema and its semantics
> > relate to the protocol as well as the model, it seemed to go in the
> > order we have it (version, lfmmodel.)  However, after reading your
> > comment and thinking about it, I suspect that you are correct about the
> > ordering, and it should be lfbmodel then 1.0.
> >
> > The second issue appears to be our use of an http uri rather than a
> > field delimited urn as the namespace.  The use of HTTP URIs seems fully
> > conformant.  And the use of web space URIs seems recommended by RFC
> > 3470.  The use of URNs for namespaces is offered there as a fallback
> > alternative.  So it is not clear to me what the problem is with our usage.
> >
>
> Not a big problem, perhaps not even little one.  I agree that HTTP URIs are
> recommended by RFC3470 but there is a reference there to ongoing work in an
I-D
> that later became RFC3688, so for me, RFC3688 trumps RFC2470 - and there, as I
> read it, urn are recommended.  Certainly I see a lot of urn and cannot recall
> when I last saw a HTTP URI, apart from here.  I will ask on apps-discuss and
> come back to you.
>
> Tom Petch
>
> > Yours,
> > Joel M. Halpern
> >
> > tom.petch wrote:
> > > I think that the URIs are not quite right yet.
> > >
> > > As per RFC3688, the targetNamespace should be a urn, eg
> > >
> > >  targetNamespace="urn:ietf:params:xml:ns:forces:lfbmodel:1.0"
> > >
> > > and the schema needs a name if it is to be registered with IANA, eg
> > >
> > > "urn:ietf:params:xml:schema:forces:xxx:1.0"
> > >
>