comments on XML 1.1 last-call draft of 2002-04-25

The XML Schema WG congratulates the XML Core WG on the publication of
the last-call working draft for XML 1.1.  We have reviewed the
specification and have a few comments.

1 On the additions to the set of name characters

We have considered the rationale for and effect of the change in
philosophy regarding the definition of the NameStartChar and NameChar
productions: in XML 1.0, these productions omit code points not yet
assigned in the Universal Character Set; in 1.1, currently unassigned
code points are included.  Some individual members of the WG expressed
a lingering uneasiness with this change, but the WG as a whole
confirms that it understands the rationale for the change and has no
objection to it.

We note that these changes to the definition of Name in XML 1.1 will
call for knock-on changes in subsequent versions of XML Schema.  We
intend to align future versions of the XML Schema specification with
the then current versions of XML.

2 On the changes to the definition of S

Like the changes to the definition of Name, changes to the definition
of S will have effects on XML Schema. Our intention is to shift our
whitespace-related rules to align with those defined here in future
versions of XML Schema.

In passing, we note that there are some reports that since the
publication of the last call specification, the XML Core WG has begun
to consider the possibility of leaving the definition of S unchanged,
precisely in order to prevent knock-on effects for other
specifications which refer to the XML definition of whitespace.  As
one WG affected in this way, we should state that we do not believe
that you should on this account refrain from making a change to XML
which you believe necessary and appropriate.  Our specification, and
others, refer to the XML definition of whitespace for convenience and
as a way of encouraging consistent definitions and behaviors across
specifications; our dependency on your specification means, in this
case, that we are perfectly willing to follow your lead in this
question.  You should not refuse to change the definition of S just
because that would mean we will have to do so, too: defining
whitespace in terms of the XML definition of S is our way of saying
that we will in fact (other things being equal) track changes in the
XML definition of S.

We are aware that any such change may require careful handling to
minimize the inconvenience of inconsistent behavior from different
processors owing to differences in language versions supported, but as
a WG we believe this is a manageable risk and task.

3 The requirement for Unicode normalization

As we have said in our comments to the I18n WG on their last-call
draft of the Character Model specification (2002-04-30), we continue
to harbor reservations about the wisdom of the
early-uniform-normalization discipline as it is defined in that spec.
On the whole, we would prefer that normalization not be required as a
condition of XML well-formedness.

If the Web community embraces early uniform normalization and it is
accepted as a fundamental rule of W3C specs, then we believe that you
should retain the requirement; if early uniform normalization is not
accepted, and is (for example) removed from the Character Model
specification, then we believe the normalization requirement should be
removed.

We suggested that the terms 'fully normalized' etc. be hyperlinked to
the appropriate definitions in the character-model specification.


Thank you for your work on this specification.

-C. M. Sperberg-McQueen
  Co-chair, W3C XML Schema Working Group
  on behalf of the XML Schema WG

Received on Friday, 12 July 2002 19:52:40 UTC