SYMM WG's comments on XLink last call

XLink WG members:

It's clear that the XLink WG has done a lot of work and has provided a
solution to a problem that works within a specific domain. Our objection is
centered around the general applicability of the proposal to languages such
as HTML and SMIL, which need other issues considered when determining an
appropriate linking syntax.

A previous draft of SMIL-Boston used the remapping facilities in an earlier
XLink WD to make SMIL-Boston conformant. This facility was dropped and no
replacement is provided in the last call draft, which leaves us and other
application designers in the position of being forced to use a specifically
recommended syntax in order to be XLink compliant. This syntax conflicts
with the syntax in the SMIL 1.0 Recommendation. Using this syntax would
require us to deprecate a significant amount of SMIL 1.0 syntax, an option
that is in direct opposition with some of our main goals for SMIL-Boston:
keeping backward compatibility with SMIL 1.0, and building SMIL-Boston on
our content author's SMIL 1.0 knowledge base.

The original requirements of XLink, as described in the requirements
document, include representing the HTML hyperlinks conveyed by the
HREF and SRC attributes, and enabling any XML document to have its
hyperlink semantics recognized, regardless of its syntax.  The XML
Linking working group has since made the decision not to meet these
requirements.  This decision is based on the group's consensus that
the technical means for meeting these requirements do not exist, and
it is outside of scope and capabilities of the XML Linking working
group to specify these technical means.

We feel that these requirements are important enough to expend the
extra required effort to meet them before releasing the XLink
recommendation. Furthermore, we feel that not meeting them will damage
Web semantic uniformity and syntactic design.  Very many existing, and
many upcoming, XML documents will not have their hyperlinks recognized
has XLinks unless these requirements are met.  Other XLink document
sets will be required to use XML syntax that is inappropriate for
their semantic domain in order to have their hyperlinks recognized has
XLinks.  The original XLink requirements indicate a mission to enable
hyperlinks of many syntaxes to be uniformly recognized as XLinks.  The
results of not meeting these requirements would be that one syntax
will be dictated for all documents sets that are to have their
hyperlinks recognized as XLinks -- effectively excluding virtually all
existing documents, documents conforming to existing recommendations,
and document sets with conflicting author design requirements.

The existing XML documents that won't have their hyperlink semantics
recognized as XLinks include those in the vast collection of HTML
documents already on the Web.  Search engines that use XLink-defined
hyperlinking will not include HTML-defined hyperlink semantics in
their query results -- a very large absence when searching on
documents located throughout the Web on the whole.  Upcoming XML
documents with non-XLink hyperlinks include those of the newly
released XHTML recommendation, a document collection that is clearly
expected to be large and substantial within the same scope as current
HTML.  Documents of the current SMIL recommendation, many of which are
already posted on the Web and used regularly, will also not have their 
hyperlinks uniformly recognized.

Because of this, we cannot support recommending XLink as a required means of
expressing linking semantics for all XML-derived languages. We could support
the current draft as a non-required means of expressing Xlinks in general
XML schemas that can use it painlessly, and support further work to make
links recognizable in the vast amount of pre-existing content on the web.

The SYMM Working Group

Received on Monday, 20 March 2000 14:36:39 UTC