[wg11] Part 28 Teleconferences

[Ed Barkmeyer]

All,

I agree that AP folk generally NEED a mapping language that specifies 
how a given XML schema is related to a reference EXPRESS schema.  I 
believe this is the intent of the German position on the "configuration 
language".  Such a language in effect specifies how to transform XML 
data that corresponds to the given XML schema back to a conforming Part 
21 file, or to an XML file that corresponds to the Part 28 default 
mapping.

The reasonable expectation is that the "given XML schema" should be 
contained in some "standard", e.g. an SC4 AP, an OMG standard, an OAGIS 
standard, an OASIS registered exchange schema, etc.  That standard would 
specify the rules for exchanges under the "given XML schema" and the 
interpretation of any corresponding "exchange document".  Thus, those 
exchanges are NOT specified by Part 28, and Part 28 need NOT be involved 
in the statement of conformance to that standard.  (That standard may 
elect to reference certain clauses of Part 28 normatively, of course, 
but that is out of our scope.)

That standard should contain an Annex that specifies the mapping from 
EXPRESS to that XML schema using the "configuration language" defined in 
Clause 8 of Part 28.  So that Annex "conforms" to Part 28, per some 
class of conformance (e.g. "conformance of a schema mapping") described 
in clause 4.  But that conformance is entirely distinct from 
"conformance of an exchange document".

What this means is that I was completely wrong when I suggested that we 
should integrate the effects of the configuration language directives 
into Clauses 6 and 7!  What Joe Fritz wanted to do -- describe all the 
effects of the configuration directives in clause 8 -- was the right 
choice!  I propose to undo my mistake.

Clause 6 and clause 7 should specify the default mapping and the 
exchanges under the default mapping, full stop.  Clause 4 should define 
"conformance of an exchange document" in terms of the default mapping 
ONLY!  As observed above, a data exchange that conforms to some 
"configured" XML schema conforms to the standard that defines that 
schema and doesn't need Part 28 for "document conformance".

For the record, NIST also agrees that there should be a separate 
conformance class for the "heavier" version of the default XML schema 
that Tom Rando has championed.  That class should include the key and 
keyref additions to the XML schema, and also the two-step derivations 
for constrained STRING and BINARY types and for redeclared attributes. 
Critical to this class is the idea that the heavier schema describes XML 
data exchanges exactly the same as those conforming to the default 
mapping and the EXPRESS schema, but enables more XML-based data validation.

Clause 8 should specify the configuration directives and their effects 
on the mapping from EXPRESS to the "derived XML schema".  Conformance 
(of a mapping, of a schema generator) to clause 8 should be a separate 
type/class of conformance to Part 28.

Note:  A long, long time ago, about January 2003, we agreed that there 
are two kinds of "configuration directives" -- those that alter the XML 
schema and those that only add extra rules to clause 7.  I'm not sure 
that any of the latter survive, but if they do, we need to think about 
how to handle them.  Given the above dichotomy, they wouldn't be 
appropriate in Clause 8 (they don't affect schema mapping), and they 
would have to be some kind of optional rule for clause 7, selected by 
implementors' agreement or something the like.

Also for the record, I see such a change as primarily editorial -- 
massive in scope, perhaps, but editorial in nature.  I envisage it as a 
change in the organization of the text of the standard whose purpose is 
to enable the clarification of the conformance classes.  In that way, we 
may consider the result not to require a second CD ballot.  We are not 
making major technical changes; we are making major editorial changes, 
but only minor technical changes.  The major technical change is in 
altering the notions of conformance, and that is a particularly 
appropriate area for comments on a DIS-FDIS ballot.

-Ed

-- 
Edward J. Barkmeyer                        Email: edbark at nist.gov
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8264                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8264                FAX: +1 301-975-4694

"The opinions expressed above do not reflect consensus of NIST,
  and have not been reviewed by any Government authority."