Making RELAX NG schemas normative rather than non-normative

[Rex Jaeschke]

When this was raised in Tokyo, I made a comment that I'll repeat here.

While the JTC 1 rules permit electronic attachments to standards, the
contents of such attachments must also be included in the "printed" version.
As such, we have electronic schemas, which are reproduced in Annexes. For
the normative XSD schemas that means we have two versions, one electronic
and one printed, which can lead to the two being out-of-sync. (And I believe
this was the case for 29500:2008.) In such cases, which version is the
"correct" one as far as conformance is concerned? That is unspecified, even
in the consolidated reprint 29500:2011.

Now if we were to add a second normative set of schemas, we'd have four
versions, two electronic and two printed, which doubles the chances of one
or more of them being out-of-sync with the others.

BTW, does anyone know of any other XML-related standard that has multiple
equivalent schemas with all being normative?

That aside, more importantly, I'd like to know just what real-world problem
does the making RELAX NG schemas normative actually solve?

Regards,

Rex



> -----Original Message-----
> From: eb2mmrt at gmail.com [mailto:eb2mmrt at gmail.com] On Behalf Of
> MURATA Makoto
> Sent: Saturday, March 12, 2011 9:37 PM
> To: SC34
> Subject: Making RELAX NG schemas normative rather than non-normative
> 
> Dear colleagues,
> 
> We are going to consider this issue in Prague.  (If nothing horrible about
> nuclear power supplies happen, I can attend.)
> 
> We discussed about this issue in Tokyo (only briefly) and decided to
discuss
> again in Prague.
> 
> Making the OOXML RELAX NG schemas normative does not mean that XSD
> schemas become non-normative.  They continue to be normative.
> 
> In my understanding, there are two implications in making RNG schemas
> normative, and there are no other implications.
> 
> First, for OOXML documents to be conformant, validity against normative
> RNG schemas will become a must.  But, as far as I know, everything in
> OOXML RELAX NG schemas is stated either in normative prose or XSD
> schemas.  More about this, see DR 10-0030.
> I believe that  RELAX NG schemas were extensively tested before the BRM,
> but I do not know whether they have been used for validating OOXML
> documents since then.
> 
> Second, if the normative RELAX NG schemas are in conflict with the XSD
> schemas or normative prose, OOXML will be defective and we will be
> required to fix this defect.  I believe that we have already assumed that
such
> differences have to be addressed.
> 
> Cheers,
> Makoto <EB2M-MRT at asahi-net.or.jp>