UTC/1999-021

Liaison report from W3C

Subject: W3C XML CG statement on annotation characters

-------------

Hello Lisa, hello Arnold,

I have received the following statement on annotation characters
from the Chair of the W3C XML Coordination Group, and I am forwarding
it here in my role as liaison from W3C to the Unicode Consortium.
I hope you can bring this to the due attention of the UTC.

Regards,   Martin.

------------

The XML Coordination Group of the World Wide Web Consortium wishes to
express its concern regarding the inclusion of annotation characters
in Unicode and to urge that the UTC take a position opposing their
adoption.

BACKGROUND: THE XML CG

The W3C XML Coordination Group continues the work of the W3C XML
Working Group (1996-1998), which designed XML, the Extensible Markup
Language.  The XML CG includes the chairs of the five current W3C XML
Working Groups (Schema, Linking, Syntax, Infoset, and Fragment) and
coordinates this work with other W3C working groups, most importantly
the Document Object Model (DOM), Extensible Stylesheet Language (XSL),
and Internationalization (I18N) Working Groups.

XML CG POSITION REGARDING ANNOTATION CHARACTERS IN UNICODE

The XML CG is deeply concerned about the inclusion of these three
annotation characters in the Unicode standard:

   U+FFF9 Interlinear Annotation Anchor
   U+FFFA Interlinear Annotation Separator
   U+FFFB Interlinear Annotation Terminator

It is our view that the functionality sought for in adding these
characters should be provided using markup, not characters, and that
if there are particular implementation needs that have caused this
proposal to be put forward, architectural mechanisms should be
identified that would satisfy them without imposing a stateful
mechanism like ISO 2022 onto the World Wide Web.  Our reasons for
taking this position are as follows:

1. The annotation characters appear to contravene a basic principle of
   Unicode design.  According to Section 3.11 of the Unicode 2.0
   specification,

      In general, Unicode does not supply formatting codes;
      formatting is left up to higher-level protocols.

   The only exception to this rule is in the case of bidirectional
   behavior, and then only because "there are circumstances where an
   implicit bidirectional ordering is not sufficient to produce
   comprehensible text."  No such need forces a violation of the
   basic design principle in this case.  The annotation characters
   solve no problems that are beyond the reach of higher-level
   protocols, but their use can impair the integrity of those
   protocols.

2. The occurrence of annotation characters in XML content will create
   rendering problems by overlaying a concurrent and conflicting set
   of formatting controls on top of the controls driven by XML
   tagging.  In particular, the annotation characters violate a basic
   assumption of XSL, which is that style boundaries coincide with
   element boundaries in the source document.  The occurrence of
   annotation characters in XML content will directly conflict with
   stylesheets applied to documents containing that content.

3. The name "annotation characters" suggests that these are intended
   to deal with interlinear annotations in general, but the mechanism
   does not scale well to nested interlinear material (interlinear
   notes on interlinear notes) or to multiple levels of interlinear
   annotation on the same base.  Nesting and multiple annotations on
   the same base text are well known in both Western and Eastern
   writing traditions; both are readily handled by higher-level
   protocols such as XML.

4. Similarly, the recursive embedding of such characters in MathML
   data could cause severe problems for rendering engines.

5. The adoption of annotation characters would set a precedent with
   harmful consequences for the users of XML, opening the door for
   the addition of Unicode characters to shift in and out of bold,
   italic, and other typographic modes.  From an architectural
   standpoint, there is no difference between shifting into
   annotation mode and shifting into 36 point type.  Such shifting is
   opaque to XML tools and, as is well known, destroys the ability to
   repurpose documents that use it.

SUMMARY

The XML Coordination Group views the inclusion of annotation
characters in Unicode as a basic layering violation with grave
consequences for the higher-level protocols that will be the most
important users of Unicode.  We urge the UTC to reject their adoption.

Jon Bosak
Chair, W3C XML Coordination Group



#-#-#  Martin J. Du"rst, World Wide Web Consortium
#-#-#  mailto:duerst@w3.org   http://www.w3.org