Re: XLINK, HLINK or neither? Possible new TAG issue

The HTML WG has asked me to address the TAG on the issue of Linking, in
particular in response to "When should I use XLink?", which says:

The XHTML specification does not use xlink, as (I understand) the working
group felt that it was too clumsy to use a different namespace, and they
wanted it to look like HTML, which uses href=. The group is (2002/06)
looking at schema annotation ways of declaring html:href to carry the
significance of an xlink. These are known as "hlink". The pros and cons of
schema annotation in general as a means to add semantics or style to a
langauge are currently under debate in the community.

The HTML WG feels there are many things wrong with the XLink spec. We think
there are architectural problems. We don't believe that for links the
datatype of the attribute should be forced to show through into the instance
(The XLink requirements stated this too: "It must be possible to apply
XML link semantics to existing documents ... requiring no modification to
the document instances themselves.") We don't believe that elements should
be restricted to one link attribute, or one date attribute, or one attribute
of any kind, with the exception of ID (though I personally don't know what
would break if an element was allowed to have multiple ID attributes).

We think there are semantic problems, since there are such clear
requirements from the existing usage of linking on the web, that XLink
should at least support those.

We communicated to the XLink group many times what we needed. We sent in
last call comments. Unfortunately the XLink group almost never replied to
us. Certainly not on our last call issues. In desperation I actually once
travelled to a linking face-to-face in France hoping to get some
coordination. At that meeting I was astonished to have to explain how
to="Rome" or date="20020219" could be links, and to hear them reject
proposed solutions as 'unaesthetic' rather than 'technicallly infeasible'.

My view is that a WG represents the community. It is not possible for all
stake-holders to be involved in all specs, so a WG is entrusted to go away
and produce a proposal for a specification. They should then check with the
community that what they have done matches what is needed. The W3 process is
there to ensure that the community gets the chance to make comments and come
to consensus on the contents of that spec. That did not happen with XLink,
and in my view is therefore not a real W3C specification.

When XLink went to Rec, the members said that a solution for linking for
XHTML should be pursued. Since that has to date not happened, we have
proceeded to work on HLink ourselves, since we need a solution.

We always wanted to use XLink, and agreed with the requirements document for
it; we put a lot of time and effort into trying to achieve a solution. On
this occasion the W3C Process failed us.

Best wishes,

Steven Pemberton

----- Original Message -----
From: "Tim Berners-Lee" <>
To: <>
Sent: Friday, June 14, 2002 9:00 PM
Subject: XLINK, HLINK or neither? Possible new TAG issue

> Martin Dürst mentioned, in response to the last call for the speech
> grammar draft, that there is a question as to whether for any URI
> parameter, xlink:href should be used.
> There was a note in later discussions that the HTML working group are
> working on a "HLINK" spec for describing xlink functionality as a schema
> annotation level solution to thius, just as several times we have looked
> at schema annotation as a compromise between RDF syntax and no semantics
> at all.
> For me this questions depends on whether the document type is a
> human-readable hypertext document, when  generic hypertext xml tools
> would benefit from knowing what is a link, and whether significance of
> the URI in question is a hypertext link or something different.
> Maybe the TAG should give some direction in this area.
> Tim BL

Received on Wednesday, 10 July 2002 09:46:55 UTC