Lists Home |
Date Index |
The XML Linking Working Group discussed your message 
in our conf. call of 2002-07-17. (My apology for the
delay in reply, I've been tardy in writing this up.)
As Henry Thompson stated in his personal response, the
group does not think a schema() pointer part is required,
although it is a very interesting idea.
So why is it not required?
With the xpointer() scheme, (and other schemes in the
XPointer Framework), there are many ways that a reader
can construct a link to a portion of a document. They
do not have to rely only on the presence of ID attributes.
(For example - if we wanted to identify the element bearing
a 'foo' attribute with a value of 'bar', we could override
the declaration of 'foo' to make it an ID. But we could do
other things, such as use basic XPath predicates, as in:
While there are many options for a document reader to identify
parts of a XML document, there is essentially only one way
for a document creator to indicate an element as a likely
target of a link. They must provide an attribute which has
somehow been declared as an ID. Because there are so few ways
for the document creator to express that intent, and because
the document readers do not actually *require* a schema override
in order to address any part of a document they might wish, it
seems best not to rush to define such an override. Given that
the linking WG's charter expires at the end of the year, we
believe it is best not to try to define such a thing at this
Please let us know if you find this answer acceptable, or
if you think we really do need to define such a thing within
the lifetime of this group.
Ron Daniel Jr.
(acting chair of the XML Linking WG)
803 11'th Ave.
Sunnyvale, CA, USA 94089
Tel: 408 530 5922
Cell: 925 368 8371
The Leader in Enterprise Content Management