Extended Link Groups and Link Chaining

Extended Link Groups and Link Chaining.

I'm trying to see how I can chain links in xmllink.  In 4.0 I read:

| A special case arises when the anchor indicated by a link-end proves to 
| be another link element. When an application traverses or displays such 
| a link, it may be desired to read and traverse that link to whatever its 
| anchor(s) is (are) and operate on the result of that traversal. This has 
| the benefit of allowing greatly increased indirection and flexibility; 
| on the other hand, it might lead to potentially unbounded delays in 
| traversing simple-looking links. This specification does not constrain 
| the behavior of applications in this situation.

What situation?  Does this language mean that this spec provides
no explicit way to chain links?  or is it about "unbounded delays"?

5.1 gives both reserved element names and fixed attribute values,
which are also reserved names.  Why the apparent redundancy?

5.2 places constraints on where mlinks may occur:

| First, multilinks may occur in the LINKS element of a document, which is 
| part of its header. So long as any part of the document remains open, 
| all anchors of all multilinks in the header shall be known to the 
| software and available to the user. Applications may apply their own 
| conventions for when a document is no longer considered "open", except 
| that any document any part of which remains in view, is open by 
| definition.

which is a piece of conformance language I'd missed; apologies for
that.  But I am confused by the notion of a (required?) LINKS element in
a header of a document the structure of which isn't constrained
by XML 1.0.  Is LINKS also a reserved name?  HEADER?  or is this
info that could be conveyed through the MIME wrapper?

Concretely, how does this work over the Internet?  Can I use a 
LINKSET document that is not pointed to by one of the documents
it points into?  How does this all get packed up and delivered
to the user?

    Terry Allen    Fujitsu Software Corp.    tallen@fsc.fujitsu.com
"In going on with these experiments, how many pretty systems do we build,
 which we soon find outselves obliged to destroy?" - Benjamin Franklin
  A Davenport Group Sponsor:  http://www.ora.com/davenport/index.html