RE: again, canonicalization, and adoption

Speaking only for myself below (but, as co-chair
of the XML Core WG, I do have some background
on the issues).

> -----Original Message-----
> From: public-xml-id-request@w3.org 
> [mailto:public-xml-id-request@w3.org] On Behalf Of Webb Roberts
> Sent: Thursday, 2005 August 25 12:59
> To: Norman.Walsh@Sun.COM; public-xml-id@w3.org
> Cc: webb.roberts@gtri.gatech.edu
> Subject: again, canonicalization, and adoption
> 
> 
> We've begun the attempt to integrate xml:id into the next revision
> of the GJXDM (Global Justice XML Data Model).  There is considerable
> concern from the field:
> 
> 1. Lack of support for xml:id.  Since xml is not yet approved,
> systems may fail to support it, or may take some time to support it,
> delaying adoption of and conformance to GJXDM.

There are various tools that do support it already:
http://www.w3.org/XML/2005/01/xml-id-implementation.html

Understandably, though, older tools do not.

If you don't have a DTD or schema, you can't have
IDs working right now.

If you do have a DTD or schema, declare xml:id as 
an ID therein, and it will work in existing tools.

I cannot see how using a non-xml namespace could
alter this concern in any way.

> 
> 2. Conflict between xml:id and Canonicalization.  The perception
> exists that Canonicalization (not exclusive canonicalization) is
> very important, and that xml:id breaks with respect to
> canonicalization, and so that xml:id is not an option.  We have to
> prove that this is not the case to move forward.

xml:id does not break.  (Non-exclusive) Canonicalization 
does.  And it's already broken with respect to xml:base.

But insofar as an existing C14N implementation will not
work properly with xml:base and xml:id, there doesn't
seem to be any easy way to address this other than
not using xml:base and xml:id or not using the C14N
implementations that have this problem.  

I understand this is not an optimal situation, and
I sympathize with your concern here. Our previous research
indicated to us (perhaps inaccurately) that exclusive
C14N--which doesn't have the problem with xml:base and
xml:id--was the much more widely used version.

Using a non-xml namespace for your IDs (and not using
xml:base) would avoid this issue.

> 
> 3. Instability.  The fact that xml:id is not yet a Recommendation
> means that adoption at this point is risky.

The PR period for xml:id ends tomorrow, and we have seen
little feedback to date that would lead us to expect 
anything other than editorial changes to the spec.  So
depending on lags just due to administrative processes,
I would expect xml:id to be a Recommendation within a
month or two.

In short, I see little risk that xml:id will not become
a Recommendation, and even if it didn't, I don't see how
using a non-xml namespace could address this concern.

paul

> 
> We're at a decision point, and a go or no-go decision must be made. 
> How does the xml:id community recommend that we address these
> specific concerns?  A no-go decision means the GJXDM NDR will use a 
> non-xml namespace for its ID attributes.
> 
> Thanks,
> Webb Roberts (currently drafting the GJXDM NDR)
> 
> (a comment of concern from our group)
> 
> jjmierwa@visionair.com (Jospeh Mierwa) wrote:
> > ...2 specific issues with the use of XML ID at this juncture.
> > 
> > First (and less importantly) is the tools concern. There is still
> > a decided lag in tools being able to support the standard. Use 
> > of ID at this point will very likely impact development schedules
> > because of it and will slow the adoption to it.
> > 
> > Secondly, and more importantly is the fact that canonical xml is 
> > an integrel part to web services security which is already very 
> > heavily used in the enterprise. My understanding is that usage of
> > xml:id with canonical xml still has compatibility issues, one of
> > which is with duplicate IDs. I also have not seen anything from
> > the W3C that would indicate the issues have been resolved. Moving
> > the GJXDM to make use of xml:id would be premature without fully
> > assessing the impacts on the transport technologies involved and
> > without identifying adequate means of mitigating/working around
> > those problems.
> > 
> > I understand that the GJXDM is not meant to be technology 
> > specific, but we must have some regard for the prevalent 
> > contemporary technologies in play, otherwise, we may see the rate
> > of adoption of GJXDM (and consequently NIEM) slowed.
> 
> 
> (previous discussion on the xml-id list)
> 
> webb.roberts@gtri.gatech.edu wrote:
> >>>> I think it would be very helpful if there was some 
> >>>> structured documentation that would help us get xml:id 
> >>>> through c14n objections. Right now, there is resistance.
> 
> Norman.Walsh@Sun.COM wrote:
> >>> The Core WG plans to address the C14N issues, but we don't 
> >>> feel that such plans need to be in the xml:id specification.
> 
> webb.roberts@gtri.gatech.edu wrote:
> >> Agreed.  But sooner is definitely better, since we have lead 
> >> times for vetting upcoming releases.
> 
> Norman.Walsh@Sun.COM wrote:
> > Indeed.
> 
> 
> -- 
> Webb Roberts (webb.roberts@gtri.gatech.edu)
> Research Scientist, Georgia Tech Research Institute
> Atlanta, GA  404-385-0181
> 

Received on Thursday, 25 August 2005 18:48:05 UTC