Experiment: Revising Part 3

[MURATA Makoto (FAMILY Given)]

Dear colleagues,

In Tokyo, I pointed out major problems in Part 3 and proposed to
revise it.  My slides are now available as WG4 N0159.  The attached
document is derived from WG N0153 (29500-3 with COR1 and AMD1
Integrated) by trying to address these major problems.  It is hoped
that this document helps us to have a better idea about the proposed
changes and estimate the amount of work required for this revision.
But please note that this document 

Among the six major problems, three have been addressed in the 
attached document, while the remaining three require more work.

1) Namespace subsumption

As agreed in WG4 e-mail discussions in the thread "Proposed fix for DR
10-0016 (MCE: Core Concepts)", namespace subsumption is underspecified
and cannot achieve interoperability.  As far as I know, there are no
implementations.

The attached document does not have namespace subsumption.

2) PreserveElements and PreserveAttributes

Since the DIS ballot, Japan believes that PreserveElements and
PreserveAttributes are underspecified and cannot achieve
interoperability.  They are explicitly disallowed in Parts 1 and 4.
As far as I know, there are no implementations of PreserveElements and
PreserveAttributes.

The attached document does not have PreserveElements and
PreserveAttributes.

3) Application-defined extension elements

Part 1 introduces ext and extLst elements.  Part 3 call such elements
"application-defined extension elements" and provides some normative
text in Clause 12.  The attached document does not have this clause.

I believe that it should be dropped from Part 3 for two reasons.
First, this clause is not useful, since the definition of ext and
extLst elements in Part 1 does not normatively reference to it.
Second, application-defined extension elements cannot be handled by a
markup preprocessor in the processing model of Part 3, since they are
intended to preserve information.  (I would argue that
application-defined extension elements in Part 3 destroy software
layering.)

4) Lack of  "MCE-aware validity"

In many places in Parts 1 and 4, we require validity against schemas
while allowing the use of MCE elements and attributes.  In other
words, what we need is not the validity of the source document but the
validity of the target document created from the source document by 
the markup preprocessor

The attached document defines "MCE-aware validity", but introductory
text and examples are still missing.

5) Processing model

The processing model should be rewritten as a normative reference model, as 
pointed out by DR 09-0315.

The attached document changes the title of the clause from
"Preprocessing Model for Markup Consumption" to "Reference Model for
Markup Consumption" and added an introductory paragraph.  However, the
rest of this section requires a lot of rewriting.

6) MCE as a thin layer between an application program and a document model

This problem should be automatically addressed if the other problems
are addressed.  However, we might want to have informative text about
the possibility of implementing MCE as a thin layer between an
application program and a document model.  The attached document does
not provide any text about such a thin layer.


Cheers,
Makoto
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Office Open XML Part 3 - Markup Compatibility and	Extensibility.docx
Type: application/octet-stream
Size: 138047 bytes
Desc: not available
URL: <http://mailman.vse.cz/pipermail/sc34wg4/attachments/20101013/7a0e875d/attachment-0001.obj>