Re: Publication of XML Linking Requirements Documents

All:

The CritSuite team (http://crit.org/) is very encouraged
by the great work that has been done on the XLink and
XPointer standards so far.  However, there are two issues
with the requirements that we feel are quite important and
would like to draw to your attention.


1.  We are concerned about linking to documents that are
embedded in a frameset context.  Although framesets have
broken the simple linking model of HTML, they are popular
with sufficiently many websites that being able to work
with them is important.  We would like to see an explicit
requirement in the XLink requirements document that says
something like this:

    An XML link must be able to express traversal to an
    end in a document within a frameset context.

The operational requirement that we would like XLink to
satisfy is that an individual could select a span of text in
a document within a frameset, send a representative XLink to
a friend, and have the friend brought to the same document
in the same context with the same span of text highlighted.

Note that we are not necessarily suggesting that the context
be specified as part of the locator.  Indeed, it is better
for the locator to specify just the true link end itself,
and to specify the frameset context as part of the traversal
behaviour of the link.  Thus it would belong under the
"Remote Resource Semantics" section of the XLink draft.

It is also possible to solve this problem by allowing the
document itself to specify the frameset context under which
it prefers to be presented.  Regardless of how the problem
is solved, however, we feel it is important that XLink be
able to work with a standard solution.  Otherwise, we
expect that a number of ad hoc and incompatible solutions
for linking to framed documents will arise, which would be
a very unfortunate situation.


2.  We are concerned about linking to spans of text within
documents.  The XPointer requirements document specifies in
section C.2 that

    All XPointers must survive purely mechanical changes
    to the target resource.

and goes on to describe mechanical changes within the SGML
tags themselves.  We would like to go a little further and
make it more clear that XPointers should also be as tolerant
as possible of mechanical changes to whitespace in the
document text.  Many authoring tools manipulate such
whitespace freely where it has no effect on the rendering of
the document.  We propose the following requirement for
addition to the "Robustness" section:

    It should be possible to create an XPointer for a span
    of text in a document that survives typical methods of
    editing the document.

The operational requirement that we would like XPointer to
satisfy is that one could be given an XPointer referring to
a span of text in a document, and expect it to work the same
way on two documents that appear and behave identically when
rendered; and one could copy and paste a region of text
surrounding the span, and expect the XPointer to point out
the same span of words in the pasted text.


Thank you very much for your consideration of these issues.


The CritSuite Team.

Received on Tuesday, 23 March 1999 22:02:39 UTC