W3C home > Mailing lists > Public > www-xml-canonicalization-comments@w3.org > March 2000

XML Schema WG response to the C14N Last Call WD

From: Biron,Paul V <Paul.V.Biron@kp.org>
Date: Fri, 24 Mar 2000 09:38:02 -0800
Message-Id: <376E771642C1D2118DC300805FEAAF4386DACC@pars-exch-1.ca.kp.org>
To: "'www-xml-canonicalization-comments@w3.org'" <www-xml-canonicalization-comments@w3.org>
Cc: "'w3c-xml-schema-ig'" <w3c-xml-schema-ig@w3.org>
XML Schema WG thanks the XML Core WG for the opportuity to review this WD
[1].  We apologize for the delay in our response due to our preoccupation
with completing our own Last Call documents.

First, the Schema WG requests that explicit mention be made (in Section 3
"Document Type Definition Processing" and elsewhere as appropriate) of
canonicalization by schema-aware processors, as it does for DTD-aware
validating processors.  Schema-aware canonicalization will have major impact
on other sections of the c14n WD.  For instance, is their a canonical
lexical representation for schema datatypes (if so, some members of the WG
believe that Schema Part 2 should define that form, while others believe
that the c14n WD should define that form)?  Is there a canonical form for
uriReference values (especially in light of potential schema/XBase
interactions)?

[Note: the Schema WG did not unanimously agree that canonicalization should
be effected by the presence of a schema]

XML Schemas introduces several "schema-related markup" constructs for use
within instance documents governed by schemas [2]: xsi:type, xsi:null and
xsi:schemaLocation.  The Core WG should consider (in consultation with the
Schema WG) whether these markup constructs should always/never be present in
the canonical form of a schema-governed instance.

The Schema WG has serious concerns with the provisions in Section 5.9
"Namespaces" that require namespace prefixes to be rewritten.  The concern
stems from the extensive use of QNames in schemas and instances governed by
schemas.  Schemas allow document authors to put QNames in attribute values
and element text content.  Unless the prefix in these QName "values" are
also rewritten serious incompatibilities will result.  There is also concern
that the canonical infoset does not contain namespace declaration info
items, particularly given the stated concerns with prefix rewriting.

QNames also appear in XPath expressions, without the aid of a schema to
distinquish them, however.  Therefore, the prefix rewriting problem is not
just a schema concern but will have potential impact on many other namespace
aware instances.

Sections 5.6 "Tags" and 5.9 "Namespaces" describe how attributes (and their
values) are reordered in start tags.  Schemas reintroduce a limited form of
the SGML & connector into content models (the named elements can appear in
any order).  c14n might consider describing a canonical ordering for such
model groups.

[Note: the Schema WG did not unanimously agree that it would be desirable to
canonicalize element order in & groups]
  
References
[1] http://www.w3.org/TR/2000/WD-xml-c14n-20000119.htm
[2]
http://www.w3.org/XML/Group/xmlschema-current/structures/structures.html#Ins
tance_Document_Constructions
Received on Friday, 24 March 2000 15:49:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 6 December 2006 18:08:09 GMT