Re: XLink: behavior must go!

Subject: Re: XLink: behavior must go!
From: "Jonathan Borden" <jborden@xxxxxxxxxxxx>
Date: Thu, 13 May 1999 12:59:14 -0400
Simon St.Laurent :

>At 09:13 AM 5/13/99 -0400, Jonathan Borden wrote:
>>    Is there anything within XLink itself that cannot be replaced by XSLT
>>now that doc() and docref() have been defined? Does XLink not become
>>something akin to a standard set of XSLT templates used for handling URI
>>traversal? doc() and docref(), as well as unification with XPointer turn
>>XSLT into a generalized graph transformation language. Could the XLink
>>itself become an XSLT include file?
>Er... just everything.
>One of the key points of XLink is that it is _not_ bonded to a particular
>style sheet language.  XLink is useful in contexts where XSLT is either too
>much or too little, and provides common vocabulary that document developers
>can use to describe links whatever final processing the documents may
>If your question is rephrased:
>"Is there anything within XLink itself that cannot be implemented by XSLT?"
>Then it might be received a little more kindly by those of us who work with
>XML in contexts where XSL (indeed style sheets, at times) is unnecessary.

        No intention to offend. You are correct, it seems to me that XLink
can be implemented via XSLT now that XSLT includes doc() and docref(). There
remains a distinction between the concept of XLink and the implementation.
The concept of XLink does appear to be related to the concept of XPointer,
both are specifications for graph traversal. My thought was that as it has
been announced that the XPointer and XSL specs are to be combined that XLink
might benefit from a similar integration. What I am struggling with is the
question as to, if behaviors are removed from XLink, what concept does XLink
contain beyond either a single URI or set of URIs? All of this, including
bi-directional links etc, comes down to graph->graph transformation. It is
unfortunate that this very important and general facility gets labeled a
'stylesheet' technology (not that I have anything against stylesheets. Since
graph->graph transformation underlies a wide range of facilities, including
RDF, a central and core definition of graph-graph transformation and
traversal facilities greatly strengthens the ability of 'XML' to deliver on
it's promises. XLink, like XPointer is a key element to graph traversal and
for this reason these technologies should be grouped together and viewed as
parts of a single larger picture.

Jonathan Borden

 XSL-List info and archive:

Current Thread